Search

This article proposed an analytical synthesis approach of generalized strongly coupled resonator triplet (GSCRT) filters, which eliminates the need for negative couplings regardless of the location of transmission zeros (TZs). The principle of removing negative couplings in GSCRTs is explained from circuit level for the first time. In contrast with conventional filter function synthesis with all in-band reflection zeros (RZs), some RZs are preassigned out of the passband, which are named redundant resonant modes (or even modes). By regulating the frequency of the redundant modes, flexible synthesis solutions can be achieved and the previously published strongly coupled resonator triplet (SCRT) becomes a special case of this new generalized version. Besides, a novel physical interpretation of GSCRT from the basis of modes is presented, which shows the modes routing clearly within the coupling matrix framework. Due to the proposed theories, additional approximation and optimization can be avoided during the synthesis procedure. Due to the redundant mode, all couplings in the GSCRT are positive, thus easing the design and manufacturing process. Also, GSCRT brings compact and in-line configuration of filters. Moreover, due to the higher flexibility provided by redundant resonant modes, reasonable synthesis solutions can be selected depending on the specifications or other physical limitations, such as bandwidth and size. A group of synthesis examples with simulations are elucidated. The simulated results show a good agreement with the synthesis theories. The simulated and measured results of an anti-interference C-band eight-pole filter with four TZs below the passband further validate the effectiveness of the proposed theories. [ABSTRACT FROM AUTHOR]

The generalized Chebyshev (pseudoelliptical) filter functions may not be optimal for certain applications due to physical constrains. With redundancy in filter polynomials, this article breaks down the “mental barrier” in conventional synthesis and directly synthesizes the “Chebyshev-like” functions to achieve desired filter performance under physical limitations. By exploiting the redundant restrictions on exact equal-ripple response, the reflection zeros can be moved to the complex plane, thus forming the “Chebyshev-like functions.” A class of inline filters with a block of second-order dangling resonators is synthesized, which can generate a pair of symmetric transmission zeros (Tzs) but is considered unrealizable using standard Chebyshev functions. With the proposed theories, an accurate and much improved filter prototype is directly derived from the “Chebyshev-like” polynomials, avoiding additional optimization in the design process. For verification, a group of examples with synthesis results are elucidated. Eight ten-pole filters with four Tzs are synthesized and fabricated. The simulated and measured results demonstrate the effectiveness and application of the proposed theories. [ABSTRACT FROM AUTHOR]

This article proposes a novel approach for extracted pole (EP) filter synthesis. The pole producing elements can be extracted without introducing phase shifters. It is made possible by extracting lowpass elements in the prototype without reducing the order of filter functions. Using the proposed elements as added redundancy, the nonideal effect in the physical model is absorbed by adjacent resonators. Those novel approaches break through the conventional framework in EP filter synthesis. The synthesized results do not introduce extra spikes caused by nonideal phase shifters in the conventional EP filters. The new theory enables the application of EP filters in broadband situations. This work also reveals the equivalence between the frequency-dependent couplings and the PPEs. The validity of the demonstrated method is illustrated via an 11-pole filter with two transmission zeros. The simulated and measured results have no extra spikes caused by the nonideal phase shifters, which verified the effectivity of the proposed theory. [ABSTRACT FROM AUTHOR]

Gradient descent optimization of learning has become a paradigm for training deep convolutional neural networks (DCNN). However, utilizing other learning strategies in the training process of the DCNN has rarely been explored by the deep learning (DL) community. This serves as the motivation to introduce a non-iterative learning strategy to retrain neurons at the top dense or fully connected (FC) layers of DCNN, resulting in, higher performance. The proposed method exploits the Moore-Penrose Inverse to pull back the current residual error to each FC layer, generating well-generalized features. Further, the weights of each FC layers are recomputed according to the Moore-Penrose Inverse. We evaluate the proposed approach on six most widely accepted object recognition benchmark datasets: Scene-15, CIFAR-10, CIFAR-100, SUN-397, Places365, and ImageNet. The experimental results show that the proposed method obtains improvements over 30 state-of-the-art methods. Interestingly, it also indicates that any DCNN with the proposed method can provide better performance than the same network with its original Backpropagation (BP)-based training. [ABSTRACT FROM AUTHOR]

This article presents a fully integrated 12-channel multiplexer with TE011 mode channel filters enabled by the unified extracted pole technique. A novel synthesis procedure is proposed to remove the limitation (a very small coupling) in the prototype with the classical approach. In addition, this article proposes a novel approach to assign the junction susceptance by simply shifting the length of the connecting waveguide section of the dangling resonators. The design process is then extended to the singly terminated case as the channel filter design. This work also proposes a novel analytical approach to derive the initial connecting phase lengths for the manifold multiplexer design. The circuit level optimization and space-mapping procedure for deriving physical dimensions are illustrated. The multiplexer is designed and constructed using a new clamshell structure, which offers low cost, small footprint, easy fabrication, and superior thermal performance. The simulated and measured results demonstrated that the proposed theory offers an excellent solution to the high-power waveguide manifold multiplexer designs. [ABSTRACT FROM AUTHOR]

This article studied the interference between two channels of a tunable diplexer and proposed a feasible solution to alleviate the effect. A novel approach to synthesize the channel filter with the first element replicating the other channel’s effect is first developed. It led a set of simple formulae for fixed frequency diplexer design. After that, a framework incorporating a redundant coupling path to generate cancellation effect is derived using the coupling matrix theory. The proposed theories are demonstrated and verified using a combline diplexer design. The simulated and measured results suggest that the interference of the two channels can be greatly mitigated. [ABSTRACT FROM AUTHOR]

This contribution establishes the relationship between the complementary condition and the singly terminated (ST) networks. The general theory is first presented in the form of bandpass/bandstop complementary diplexers. Furthermore, the well-known limitation of a complementary diplexer, i.e., using maximally flat prototypes, is overcome. From the first principle, it is rigorously proven that predefined complementary features can be obtained by connecting two channel filters in an ST prototype of the same filter function but exchanging the transfer and reflection part. This paper leads to a general design procedure for bandpass and bandstop complementary diplexers using arbitrary functions. A complementary diplexer design example with full electromagnetic simulated results and measured results verified the accuracy of the proposed theory. Finally, the theory is further extended to N-ports networks. A new “partially-complementary” condition is proposed. The new theory leads to an extremely simple and “direct approach” to estimate the performance of contiguous multiplexers without any optimization. [ABSTRACT FROM AUTHOR]

The extreme learning machine (ELM), which was originally proposed for “generalized” single-hidden layer feedforward neural networks, provides efficient unified learning solutions for the applications of regression and classification. Although, it provides promising performance and robustness and has been used for various applications, the single-layer architecture possibly lacks the effectiveness when applied for natural signals. In order to over come this shortcoming, the following work indicates a new architecture based on multilayer network framework. The significant contribution of this paper are as follows: 1) unlike existing multilayer ELM, in which hidden nodes are obtained randomly, in this paper all hidden layers with invertible functions are calculated by pulling the network output back and putting it into hidden layers. Thus, the feature learning is enriched by additional information, which results in better performance; 2) in contrast to the existing multilayer network methods, which are usually efficient for classification applications, the proposed architecture is implemented for dimension reduction and image reconstruction; and 3) unlike other iterative learning-based deep networks (DL), the hidden layers of the proposed method are obtained via four steps. Therefore, it has much better learning efficiency than DL. Experimental results on 33 datasets indicate that, in comparison to the other existing dimension reduction techniques, the proposed method performs competitively better with fast training speeds. [ABSTRACT FROM AUTHOR]

This paper proposes a novel and unified synthesis approach and design procedure for microwave filters with extracted pole and nonresonating node (NRN) technique. By introducing a general circuit topology, the traditional extracted pole and NRN filters can be unified to a general prototype. With the new design process, the exact mapping relationship between the electromagnetic (EM) model of the pole producing element and the circuit model can be achieved. All the physical dimensions of the filter structure can be obtained by a fully analytical approach. No full EM optimization or adjustment is needed in the filter design. A three pole TE \mathrm { {101}} mode waveguide filter with one finite transmission zero is designed to illustrate the new procedure. In addition, a seven-pole TE \mathrm { {011}} mode filter with four finite transmission zeros covering all three possible types of arrangements is designed and fabricated to demonstrate the versatility of the proposed method. The simulated and measured results verified the efficiency and the accuracy of the proposed design method. [ABSTRACT FROM AUTHOR]

High-frequency surface wave radar (HFSWR) has become the cornerstone of maritime surveillance because of its low-cost maintenance and coverage of wide area. However, when it comes to the extraction of key areas, such as vessel-target detection and vessel-path tracking, the HFSWR signal is strongly interfered by clutters and noise, which makes maritime surveillance a challenging task. This article proposes a hierarchical key-area perception model for maritime surveillance harnessing range-Doppler (RD) image from HFSWR, Laplacian kernel, a linear classifier (LC), and a subnet-based multilayer representation learning framework (SMRLF). First, a weak LC with a Laplacian kernel is utilized to capture the plausible vessel regions (PVRs). Then, a novel SMRLF is proposed to localize the vessel targets from the PVRs. To handle the noise, a maximum correntropy criterion with variable centers (MCC-VC) is incorporated in the subnet-based learning model. A thorough experimental analysis on cross-domain samples from radar dataset to scene classification dataset shows that the proposed HKPM performs competitively. The model shows a superior performance over most of the state-of-the-art vessel-target detection algorithms with a vessel-target detection accuracy of 94%. The extended analysis on image classification problem proves that the proposed model has great adaptivity and scalability. [ABSTRACT FROM AUTHOR]

This article offers a novel approach to realize dual-polarized diplexers (DPDIPs), which can duplex horizontally and vertically polarized signals separately. The proposed device is composed of a turnstile junction (TJ) and four-channel filters. Compared with conventional orthomode transducer (OMT) and two diplexers scheme, it has an ultracompact size and comparable performance. A new arrangement for the TJ that is suitable for the DPDIP is analyzed and simulated accordingly. A Ka-band DPDIP is designed and fabricated as a verification example. Simulated and measured results have verified the accuracy and efficiency of the proposed method. [ABSTRACT FROM AUTHOR]

This article proposes a computer-aided tuning approach for Butler matrices (BMs). It is implemented in two stages. First, extract the amplitude imbalance of each individual coupler from the overall BM measurement results using the proposed diagnosis procedure. The extracted results led to the instructions for fine adjustments. After that, each coupler is adjusted using the proposed method. As a result, the nonideal effect of a BM can be tuned out to achieve significant performance improvement. An $8 \times 8$ BM was designed, fabricated, and tuned using the proposed approach. The final results show excellent amplitude imbalance performance, which is improved by more than 80% relative to the initial results. For the first time, a systematic tuning method is demonstrated to achieve excellent amplitude imbalance for a BM. [ABSTRACT FROM AUTHOR]

Human action recognition (HAR) is one of most important tasks in video analysis. Since video clips distributed on networks are usually untrimmed, it is required to accurately segment a given untrimmed video into a set of action segments for HAR. As an unsupervised temporal segmentation technology, subspace clustering learns the codes from each video to construct an affinity graph, and then cuts the affinity graph to cluster the video into a set of action segments. However, most of the existing subspace clustering schemes not only ignore the sequential information of frames in code learning, but also the negative effects of noises when cutting the affinity graph, which lead to inferior performance. To address these issues, we propose a sequential order-aware coding-based robust subspace clustering (SOAC-RSC) scheme for HAR. By feeding the motion features of video frames into multi-layer neural networks, two expressive code matrices are learned in a sequential order-aware manner from unconstrained and constrained videos, respectively, to construct the corresponding affinity graphs. Then, with the consideration of the existence of noise effects, a simple yet robust cutting algorithm is proposed to cut the constructed affinity graphs to accurately obtain the action segments for HAR. The extensive experiments demonstrate the proposed SOAC-RSC scheme achieves the state-of-the-art performance on the datasets of Keck Gesture and Weizmann, and provides competitive performance on the other 6 public datasets such as UCF101 and URADL for HAR task, compared to the recent related approaches. [ABSTRACT FROM AUTHOR]

Anomaly detection refers to the identification of cases that do not conform to the expected pattern, which takes a key role in diverse research areas and application domains. Most of existing methods can be summarized as anomaly object detection-based and reconstruction error-based techniques. However, due to the bottleneck of defining encompasses of real-world high-diversity outliers and inaccessible inference process, individually, most of them have not derived groundbreaking progress. To deal with those imperfectness, and motivated by memory-based decision-making and visual attention mechanism as a filter to select environmental information in human vision perceptual system, in this paper, we propose a Multi-scale Attention Memory with hash addressing Autoencoder network (MAMA Net) for anomaly detection. First, to overcome a battery of problems result from the restricted stationary receptive field of convolution operator, we coin the multi-scale global spatial attention block which can be straightforwardly plugged into any networks as sampling, upsampling and downsampling function. On account of its efficient features representation ability, networks can achieve competitive results with only several level blocks. Second, it’s observed that traditional autoencoder can only learn an ambiguous model that also reconstructs anomalies “well” due to lack of constraints in training and inference process. To mitigate this challenge, we design a hash addressing memory module that proves abnormalities to produce higher reconstruction error for classification. In addition, we couple the mean square error (MSE) with Wasserstein loss to improve the encoding data distribution. Experiments on various datasets, including two different COVID-19 datasets and one brain MRI (RIDER) dataset prove the robustness and excellent generalization of the proposed MAMA Net. [ABSTRACT FROM AUTHOR]

The quality detection of pharmaceutical liquid products is inevitable and crucial in drug manufacture because drugs contaminated with foreign particles are definitely not to be used. However, with the current detection methods, it is still a challenge to detect and identify the small moving particles using an imaging system. In this article, a deep multimodel cascade method combining single-frame image and multiframe images processing method to detect and identify foreign particles is proposed. The proposed method consists of three stages. First, a Faster R-CNN convolutional neural network is adopted to detect and localize the multiple suspected foreign particles of each single-frame image. Then, the k-means clustering algorithm is used to cluster the trail of that detected multiple suspected foreign particles in the eight sequential images to obtain the moving object trajectory. Finally, trajectory features are extracted and the random forest (RF) classifier is used to distinguish noises and foreign particles according to the motion feature of the moving object trajectory. Experimental results demonstrate that the proposed multitask stepwise method improves the accuracy of foreign particles detection and reduces the rate of omission in the case of strong noise, which proves the effectiveness of this method. [ABSTRACT FROM AUTHOR]

Rail inspection system (RIS) remains an emergent instrumentation for railway transportation, with its capacity of measuring surface defect on steel rail. However, detecting technique and interpretation of RIS constitute a challenging problem since traditional technologies are expensive and prone to errors. In this paper, a deep multimodel RIS (DM-RIS) is established for surface defect where fast and robust spatially constrained Gaussian mixture model is presented for segmentation proposal and Faster RCNN is utilized for objective location in a parallel structure. First, we incorporate spatial information between pixels into an improved Gaussian mixture model based on Markov random field (MRF) for accurate and rapid defect edge segmentation. Specifically, a direct parameter-learning in expectation–maximization (EM) algorithm is proposed. Meanwhile, to remove nondefect, numerous labeled samples with weak illumination, inequality reflection, external noise, rust, and greasy dirt are fed into Faster RCNN so that DM-RIS is robust environmentally to various light, angle, background, and acquisition equipment. Finally, the joint hit area refers to a real defect. The experimental results demonstrate that the proposed method performs well with 96.74% precision, 94.13% recall, 95.18% overlap, and 0.485 s/frame speed on average, and is robust compared with the related well-established approaches. [ABSTRACT FROM AUTHOR]

Machine learning algorithms such as those for object classification in images, video content analysis, and human action recognition are used to extract meaningful information from data recorded by image sensors and cameras. Among the existing machine learning algorithms for such purposes, extreme learning machines (ELMs) and online sequential ELMs (OS-ELMs) are well known for their computational efficiency and performance when processing large datasets. The latter approach was derived from the ELM approach and optimized for real-time application. However, OS-ELM classifiers are computationally demanding, and the existing state-of-the-art computing platforms are not efficient enough for embedded systems, especially for applications with strict requirements in terms of low power consumption, high throughput, and low latency. This paper presents the implementation of an ELM/OS-ELM in a customized system-on-a-chip field-programmable gate array-based architecture to ensure efficient hardware acceleration. The acceleration process comprises parallel extraction, deep pipelining, and efficient shared memory communication. [ABSTRACT FROM AUTHOR]

The quality detection of pharmaceutical liquid products is inevitable and crucial in drug manufacture because drugs contaminated with foreign particles are definitely not to be used. However, with the current detection methods, it is still a challenge to detect and identify the small moving particles using an imaging system. In this article, a deep multimodel cascade method combining single-frame image and multiframe images processing method to detect and identify foreign particles is proposed. The proposed method consists of three stages. First, a Faster R-CNN convolutional neural network is adopted to detect and localize the multiple suspected foreign particles of each single-frame image. Then, the $k$ -means clustering algorithm is used to cluster the trail of that detected multiple suspected foreign particles in the eight sequential images to obtain the moving object trajectory. Finally, trajectory features are extracted and the random forest (RF) classifier is used to distinguish noises and foreign particles according to the motion feature of the moving object trajectory. Experimental results demonstrate that the proposed multitask stepwise method improves the accuracy of foreign particles detection and reduces the rate of omission in the case of strong noise, which proves the effectiveness of this method. [ABSTRACT FROM AUTHOR]

Rails are among the most important components of railway transportation, and real-time defects detection of the railway is an important and challenging task because of intensity inhomogeneity, low contrast, and noise. This paper presents an automatic railway visual detection system (RVDS) for surface defects and focuses on several key issues of RVDS. First, in view of challenges such as complex condition and orbital reflectance inequality, we put forward a region-of-interest detection region extraction algorithm by vertical projection and gray contrast algorithm. In addition, a curvature filter equipped with implicit computing and surface preserving power is studied to eliminate noise and keep only the details. Then, an improved fast and robust Gaussian mixture model based on Markov random field is established for accurate and rapid surface defect segmentation. Additionally, an expectation–maximization algorithm is applied to optimize the parameters. The experimental results demonstrate that the proposed method performs well with both noisy and railway images, which enables identification and segmentation of the defects from rail surface, achieving detection performance with 92% precision and 88.8% recall rate on average, and is robust compared with the related well-established approaches. [ABSTRACT FROM AUTHOR]

The particle matter inspection for pharmaceutical injection is inevitable in the field of pharmaceutical manufacturing, as it has the direct impact on the quality of the drugs. It is a challenge to inspect the contaminated injection online using an imaging system. This paper introduces a novel and effective inspection machine consisting of three modules, a mechanical system with 120 carousel grips, an image acquisition system with multihigh resolution cameras and a multilight sources station, and a distributed industrial electrical computer control system. Particle visual inspection machine first acquires image sequence using the high-speed image acquisition system. The image capture process at each camera module is alternately synchronized with different LED illumination techniques (light transmission method and light reflection method), enabling independent capture of particle images from the same container. Then, a set of novel algorithms for image registration and fast segmentation are proposed to minimize false rejections even in sensitive conditions, which enable the identification of all the tiny potential defects. Finally, a particle tracking and classification algorithm based on an adaptive local weighted-collaborative sparse model is also presented. The experiments demonstrate that the proposed inspection system can effectively detect the particles in the pharmaceutical infusion solution online, and achieve a performance rate of above 97% average accuracy. [ABSTRACT FROM PUBLISHER]

Series hybrid electric vehicles improvements in fuel consumption and emissions directly depend on the operating point of the auxiliary power unit (APU). A new APU operating point optimization approach based on dynamic combined cost map (DCM) and particle swarm optimization (PSO) is presented in this paper. The influence of coolant temperature, catalyst temperature, and air/fuel (A/F) ratio on fuel consumption characteristics and HC, CO, NOx emission characteristics are quantitatively analyzed first. Then, the DCM is derived by combining the individual cost maps with predefined weighting factors, so as to balance the potentially conflicting goals of fuel consumption and emissions reduction in the choice of operating point. The PSO is utilized to search the optimum APU operating point in the DCM. Finally, bench experiments under three typical driving cycles show that, compared with the results of the traditional static steady-state fuel consumption map-based APU operating point optimization approach, the proposed DCM and PSO-based approach shows significant improvements in emission performance, at the expense of a slight drop in fuel efficiency. [ABSTRACT FROM AUTHOR]