Search

Artificial intelligence (AI) ethics has emerged as a global discourse within both academic and policy spheres. However, translating these principles into concrete, real-world applications for AI development remains a pressing need and a significant challenge. This study aims to bridge the gap between principles and practice from a regulatory government perspective and promote best practices in AI governance. To this end, we developed the Ethical Regulatory Framework for AI (ERF-AI) to guide regulatory bodies in constructing mechanisms, including role setups, procedural configurations, and strategy design. The framework was developed through a systematic review, thematic analysis, and the integration of interdisciplinary concepts. A comprehensive search was conducted across four electronic databases (PubMed, IEEE Xplore, Web of Science, and Scopus) and four additional sources containing AI standards and guidelines from various countries and international organizations, focusing on studies published from 2014 to 2024. Thematic analysis identified and refined key themes from the included literature and integrated concepts from process control theory, computer science, organizational management, information technology, and behavioral psychology. This study adhered to the PRISMA guidelines and employed NVivo for thematic analysis. The resulting framework encompasses 23 themes, particularly emphasizing three feedback-loop processes: the ethical review process, the incentive and penalty process, and the mechanism improvement process, offering theoretical guidance for the construction of ethical regulatory mechanisms. Based on this framework, a seven-step process and case examples for mechanism design are presented, enhancing the practicality of ERF-AI in developing ethical regulatory mechanisms. Future research is expected to explore customization of the framework to remain responsive to emerging AI trends and challenges, supported by empirical studies and rigorous testing for further refinement and expansion.

Air conditioning load has become a crucial demand response resource in power systems. However, due to its diversity in types and decentralized integration, the dispatch center faces challenges in directly accessing its aggregated power and conducting scheduling control, limiting the full potential of its response. To address this issue, this paper proposes a dual-layer control framework that combines multiple types of resources, considering the aggregation response potential of air conditioning load, and integrates precise control into the scheduling process. In the day-ahead scheduling layer, an approximate aggregation model is used to determine the aggregated power of air conditioning load. Considering factors such as user thermal comfort, willingness, and controllability, an evaluation model for air conditioning load aggregation response potential is established. This model, combined with the response characteristics of fundamental flexible loads, constitutes a unified scheduling model, effectively leveraging the potential of various demand-side resources in system regulation. In the intra-day control layer, to tackle the power drop phenomenon in air conditioning groups during load reduction and temperature control strategy execution, a variable-state queuing model is introduced. By introducing preparation time, heterogeneous air conditioning clusters are grouped for controlled operation, ensuring that air conditioning load follows the scheduling plan. This enhances control precision and mitigates the impact of power drops on system operation. Lastly, based on simulation analysis of a simplified distribution network system, the results indicate that the proposed two-layer control framework can effectively harness and direct the utilization of air conditioning load response potential at the scheduling layer. At the control layer, it achieves precise control and mitigates the negative effects of power drops, demonstrating substantial practical value in engineering applications.

With the continuous advancement of industry 4.0 and intelligent manufacturing, there is a growing need for automation and intelligence in industrial production processes. Anomaly detection of industrial product surface images is a key technology in achieving this goal and has thus become a significant area of research. However, this endeavor still encounters challenges such as scarcity of abnormal samples, complexities in data labeling, and uncertainties stemming from unknown factors and randomness. To this end, we propose the Feature Extraction-based for Generative Adversarial Network (FEGAN), aimed at detecting and precisely localizing surface anomaly in industrial products. FEGAN focuses on the deep features of an image, and it builds a feature extraction network and an improved generative adversarial network based on VGG19, respectively. We also jointly determines the anomaly score through the deep feature space as well as the Euclidean distance in 2D image space to better identify and locate the anomaly. Furthermore, we introduce a novel Multi-scale Self-Enhancement (M-SE) strategy to bolster the model’s generalization capabilities. We conducted training and testing on the MVTec and Bottle-Cap public datasets. A plethora of experimental results indicate that the proposed method outperforms existing methods significantly in terms of anomaly detection. Additionally, through an evaluation of the model’s localization accuracy, we demonstrate that FEGAN exhibits certain competitive advantages.

With the continuous advancement of deep neural networks, leveraging deep learning for the classification and diagnosis of medical images to aid physicians in patient diagnosis has gained significant traction. However, the efficacy of deep learning networks is not solely contingent upon the network architecture, it also relies on the appropriate setting of hyperparameters. Properly tuning hyperparameters can markedly enhance the performance of deep neural networks. To propose an efficient deep learning method for medical image classification, this study introduces a deep residual neural network optimized using an enhanced Heap-based Optimization (HBO) algorithm. The enhanced HBO algorithm integrates elements grey wolf mechanism and orthogonal learning mechanism to augment algorithm convergence and bolster the equilibrium between global exploration and exploitation. The proposed algorithm is applied to optimize the hyperparameters of the deep residual network, thereby enhancing accuracy and reliability in medical image classification. To evaluate the efficacy of the proposed model, this study conducts various experiments on the CEC 2017 benchmark functions for the updated HBO algorithm, encompassing validation of strategies, scalability assessments, balance analyses, and comparisons with state-of-the-art similar algorithms. The performance of the proposed evolutionary deep learning method is assessed across three medical image datasets. The three medical image datasets encompass the colorectal cancer dataset, brain tumor dataset, and chest X-ray image dataset. These datasets represent tissue images, magnetic resonance imaging (MRI), and computed tomography (CT) images, respectively. The classification accuracies on the three datasets reached 95.6%, 97.64%, and 96.48%, respectively, outperforming other competing algorithms.

Spectral clustering (SC) is a hot research topic in graph theory today, which is centered on constructing similarity matrix. Many previous works have studied how to construct a higher-quality similarity matrix. However, most of these works ignore the connection between constructing similarity matrix and the overall distribution of the datasets. This problem will lead to low quality and lack of robustness in the similarity matrix when the data volume and dimensionality are expanded, which leads to poor clustering performance. Therefore, to solve the above problems, an Adaptive Fuzzy Spectral Clustering (AFSC) model is proposed in this paper. The model considers the local neighbor and global fuzzy affiliation information of the samples to construct the objective function, which can obtain the similarity matrix that satisfies the overall distribution of the datasets. Meanwhile, we propose a weight definition method to balance the above two kinds of information, which can prevent the adverse effect of information loss on the quality of the similarity matrix. The similarity matrix obtained by the AFSC model will have high quality and strong robustness, which leads to excellent clustering performance. In experiments, we verified that the AFSC model can obtain excellent clustering performance and outperform some state-of-the-art clustering models as it is applied to several datasets of different sizes and dimensions.

The air pollution problem has been a serious issue for public health and city development in recent years, which rises an urgent demand for accurate air pollution prediction models. Traditional time series prediction scheme has been challenged in such case because the air pollution signals can be highly dynamic and uncertain. Distinct latent physical processes and complex environment changes over time make it hard for one fixed model to give consistently good performance. Also probabilistic prediction and uncertainty-aware estimation are critical when the model is used for public decisions in practice. However, few previous works can simultaneously provide adaptive and probabilistic air pollution prediction abilities, which are important practical issues for the real-world implementation of data-driven air pollution prediction models. In this paper we propose an adversarial meta learning approach to address these concerns. After reformulating the target signals as a collection of independent but similar mini-tasks along time, we adversarially learn a meta model that includes an implicit generator to adaptively give more informative probabilistic predictions across tasks. We further provide a bayesian meta learning interpretation that recasts the proposed model as an approximate minimizer for the Wasserstein distance between a generative latent model and the true data distribution. Our model provides an algorithm that explicitly models how the prediction distribution is conditioned on underlying data patterns and simultaneously gives adaptive uncertainty estimation. Experiments on both synthetic and real-world air pollution datasets show that the proposed model can simultaneously provide better probabilistic and adaptive prediction while keeping stratifying point prediction error.

Real-time visual-aided navigation and path strategy for pneumonoconiosis and efficient 3D visualization of pulmonary vessels are of great research and clinical significance in the treatment of lung diseases. The complex structure of lung tissue limits the application of deep learning in pulmonary vascular visualization due to the lack of vascular labeling datasets. Also, the existing methods have large computational complexity and are low efficiency. This study proposes a method for high-quality 3D visualization of pulmonary vessels based on low-cost segmentation and fast reconstruction, consisting of three steps: 1) Pulmonary vessel feature extraction from lung CT images using self-supervised learning, 2) Segmentation of pulmonary sparse vessels in lung CT images using self-supervised transfer learning, and 3) 3D reconstruction of pulmonary vessels based on segmentation results of step (2) using interpolation. The accuracy of pulmonary vascular contour segmentation was improved from 91.31% using the sparse coding to 98.65% using our proposed method (27,270 test sample points); the classifier evaluation accuracy was improved from 95.33% to 98.26%, and the average running time of the model with the test set data was 44 ms per slice. the segmentation results can automatically generate a complete vascular tree model with an average time of 10.8s ± 1 1.6s. The results demonstrate that the proposed method provides fast and accurate 3D visualization of pulmonary vessels, and is promising for more precise and reliable information for pneumonoconiosis patients.

Fourier ptychographic microscopy (FPM) is a recently developed computational imaging technique which can perform complex amplitude imaging with both large field of view and high resolution by using a simple microscope setup. Here, we propose a transformer based neural network named as FP-transformer, which takes the low-resolution amplitude (LRA) images as the sequential input and uses self-attention mechanism to compute the relationship among them. The high-resolution FPM complex amplitude reconstruction is the end-to-end output of the FP-transformer. We apply the image library of div2k to generate the FPM LRA images with the physical model, and then perform the training and validation with this dataset containing ground truth. We also perform the validation with the experiment images and it is found that the high-quality FPM complex amplitude image pairs can be obtained. Therefore, the FP-transformer creates a new platform for the FPM deep learning reconstruction, which has the better dependability and adaptability.The code of this work will be available at https://github.com/zhaolin6/FPTransfomer for the sake of reproducibility.

In recent years, deep learning-based hashing techniques have garnered significant attention within the academic community, owing to their impressive performance achievements. However, the current deep hashing method is limited by its use of neural networks cannot make good use of the important feature information of the image, thus limiting the retrieval accuracy. In this paper, we design a two-channel feature extraction network called the local and global feature fusion network (LGDH). Specifically, the features extracted by visual geometry group (VGG) network and con-volution-enhanced image transformer (CeiT) network are deeply orthogonal fusion of local and global features, which solves the problem of lack of semantic correlation in existing deep hash methods. To fully harness the insights gained from previous hash function learning, we employ long short-term memory (LSTM) decoder to proficiently shape and create a resilient public binary code space. Through detailed experiments, our proposed method has achieved the Sota level in three public datasets CIFAR10, NUS-WIDE and MIRFLICKR compared with multiple cutting-edge algorithms, which can prove the effectiveness of our pro-posed method.

Anomaly detection within the realm of industrial products seeks to identify regions of image semantics that deviate from established normal patterns. Given the inherent challenges associated with collecting anomaly samples, we exclusively extract features from normal semantics. Our proposed solution involves a Semantic CopyPaste based Generative Adversarial Network (SCGAN) for unsupervised anomaly detection. To enable the comprehensive acquisition of semantic features within intricate real-world images, we embrace an encoder-decoder-encoder as the fundamental network structure. In practical terms, our approach commences with the input image being subjected to the CopyPaste augmentation module. Here, we strategically copy N patches, each constituting 1% of the image’s area, from normal samples. These patches are then randomly pasted into different regions of the original image. Subsequently, a generative adversarial network is trained to facilitate sample reconstruction. A noteworthy augmentation to the network’s channel attention capabilities entails the incorporation of a multi-scale channel attention module within the first encoder. This module serves to emphasize contextual features across varying scales within the image. During the test, we detect anomalous regions by meticulously comparing residuals between the input image and its reconstructed counterpart. Our methodology is rigorously validated through diverse experiments conducted on challenging MVTec and BTAD public datasets. The results conclusively affirm the state-of-the-art performance achieved by our proposed method in the domain of anomaly detection.

Object detection based on deep learning is a popular trend, and it includes object recognition and positioning. This paper proposes a method that can accurately obtain object type and accurate three-dimensional position. The method is divided into three parts: object recognition and coarse positioning based on deep learning, precise positioning based on deep learning combined with B-spline level set in color images, and precise three-dimensional positioning with depth information of RGB-D camera. The precise positioning of the object provides accurate end pose information for the autonomous grasping of the robotic arm, and it has great significance to the gripping of the robotic arm. Performance metrics include mAP (mean average precision) and IOU (intersection of a union). Experimental results show that the mAP value of Yolo-v3 in this paper can reach 87.62%, the average IOU of Yolo-v3 in this paper can reach 66.74%, the average IOU of Yolo-v3 and B-spline level set can reach 100%, and can get accurate 3D location in the real scene. In addition, the experiments comparisons between VOC dataset and our own dataset validate that our dataset can take higher mAP and average IOU values.

State-of-the-art Feature Pyramid Networks (FPNs) often focus on extracting features across different levels. In this paper, we propose a novel architecture, Bidirectional Parallel Feature Pyramid Network (BPFPN), to capture multi-scale spatial information from each level of FPN effectively. BPFPN consists of two blocks: Cross-level Channel Attention-Refinement (ClCSAR) Block and Weighted Parallel Feature Aggregation (WPFA) Block. ClCSAR block uses a channel attention mechanism to strengthen the context information of lower-level feature with aid from the upper-level feature. WPFA block exploits discriminating information from variable receptive fields via integrating multi-branch by employing dilated convolutions and using attention mechanisms to capture the salient dependencies over branches. Considering the incremental computation, we also give a lightweight version of BPFPN, namely BPFPN-Lite, integrated with an Efficient WPFA (E-WPFA) to improve detection accuracy while maintaining efficiency. Our proposed network can be easily plugged into existing object detection models and outperforms different feature pyramids methods by $0.2\sim 2.1$ on the COCO test-dev benchmark without bells and whistles.

Nowadays, digital images are widely used in various services. The emergence of more and more image editing algorithms has made image forensic approaches to be severely challenged. Driven by the emergence of forged images, more and more image forensic methods are proposed to evaluate the authenticity of digital images. However, in some privacy-related image forensics areas, the scarcity of data affects their development. In this paper, we investigate a document image generation scheme based on face swapping and distortion generation to generate document image databases at a low cost. First, we propose an IDNet for editing face content and text content in digital document images. Second, we propose a Distortion Simulation Network (DSNet) for simulating print-and-scan distortion, and the generated data can be used to study a novel document attack type called recapture attack. Third, we use the database generated by our proposed method to assist in training document image recapture forensic networks. A qualitative comparison with existing methods illustrates that the image content generated by our proposed method maintains more complete semantic information and higher image quality. Quantitative results have confirmed that with the addition of the generated database, the performance of the model improves by about 8% measured in AUC.

Fine-grained air quality can provide essential urban environmental information for administrators and residents. With advances in communication and sensing technologies, low-cost portable sensors installed on vehicles enable high-coverage air quality monitoring. However, data collected by low-cost mobile sensors may be inaccurate and inconsistent in complex operation environments, which brings the issue of data uncertainty. Moreover, due to uncontrolled vehicles and human activities, the coverage of mobile nodes is dynamic over time, leading to uneven or sparse spatial distribution. To address these challenges, we propose AQI-M3, a novel framework for fine-grained air quality inference via multi-view learning with mobile sensing memory. Specifically, an encoder-decoder structure is applied in the region view for modeling the spatial dependencies in pollution maps. More importantly, sensing gradients are extracted in the trajectory view to enable the utilization of uncertain mobile sensing data. In addition, a memory network is designed to capture the spatial patterns from the historical sensing data and provide the global patterns as a complemental guide to overcome dynamic coverage sampling. Extensive experiments are conducted on three real-world deployments of hybrid sensing systems with both static and mobile sensors. Experimental results show that our proposed approach outperforms competitive baselines with 17% $\sim 29$ % reduction in mean absolute error. Furthermore, detailed evaluations demonstrate the effectiveness and robustness of the proposed framework under dynamic coverage.

According to the transmission mode of polarized light in Mueller ellipsometry, a characterization method for the thickness and optical constants of isotropic nano films based on Self-Adaptive Differential Evolution algorithm (SADE) is proposed. By establishing the least square model of the output light intensity with respect to the Mueller matrix of the standard sample to be measured, the elements of the Mueller matrix are solved by using the Sade algorithm, and the Mueller spectral curve obtained by fitting is compared with that measured by DRC-MME, and the film thickness is calculated by using the transfer matrix. The SiO2 / Si standard samples with calibration values of 100.4 nm and 121.56 nm are simulated and calculated. The experiment shows that: when numbers of iterations accumulated to 65 and 80 respectively, the residual square sum of the light intensity of the objective function converges to the minimum values of 1.24 and 1.02. The calculated film thickness values are 101.25nm and 120.53nm respectively, and the relative errors are both less than 1%. It proves the characteristics of simple calculation, fast convergence and accurate global optimal solution.

Multiple sensors are often used in robotic applications for better situational awareness. Hence, sensor fusion becomes a key technology to manage multiple sources of information and plays a critical role to the success in robotic tasks such as object detection and tracking, autonomous navigation, and interaction with humans. With these capabilities, wheeled autonomous vehicles can be used to automate some public services. However, there are still challenges for wheeled vehicles to safely and agilely maneuvering in human-centered environments. One of these challenges is lacking the capability of autonomously opening doors and traversing doorways without using a general-purpose robotic arm (manipulator). An autonomous door-opening operation is a complex task consisting of identifying the door and door handle, navigating the vehicle to the door, operating the door handle, and pulling or pushing the door to open while traversing the doorway. A self-closing door adds significant difficulty for the last step because the door usually needs to be held open while the vehicle is traversing the doorway. This paper presents a method using force-vision sensor fusion to enhance a deep reinforcement learning (RL) process for a wheeled vehicle to perform the most difficult step of a door-opening and pass-through operation. That step is to pull and hold a self-closing door open while the vehicle is traversing the doorway. In our solution, the vehicle is equipped with a camera, a force sensor, and a concise door-opening mechanism. The method was simulated in Gazebo and the results demonstrated that the deep RL-based force-vision sensor fusion method can be successfully applied to the task of self-closing door pulling by a wheeled vehicle without using a robotic arm and without a pre-planned trajectory. The vehicle control was trained without using domain randomization, but it still works in variant environments.

The boom in online food delivery(OFD) platforms brings convenience to both consumers and restaurants but leads to pollution from nonbiodegradable plastic food containers. To reduce pollution, the Chinese government enacted a ban on using multiple single-use plastics, including nondegradable OFD food containers. There may be various interests involved in enforcing the ban, so careful analysis is required before implementing the ban. This study applied evolutionary game theory to examine the interaction mechanism of the problematic behaviors between OFD platforms and restaurants. By theoretical research and simulation, we analyzed the conditions under which the game can converge to the expected outcome. In addition, we discussed the measures that are most important to promote restaurants using degradable food packaging in their service.

To mitigate the temperature sensitivity of photonic devices, materials with a negative thermo-optical coefficient (TOC) are integrated to optical waveguides. However, previously reported waveguides are made athermal at only one wavelength. In this paper, we theoretically propose a new broadband athermal waveguide, which consists of TiO2 with a negative TOC and SiC with a positive TOC in its core. This composite core shows a near-zero broadband effective TOC, i.e., ±1x10-6/K over a 780-nm bandwidth from 1280 to 2060 nm. Furthermore, it also has low anomalous dispersion, from 66 to 329 ps/nm/km in the same wavelength range. This new athermal waveguide, when used to form microresonators, enables us to achieve broadband nonlinear applications with negligible intra-cavity thermal dynamics on a chip. We also show that the proposed waveguide can be tightly bent without suffering from a large bending loss/substrate leakage, which is suitable for dense integration.

Traditional defensive techniques are usually static and passive, and appear weak to confront highly adaptive and stealthy attacks. As a novel security theory, Cyberspace Mimic Defense (CMD) creates asymmetric uncertainty that favors the defender. CMD constructs multiple executors which are diverse functional equivalent variants for the protected target and arbitral mechanism. In this way, CMD senses the results of current running executors and changes the attack surface. Although CMD enhances the security of systems, there are still some critical gaps with respect to design a defensive strategy under costs and security. In this paper, we propose a dual model to dynamically select the number of executors being reconfigured according to the states of the executors. First, we establish a Markov anti-attack model to compare the effects of CMD under different types of attack. Then, we use a dynamic game of incomplete information to determine the optimal strategy, which achieves the balance of the number of reconfiguration and security. Finally, experimental results show that our dual model reduces defensive costs while guarantees security.

The atmospheric turbulence (AT) causes distortion of phase fronts of orbital angular momentum (OAM) beams, which hinders the recognition of OAM modes. Convolutional neural network (CNN), a deep learning (DL) technique, can be utilized to realize the effective recognition of OAM modes. In this article, we propose a properly designed six-layer CNN model that can achieve high recognition accuracy (RA) of OAM modes at a reasonable computing complexity. We used intensity images of Laguerre-Gaussian (LG) beams to train our CNN model and explored the relationship between the RA for different single OAM modes and many factors including the number of training epochs, AT intensity, transmission distance, and the number of single OAM modes. Our CNN model can obtain an RA of 97.1% under moderate turbulence and 80% under strong turbulence, which are better than some CNN models proposed previously. Compared with these previous CNN models, our CNN model can also reduce the time consumption for at most 70%. Our research could contribute in achieving higher data capacity in OAM-based free space optical (FSO) communication systems.

Remote sensing image (RSI) scene classification is the foundation and important technology of ground object detection, land use management and geographic analysis. During recent years, convolutional neural networks (CNNs) have achieved significant success and are widely applied in RSI scene classification. However, crafted images that serve as adversarial examples can potentially fool CNNs with high confidence and are hard for human eyes to interpret. For the increasing security and robust requirements of RSI scene classification, the adversarial example problem poses a serious problem for the classification results derived from systems using CNN models, which has not been fully recognized by previous research. In this study, to explore the properties of adversarial examples of RSI scene classification, we create different scenarios by testing two major attack algorithms (i.e., the fast gradient sign method (FGSM) and basic iterative method (BIM)) trained on different RSI benchmark datasets to fool CNNs (i.e., InceptionV1, ResNet and a simple CNN). In the experiment, our results show that CNNs of RSI scene classification are also vulnerable to adversarial examples, and some of them have a fooling rate of over 80%. These adversarial examples are affected by the architecture of CNNs and the type of RSI dataset. InceptionV1 has a fooling rate of less than 5%, which is lower than the others. Adversarial examples generated on the UCM dataset are easier than other datasets. Importantly, we also find that the classes of adversarial examples have an attack selectivity property. Misclassifications of adversarial examples of RSIs are related to the similarity of the original classes in the CNN feature space. Attack selectivity reveals potential classes of adversarial examples and provides insights into the design of defensive algorithms in future research.

Fine-grained air pollution monitoring has attracted increasing attention worldwide. Even with an increasing amount of both static and mobile sensing systems, an inference algorithm is still essential to achieve a comprehensive understanding of the urban atmospheric environment. Conventional physical model-based methods are unable to involve all the influencing factors with limited prior knowledge, and datadriven methods lacking physical interpretation may result in bad generalization ability. This paper presents a multi-task learning scheme, which combines the physical model and the data-driven model with both merits. It enhances the data learning of a neural network with the aid of prior knowledge on atmospheric dispersion, and also controls the impact of the knowledge with a tunable weighting coefficient. Evaluations over a real-world deployment in Foshan, China show that, with the resolution of 500m×500m×15min, the proposed method outperforms the state-of-the-art ones with 7.9% error reduction and 6.2% correlation increase. Benefited from the physical knowledge, the neural network obtains stable performance with lower variance, as well as higher robustness against negative background conditions.

Object transportation could be a challenging problem for a single robot due to the oversize and/or overweight issues. A multi-robot system can take the advantage of increased driving power and more flexible configuration to solve such a problem. However, an increased number of individuals also changed the dynamics of the system which makes control of a multi-robot system more complicated. Even worse, if the whole system is sitting on a centralized decision making unit, the data flow could be easily overloaded due to the upscaling of the system. In this research, we propose a decentralized control scheme on a multi-robot system with each individual equipped with a deep Q-network (DQN) controller to perform an oversized object transportation task. DQN is a deep reinforcement learning algorithm, thus does not require the knowledge of system dynamics, instead, it enables the robots to learn appropriate control strategies through trial-and-error style interactions within the task environment. Since analogous controllers are distributed on the individuals, the computational bottleneck is avoided systematically. We demonstrate such a system in a scenario of carrying an oversized rod through a doorway by a two-robot team. The presented multi-robot system learns abstract features of the task and cooperative behaviors are observed. The decentralized DQN-style controller is showing strong robustness against uncertainties. In addition, We propose a universal metric to assess the cooperation quantitatively.

The power consumption at the receiver side will be dramatically increased in the millimetre-wave and massive multiple-input-multiple-output (MIMO) communication systems due to the wide bandwidth and a large number of antennas adopted. A half phase-only MIMO (HPO MIMO) scheme, in which the base station (BS) acquires $\pi $ -periodic phase measurements of the complex envelop signals was proposed very recently to overcome the above problem. Due to the non-linear nature of HPO MIMO, the valuation of the achievable rate is very challenging. The purpose of the paper is to provide an efficient method for calculating the achievable rate of the HPO MIMO system. By the mutual information theory, we transform the achievable rate into a sum of two high-dimensional integrations. However, calculating those integrations suffers from the enormous computational burden when using the traditional Monte-Carlo method. In order to increase efficiency, a new method by combining quasi-Monte Carlo with a variance reduction technique is proposed. Besides, we derive the probability density function (PDF) of the HPO MIMO system and analyze the uplink achievable rate of the HPO MIMO scheme. Numerical results show that our proposed method is efficient for calculating the achievable rate of the HPO MIMO system. With the proposed method we confirm that HPO MIMO is a promising technology in future low-power communication scenarios.

In recent years, marine exploration has become one of the most popular research subjects. At present, gliders in Underwater Acoustic Sensor Networks (UASNs) are equipped with a variety of sensors, which can play an important role in marine detection and monitoring. In addition, gliders also have the ability of data collection and storage. When coexisting with the sensor nodes, they can be regarded as special sensor nodes with semi-determined sawtooth trajectories. In hybrid UASNs including sensor nodes and gliders, a robust routing protocol is required to improve the poor network connectivity caused by long transmission delay, high bit error rate and unreliable transmission links. In this article, the Fuzzy Logic Algorithm (FLA) is used to convert multiple input parameters into one output value, thus reducing the storage burden of the network. Therefore, a glider-assist routing scheme with trajectory prediction is proposed to improve the connectivity of the hybrid network. The simulation results show that our scheme is superior to other protocols in terms of delivery ratio, end-to-end latency and lifetime.

This paper presents a novel algorithm for unsupervised video object segmentation (UVOS) in unconstrained scenarios. Although a large variety of methods have been proposed in the literature, segmenting generic objects is still challenging because different methods often perform well in different situations, and no single method can outperform the others in all cases. To address this, we propose to solve the problem of UVOS in a crowd-sourcing setting. We claim that one can achieve superior results by aggregating the predictions of multiple imperfect methods in a reasonable way. Specifically, we propose a latent regression algorithm for ensemble-based segmentation by jointly labelling pixels in a sequence and learning an adaptive weight for each single method in an ensemble. The pixel labellings offer the outcome (pseudo groundtruth) for regression and thus promote the procedure of weight learning, while the learnt weights could provide better shape priors for labelling, resulting in more accurate segmentation. Besides, Laplacian regularization is introduced into the regression to facilitate a stable learning of the weights. The most distinct feature of our algorithm is that it adaptively learns the contributions of different single methods for each test sequence, thus is capable of capturing the advantages of those methods while avoiding their weaknesses. In the experiments, our algorithm is built on 14 non-deep learning segmentation methods which are based on handcrafted features and require no training data. Experimental results on popular benchmarks show that our algorithm achieves compelling performance, even in comparison with deep learning-based methods. Furthermore, benefiting from the adaptive weight learning mechanism, our algorithm can achieve good flexibility and usability by choosing the most complementary single methods without losing too much performance.

One of the main challenges that mobile crowdsensing systems must solve is reducing data collection costs while still holding high data delivery probability. Compared with cellular networks, opportunistic networks can significantly reduce data transfer costs at the cost of damaging data delivery probability. This paper proposes an optimal data collection scheme for mobile crowdsensing, which utilizes integrated cellular and opportunistic networks to implement data collection. We use data collecting path to describe how the sensing data are collected and sent to the back-end platform, though cellular networks directly or through multi-hop opportunistic networks. An optimal data collection problem is then formulated as choosing specific data collecting paths from candidate path set to minimize the total crowdsensing cost under the data delivery constraints, which can be considered as a minimum set covering problem. To solve this NP-hard problem, we design and implement a greedy heuristic algorithm that constructs the solution in multiple steps by making a locally optimal decision in each step. We conduct extensive simulations based on three real-world traces: Cambridge, Infocom06, and UPB. The results show that, compared with other data collection approaches, our approach achieves a better tradeoff between cost and data delivery.

Safety message dissemination in a vehicular communication network requires vehicle-to-vehicle (V2V) communication with low latency and high reliability. To fulfill the stringent requirements in terms of reliability associated with safety applications, a control information assisted three-step (CIAT) resource selection scheme based on cellular vehicle-to-everything (C-V2X) technology for future vehicular communication network is investigated in this paper. To get a mathematical understanding of the process and performance of the CIAT scheme, in this paper, we develop an effective analytical approach to characterize the properties of the CIAT resource selection scheme in vehicular communication network by using the tool of stochastic geometry. In particular, the resource selection process and the key performances are analyzed mathematically; The closed-form expressions for the key performances are derived, which include the resource exclusion ratio, the average successful transmission probability and the packet inter-reception time; Furthermore, the performances of CIAT resource selection scheme are compared with those of random resource selection scheme mathematically. All the analytical results in this paper are validated through extensive Monte Carlo simulations, and the results provide insights to proper parameter settings for future vehicular communication network.

Low power and ultrafast multi-state storage resistive switching memory (RRAM) device had been developed based on Al/Al2O3/AlxOy/Al structure. Both of Al2O3 and Al nanocrystal (nc-Al) induced AlxOy thin films were deposited by RF sputtering. The nc-Al AlxOy based RRAM device showed typical unipolar switching behavior which was due to conductive filaments (CFs) connected and broke in Al2O3/AlxOy layers. An additional 30 nm Al2O3 thin film would deposit on AlxOy film to form bi-layer structure, in which the multi-state switching could be observed by applying different voltage pulses on it. In this study, a 15 V pulse with 600 ps width could trigger RRAM device switch from high resistance state(HRS)to next intermediate resistance state (IRS), the device could finally switch to LRS after continuous pulse simulation. Such switching from HRS to LRS was called “writing” process as data would be stored in RRAM device after this process. A longer but lower amplitude voltage pulse was required to make device switch from LRS to HRS which was called “erasing” process, as data would be eliminated after this process. The multi-state switching was corresponding internal switching between these IRSs during “writing” and “erasing” process. The multi-level resistances might be caused by partially formed CFs in Al2O3/AlxOy layers. The distribution of CFs could be controlled by controlling the shape of pulse voltage to achieve this multi-state storage. This bilayer structured RRAM device had good endurance and retention performances at both room and high temperatures.

As an essential and challenging technology of fault prediction and health management(PHM), fault prediction technology has been a research focus in the field of fault diagnosis. However, the current model-based fault prediction technology and data-driven fault prediction technology have some limitations, and it is difficult to effectively apply them in practice. Therefore, this paper combines the advantages of two kinds of fault prediction technology, sets the fault distribution function as the membership function of the adaptive fuzzy neural network based on the full analysis of the fault mechanism. The use of the fault distribution function to highly generalize the law of fault occurrence, and the strong self-learning ability of the neural network can effectively tap the potential fault information of the fault data, thereby using the fault distribution function to fit the fault data, and forming a set of membership functions by presetting a variety of membership functions, so as to expand the applicability of the proposed model in fault prediction. The experimental results show that the fault prediction model proposed in this paper has the advantages of high prediction accuracy, fast convergence speed and good applicability.

The beam pattern synthesis of antenna arrays is one of the most significant optimization problems in the electromagnetics community, and it affects the performance of wireless communication systems. In this article, we formulate the beam pattern optimization problems, and propose an improved invasive weed optimization with random mutation and Lévy flight (IWORMLF) algorithm to synthesize the beam patterns of linear antenna arrays (LAAs) and circular antenna arrays (CAAs). IWORMLF introduces the random mutation operator and the Lévy flight mechanism to enhance the efficiency and to balance the exploration and exploitation of the algorithm for optimization problems. Several simulations are conducted to evaluate the performance of the proposed approach. First, IWORMLF is tested by a variety of benchmark functions, and the results indicate that it achieves the best performance in most of the functions compared to some other algorithms. Second, IWORMLF is utilized for the maximum sidelobe level (SLL) reduction and the joint maximum SLL and mainlobe beamwidth reductions. The results show that it achieves better performances in terms of the convergence rate, stability, and accuracy compared to some evolutionary algorithms for these optimization cases. Third, the efficiencies of the improved factors are verified. Finally, electromagnetism simulations are conducted to evaluate the performance of IWORMLF for the beam pattern optimizations with considering mutual coupling.

Structural reliability analysis is the key approach to assess uncertainties so that to increase the safety of engineering structures. Quantifying the failure probability (FP) is central to direct the result of the reliability assessment. In aerospace and military fields, normal samples collected from a structure are easily available, however, failure samples are extremely limited. Such imbalanced samples circumstance may lead to a large approximate bias of the failure probability. The critical problem in structural reliability analysis is how to use a smaller number of samples to get more precise failure probabilities. Although the Monte Carlo simulation (MCS) is the recognized benchmark, the high computational cost of calling limit state function has forced people to seek alternative ways. On the other hand, the surrogate model method, such as the adaptive Kriging model (abbreviated as AK-MCS), has been proposed to reduce the computational burden. To evaluate small failure probability, however, the number of the candidate points must be large for a convergent solution. To this end, this paper constructs a security domain identified model (SDIM) based on the one-class support vector machine (SVM) and imbalanced data. Different types of misjudged samples and approximate errors of the trained SDIM are analyzed. Two schemes are proposed accordingly to reduce the tested estimate error of the failure probability. Comparing with MCS and AK-MCS, the numerical and engineering examples demonstrate the accuracy and efficiency of the proposed method under different scenarios.

Metasurfaces become increasingly important for wavefront shaping and beam steering with high efficiency. We propose a wide-angle beam steering device based on all-dielectric metasurfaces consisting of two symmetrical blazed gratings for optical trapping of particles, which can convert the input Gaussian beam to a special beam with intensity gradient along the propagation direction. Two major types of metasurfaces (Pancharatnam-Berry phase and nanoposts) are compared, and dramatic difference is found between them, especially at oblique incidence. We show that the Pancharatnam-Berry phase-based metasurface is more tolerant to the incident angle and that significant optical gradient force can be generated over a large angle range of -50° to 50°, and the maximum attractive force is up to tens of pN/W, with a trapping range of 14 gym for particles of 2.5 μym in diameter. The proposed scheme exhibits great potential to trap a large fraction of particles floating in a microfluidic channel when the beam is dynamically steered.

Decentralization of grid systems plays an important role in improving their efficiency and fault tolerance. To enhance the performance and stability of grid and mitigate the problems of centralized grid, an agent-based decentralized gird model (ADGM) with universality and functional integrity is proposed. In this paper, we build an agent-based grid structure and propose an agent-based grid consensus algorithm (AGCA). A group membership protocol, a consistency protocol and a view change protocol in AGCA are also designed. Furthermore, based on Pi calculus, the information registration, resource sharing and error recovery of the grid model in the decentralized environment are formally specified. Finally, we analyze the performance of the AGCA algorithm and compare it with other consensus algorithms, the simulation results demonstrate that the AGCA algorithm achieves better comprehensive performance and more symmetrical time performance, space performance and fault tolerance than other consensus algorithms.

We conduct a research on free space optical (FSO) communication system that is applied for transmission in wireless sensor networks (WSN), which is based on the hybrid pulse position modulation-binary phase shift keying-subcarrier intensity modulation (PPM-BPSK-SIM) undergoes novel Exponentiated Weibull fading channels. When battery-charged sensor transmission nodes with limited energy is transmitted, it is vital to study the performance of the transmission link. We have derived the exact joint probability function of the transmission link when the atmospheric turbulence and pointing errors are considering. On the basis of them, the unconditional average bit error rate (BER) for hybrid PPM-BPSK-SIM is derived, then the closed expression of the outage probability and the average channel capacity are also derived. Research results indicates that increasing the receiver aperture by aperture averaging effect can distinctly improve the performance of the link. Additionally, for any circumstances of atmospheric turbulence and pointing errors, by combining the utilization of hybrid PPM-BPSK-SIM modulation and symbol with average length greater than eight, the average BER performance can be conspicuously enhanced. The outage probability and average channel capacity of the FSO link are jointly affected by atmospheric turbulence and pointing error, and optimization strategies in different scenarios are given by our research.

Based on the temperature characteristics of the air, ice and water in the vertical direction of the river ice cover, an automatic temperature measuring instrument for ice cover profile was designed. The sensor used NTC thermistor as a temperature detection unit. In order to improve the actual detection accuracy of the sensor units in a wide temperature range [223.15 K (-50°C), 293.15 K (20°C)], a calibrating method of the thermistor using the resistance residual compensation method was proposed and compared with the existing seven calibration methods. The results show that the accuracy of the calibrated temperature sensor units can reach 0.1 K in the testing range, which can satisfy the accuracy requirements in thermodynamics research of ice cover profile. We further designed a temperature measuring instrument and carried out a deployment of the instrument and in-situ observation of data in cold region of the upstream of Heilongjiang River in late December 2017. The evolution process of ice thickness was analysed and discussed by the collected data. It can be proved that this instrument is an effective scheme for intelligent detection of river ice environment in cold regions.

A silicon nitride horizontal slot waveguide with silicon dioxide cladding is designed to achieve flat negative dispersion (-3 W. After passing through another ideal positive dispersion element, we find that the output pulse peak power can be recompressed close to 1 W in the case of different input pulse widths. Moreover, for the input pulse with a much higher peak power, the pulse width of the output pulse can be compressed to a much narrower level because of the spectral broadening introduced by optical nonlinearity. A 100-fs input pulse with 10-kW peak power can be compressed to 12.5 fs with a peak power of 63 kW. Considering the time-domain broadening and the peak power reduction, such a slot waveguide with a flat negative dispersion could be used as a time stretcher for 10-fs level pulse in on-chip chirped pulse amplification (CPA) system.

In this paper, a broadband transparent microwave absorber with a fractional bandwidth of 118.1% was designed. In the proposed absorber, a gear-shaped unit cell consisting of optically transparent indium tin oxide-polyethylene terephthalate films fabricated on a polymethyl methacrylate is developed to guarantee electromagnetic wave to well enter into and is dissipated in the absorbing structure. Two resonant frequencies at 4.8 GHz and 13.6 GHz, which correspond to electromagnetic and electric resonances, respectively, are generated to achieve a wide absorbing band from 3.86 GHz to 15.04 GHz covering S, C, X and Ku bands. The proposed absorber has the advantages of low profile, polarization insensitivity, and angular stability. The prototype of the absorber has been fabricated and measured, and the experimental results are in good agreement with the simulation results.

Given the fusion problem of uncertain information on the time domain, this paper proposes a time-domain evidence fusion method involving the time sequence of the decision-maker based on evidence theory to give full expression to the influence of dynamic features and the time factor of time-domain information fusion on the integration. Time sequence preference of the decision maker is integrated into the time domain evidence fusion. The time sequence preference of the decision-maker is measured by analyzing the features of time domain evidence sequence based on the definition of time sequence memory factor. An optimization model is constructed to solve the time sequence weight. In addition, the concept of the evidence belief degree is combined to amend the evidence source. Finally, the amended evidence is integrated using the Dempster combination rule. Compared with the fusion method without the time factor, numerical results indicate that the evidence fusion method considering the time preference of decision maker can effectively handle the conflict information in the time-domain information sequence, and reasonable fusion results can be obtained. The results of comparative analysis indicate that the proposed method, in which the belief degree of time domain evidence sequence and the subjective preference of decision maker are fully considered, can demonstrate the influence of subjective factors of the decision maker on the time domain evidence fusion and better manifest the dynamic feature of time domain evidence fusion.

The planar three-link passive-active-active (PAA) under-actuated manipulator does not satisfy the small-time local controllability (STLC), which makes it more difficult to control than other planar under-actuated manipulators. This paper presents a two-stage switching hybrid control method based on improved particle swarm optimization (PSO) for the PAA under-actuated manipulator. According to the model reduction strategy, the control process is divided into two stages. The workspace of the manipulator is analyzed to guarantee the target position within the reachable region. In order to reduce the probability of falling into local optimum, we introduce the Metropolis criterion of simulated annealing algorithm and crowding factor of artificial fish swarm algorithm into PSO algorithm to calculate the target angles of two active joints. At each control stage, we design the Lyapunov function to maintain the angle of an active joint unchanged while we build the sliding surface and select the power function as reaching law to make the other active joint converge to the target angle. The control law switches to next stage when the rotated joint converges to the target angle. The proposed method not only retains the rapidity of sliding mode control, but also reduces the total chattering of the system by the Lyapunov function method. Simulation results demonstrate that the proposed method has shorter steady-state adjustment time than that of existing typical control methods, and its overshoot is almost zero.

Non-orthogonal multiple access (NOMA) is envisioned as a key technology to enhance the spectrum efficiency for 5G cellular networks. Meanwhile, ambient backscatter communication is a promising solution to the Internet of Things (IoT), due to its high spectrum efficiency and power efficiency. In this paper, we are interested in a symbiotic system of cellular and IoT networks and propose a backscatter-NOMA system, which incorporates a downlink NOMA system with a backscatter device (BD). In the proposed system, the base station (BS) transmits information to two cellular users according to the NOMA protocol, while a BD transmits its information over the BS signals to one cellular user using the passive radio technology. In particular, if the BS only serves the cellular user that decodes BD information, the backscatter-NOMA system turns into a symbiotic radio (SR) system. We derive the expressions of the outage probabilities and the ergodic rates and analyze the corresponding diversity orders and slopes for both backscatter-NOMA and SR systems. Finally, we provide the numerical results to verify the theoretical analysis and demonstrate the interrelationship between the cellular networks and the IoT networks.

Nonnegative matrix factorization, as a classical part-based representation method, has been widely used in pattern recognition, data mining and other fields. However, the traditional nonnegative matrix factorization directly factoring decomposes the original data, and the original data often contains a lot of redundancy and noise, which seriously affect the subsequent processing of the data. In this work, we propose an adaptive graph regularization discriminant nonnegative matrix factorization (AGDNMF) for image clustering. The AGDNMF algorithm makes full use of local structure information and a small amount of label information. In AGDNMF, the local structure information can be more accurate and the label information can prevent the points with the same label from being merged into one point. These two items are combined into the objective function of NMF. In addition, we provide the update rules for the corresponding optimization functions and prove its convergence. A large number of experiments on different data sets show that the proposed algorithm has good clustering performance.

In this paper, we are interested in a symbiotic radio (SR) system, in which a passive full-duplex backscatter device (BD) is parasitic in an active primary transmission. The primary transmitter (PT) with multiple antennas is designed to broadcast common messages to the primary receiver (PR) and the BD, as well as to support passive information transmission from the BD to the PR. To do so, the full-duplex BD absorbs a fraction of the incident signal from the PT to decode the common messages and simultaneously transmits its own information to the PR by backscattering the remaining part of the incident signal. We derive the achievable rates of the BD transmission with Gaussian and quadrature amplitude modulation codewords. We also formulate a transmit power minimization problem by jointly designing the beamforming vector at the PT and the power splitting factor at the BD. This problem is first solved by the semi-definite relaxation technique together with a one-dimensional linear exhaustive search over the power splitting factor. Then, a suboptimal but low-complexity solution with closed-form expressions is proposed. The simulation results have shown that the proposed suboptimal solution achieves almost the same performance as that obtained by the exhaustive search. In addition, our proposed SR system with the full-duplex BD outperforms the half-duplex system with the time-division-multiplexing mode in general.

This paper concerns the control algorithm oriented to nonparallel-ground-track-imaging for agile optical satellite. Firstly, for obtaining the desired trajectory, the mapping relationship between the satellite attitude and the ground stripe is established by using space vector method. According to the mapping function, an attitude adjustment strategy via constant scanning velocity is proposed. Then, considering the exact information of the external disturbance cannot be obtained and the input saturation problem, a robust anti-saturation tracking controller is designed via fixed-time disturbance observer and integral sliding mode. Finally, the effectiveness of the proposed control algorithm is illustrated by numerical simulation.

In this paper, an adaptive dual-threshold sparse Fourier transform (ADT-SFT) algorithm is proposed, which enables the application of the SFT and robust SFT (RSFT) to the moving target detection in clutter background. Two levels of detection are introduced in this algorithm. First, a scalar constant false alarm rate (CFAR) detection is employed in each frequency channel formed by subsampled fast Fourier transform (FFT) to suppress the influence of strong clutter points on the sparsity and frequencies estimation. Second, the subspace detector constructed by suspected target Doppler frequencies is adopted to complete the target detection. The simulation analysis and results of the measured sea clutter data show that the ADT-SFT algorithm is more suitable for the clutter background and can obtain better detection performance than SFT and RSFT. In addition, compared with the conventional subspace detection (SD) algorithm, which needs to search all the Doppler frequencies one-by-one to establish the detector, the ADT-SFT algorithm only needs to search a small number of suspected target Doppler frequencies, and therefore, the computational complexity can be greatly reduced.

Automatic steering system of self-driving vehicles has significant influence on vehicle safety. The performance of automatic steering system is affected by external disturbance, parameter perturbation, and real-time ability of control algorithm. First, to develop a high-performance automatic steering control strategy, a vehicle-road system model based on vehicle dynamics and kinematics is established. Then, the system state equation with direction-angle and lateral-position deviations is derived to describe the tracking accuracy. Finally, based on the sliding mode variable structure control method, an automatic steering control algorithm is proposed and verified. The results show that the dynamic quality of the system is improved with the optimization of the sliding mode. Furthermore, good robustness against vehicular velocity, real-time performance, and tracking accuracy is achieved.

Extracting valuable information to enhance the performance of forecasting models from the imbalanced and big data requires the scalable implementation of advanced statistical learning methods. This paper proposes the online mixture model (OMM) and applies it to the Mach number forecasting. Treating the key variable (e.g., Mach number) forecasting under all working conditions as an entire task, and viewing that of each individual working condition as a subtask, the OMM separates the dense samples from the sparse ones on the basis of subtasks. The subtask models are independently learnt on the samples with reduced volume, and updated for the new working conditions without retaining samples from the old working conditions. Moreover, the tree-structure ensemble (TSE)-feature subsets ensembles (FSEs) algorithm is presented to fit the nonlinear function of a subtask model, where the FSE local models with low-dimensional input features are established on the non-overlapping sample subsets constructed by the TSE method. The TSE-FSEs not only reduce the volume of data but also perform distributed computing with parallel structure, and thus has the advantage of the learning of big data. Experiments carried out on the measurement data of wind tunnel indicate that the OMM with the TSE-FSEs outperforms other learning algorithms for the Mach number forecasting, and meets the precision and forecasting speed requirements in engineering.

This paper investigates the finite-time consensus tracking problem for nonstrict feedback nonlinear multi-agent systems with unknown dynamics and control directions. Firstly, the backstepping scheme based on command filter and fractional power control law is proposed, which can not only solve the problem of computational explosion, but also ensure that the state of each agent can follow the leader's output within a finite time. Then, the error compensation signals are designed to compensate the error caused by filtering. The unknown nonlinear dynamics are approximated by the fuzzy logic system. The state variables problem in the non-strict feedback system is well solved through the scaling of inequalities. By introducing the Nussbaum function, the intermediate input signal and control input signal based on adaptive control law are designed respectively, and the problem of unknown control directions is solved. Simulations are given to show the effectiveness of the presented method.

With the development of technologies, such as big data, cloud computing, and the Internet of Things (IoT), digital twin is being applied in industry as a precision simulation technology from concept to practice. Further, simulation plays a very important role in the healthcare field, especially in research on medical pathway planning, medical resource allocation, medical activity prediction, etc. By combining digital twin and healthcare, there will be a new and efficient way to provide more accurate and fast services for elderly healthcare. However, how to achieve personal health management throughout the entire lifecycle of elderly patients, and how to converge the medical physical world and the virtual world to realize real smart healthcare, are still two key challenges in the era of precision medicine. In this paper, a framework of the cloud healthcare system is proposed based on digital twin healthcare (CloudDTH). This is a novel, generalized, and extensible framework in the cloud environment for monitoring, diagnosing and predicting aspects of the health of individuals using, for example, wearable medical devices, toward the goal of personal health management, especially for the elderly. CloudDTH aims to achieve interaction and convergence between medical physical and virtual spaces. Accordingly, a novel concept of digital twin healthcare (DTH) is proposed and discussed, and a DTH model is implemented. Next, a reference framework of CloudDTH based on DTH is constructed, and its key enabling technologies are explored. Finally, the feasibility of some application scenarios and a case study for real-time supervision are demonstrated.

This paper concerns the optimal control problem of a general affinely pseudo-linearized nonlinear system and puts forward the solving scheme based on adaptive dynamic programming and neural network. In detail, a new more general affinely pseudo-linearized nonlinear system model is presented and covers more practical scenarios than existing models in engineering. In consideration of the Bellman optimal principle, the corresponding model particulars are then analyzed and its adaptive dynamic programming solving scheme is derived. That is, the modified policy iteration is used to estimate the value function and optimal control at each sampled state, and then the calculated state-control pairs and state critic values are utilized to train two neural networks (one works as state evaluator and the other works as optimal control generator, respectively). Through the above, the discrete state space is smoothed into a continuous state space, and the notorious dimensional curse problem could be addressed. Moreover, in order to improve the efficiency and precision of the network training, the samples calculating and network training processes are simultaneously carried out in this paper. In the end, a related simulation experiment is applied and the results demonstrate that the proposed method has more effectiveness and validation for the optimal control problem of the newly proposed general system model than the compared methods.