Search

The development of intelligent robot has always been an important research direction in the field of artificial intelligence, and the object detection of robot is the basis of intelligent perception and autonomous action. This study proposes an improved object detection algorithm which integrates two kinds of intelligent robot object detection techniques. In this process, the relationship between the center position of the target in the frame used for object detection and tracking is analyzed. It uses Laplace probability density as the parameter to calculate the center position relationship, and joins the correction network to correct the feature point positioning. The object re-capture function is added to solve the problem of object loss in the long-term object detection task, and the classifier is used to realize the object recognition. The results show that when the number of targets in the image reaches 20, the capture accuracy of the two data sets remains above 98.7%. In the intersection to union ratio test, when the real rectangular box contained 2M pixels, the intersection to union ratio of the method proposed in this study remains at or above 0.989. When conducting actual application memory usage tests, the proposed method maintains a memory usage of less than 2000Mb at runtime. It is shown that this method has better target detection efficiency and quality, and the requirement of hardware is lower.

Existing high-power semiconductor lasers widely use asymmetric large cavity structures or designs with short cavity lengths to enhance device performance, inevitably increasing the device threshold current density and suppressing further improvement of device performance. We have utilized rate equations analysis to investigate the impact of threshold current density on the threshold characteristic temperature of GaAs-based laser diodes at short wavelengths. A simplified carrier escape model has been employed, which is more accurate compared to the classical model of thermal carrier emission. Compared to long-wavelength laser diodes, short-wavelength laser diodes exhibit higher temperature stability due to their smaller Auger recombination coefficients. Increased internal losses caused by carrier escape and decreased injection efficiency are the main reasons for the decrease in threshold characteristic temperature with increasing temperature. The increase in threshold significantly enhances carrier escape, leading to a decrease in the threshold characteristic temperature and an increase in sensitivity to temperature. Moreover, high thresholds limit the increase in the device's decoupling factor, causing a decrease in current inject efficiency and further degradation of device performance. By increasing the barrier height, carrier escape can be effectively suppressed, thereby enhancing device performance. The simulation results agree well with the experimental data of InAlGaAs laser diodes at a wavelength of 780 nm, accurately predicting the threshold characteristics of devices with different cavity lengths at various temperatures.

In recent years, with the widespread popularity of the Internet, social media has become an indispensable part of people’s lives. People regard online social media as an essential tool for interaction and communication. Due to the convenience of data acquisition from social media, mental health research on social media has received a lot of attention. The early detection of psychological disorder based on social media can help prevent further deterioration in at-risk people. In this paper, depression detection is performed based on non-verbal (acoustics and visual) behaviors of vlog. We propose a time-aware attention-based multimodal fusion depression detection network (TAMFN) to mine and fuse the multimodal features fully. The TAMFN model is constructed by a temporal convolutional network with the global information (GTCN), an intermodal feature extraction (IFE) module, and a time-aware attention multimodal fusion (TAMF) module. The GTCN model captures more temporal behavior information by combining local and global temporal information. The IFE module extracts the early interaction information between modalities to enrich the feature representation. The TAMF module guides the multimodal feature fusion by mining the temporal importance between different modalities. Our experiments are carried out on D-Vlog dataset, and the comparative experimental results report that our proposed TAMFN outperforms all benchmark models, indicating the effectiveness of the proposed TAMFN model.

To inhibit commutation failure (CF), in this paper, the influence factors of CF were analyzed, and considering different fault types and different valves, the mathematical expression of extinction angle was deduced. In addition, by analyzing existing control strategies for inhibiting CF, it is indicated that existing control strategies cannot quickly and accurately obtain the AC voltage amplitude and zero-crossing phase shift angle. To solve this problem, this paper proposed a fast AC voltage measurement method, and the zero-crossing phase shift angle can be calculated by AC voltage amplitude. On this basis, a commutation failure predictive control method (CFPCM) was proposed to inhibit CF. The CFPCM can adapt to all fault types. Finally, based on the CIGRE HVDC benchmark model, the commutation failure prevention module (CFPREV) and CFPCM simulation model were built using PSCAD/EMTDC to verify the effectiveness of the proposed CFPCM. The simulation results showed that the CFPCM can not only inhibit CF more effectively, but also reduce the fluctuation of DC current, AC voltage, and extinction angle with a fault occurred in the inverter AC system.

In the process of acquiring 3D point cloud data, due to environmental interference or unstable scanning equipment, the acquired data often have noisy points. Recently, with the development of neural networks for point clouds, great progress has been made in deep learning-based point cloud denoising. However, most of the existing methods adopt a pointnet-like structure to predict point offsets. Simple pooling operation loses much important information, such as local neighborhood information, and global information. The loss of information makes the algorithm ineffective when dealing with some complex cases. In order to solve the above problems, we propose a self-attention-based point cloud denoising network architecture, through the Transformer structure, to establish long-range dependencies of the points. In addition, we propose a local information embedding module to fast select meaningful points and serve as the input of the Transformer. We also consider the correlation between channels of point cloud features and further introduce a channel attention module. Extensive experiments prove that our method outperforms existing methods, and maintains a high running speed.

Multi-microgrids (MMGs) suffer from power shortages due to the loss of utility grid support when an unintentional transition occurs. This imposes a transient shock on the system voltage and frequency. To maintain the frequency stability and power balance of an islanded MMG, this paper presents an underfrequency load shedding (UFLS) strategy with adaptive variation. A comprehensive load evaluation method based on a composite least squares support vector machine (CLS-SVM) is proposed to ensure uninterrupted power for critical loads. This method considers the comprehensive evaluation influence factors (CEIFs) of loads. Then, a least squares support vector machine (LS-SVM) provides the load shedding determination factors, transforming the problem of determining critical loads into a 0-1 planning problem. A method with adaptive variation is proposed to solve the UFLS model. The effectiveness of the proposed strategy is verified for an MMG model based on a modified IEEE 33-bus system. The test results show that: 1) the average accuracy of the proposed method is 21.05% higher than that based on LS-SVM; 2) compared with UFLS strategies based on the load level alone and on an intelligent algorithm, the frequency fluctuation range of the proposed strategy is 12.50% and 19.23% lower, respectively, and the frequency recovery time is 3.90% and 5.73% shorter, respectively; 3) compared with PSO, GOA and GA, the standard deviation of the iterative mean of the proposed algorithm decreases by 36.22%, 53.42%, and 34.00%, respectively. The proposed strategy can reduce the frequency fluctuation range and frequency recovery time while maintaining strong adaptability.

The maximum total complex correntropy (MTCC) algorithm improves the performance of adaptive filtering under the error in variable (EIV) model by integrating both input and output noise information into the total complex correntropy. However, the MTCC algorithm cannot be applied to the widely linear model, directly. Compared with a strictly linear model, the widely linear model provides the complete statistical information of signals for adaptive filtering. This paper first constructs a novel cost function for the widely linear model by incorporating the input and output noise information into the improper complex correntropy. Then, a novel widely linear adaptive filter algorithm named widely linear maximum total improper complex correntropy (WL-MTICC) is proposed using the gradient descent method under the maximum total improper correntropy criterion. Finally, the analysis of local stability and convergence in the mean sense regarding the proposed WL-MTICC algorithm is provided. Simulations are used to show the performance advantage of the proposed WL-MTICC algorithm in the presence of both input and output noises.

With the continuous increase of offshore wind farms and other offshore regional buildings, the contradiction between them and the navigation safety in the densely distributed areas of offshore ship tracks is increasingly obvious. It is urgent to build a safe distance model based on the actual sea condition to reduce the mutual influence. Based on the drift distance of ships in the boundary water areas of the wind farm under the actual meteorological and hydrological conditions, as well as the tolerable collision probability, reliability model and the normal distribution trend of ships in the ship routes, and the probability model of safe distance between ship routes and wind farm is built in the end. Based on the example and the tolerable collision probability of ships, the safe distance is analyzed, the analysis results show that the calculation model can determine the safe distance according to the control requirements of the collision probability between the ship and the offshore wind farm. In addition, the collision probability and the safe distance obtained are basically consistent with the safe distance published by the local maritime authority in combination with the changes of the output results caused by the velocity difference of the ship. The model lays a theoretical foundation on offshore wind farm site selection optimization, route boundary demarcation, ship safety and maritime supervision.