Chinese Journal of Electronics

Defects Recognition of Train Wheelset Tread Based on Improved Spiking Neural Network

ZHANG Changfan, HUANG Congcong, HE Jing

2023, 32(5): 941-954. doi: 10.23919/cje.2022.00.162

Abstract(697) HTML (341) PDF(102)

Abstract:
Surface defect recognition of train wheelset is crucial for the safe operation of the train wheel system. However, due to the diversity and complexity of such defects, it is difficult for existing algorithms to make rapid and accurate recognitions. To solve this problem, an improved spiking neural network (SNN) based defect recognition method for train wheelset tread is proposed. Specifically, a hybrid convolutional encoding module is first designed to conduct image-to-spike conversion and to create multi-scale sparse representations of the features. Second, a residual spiking convolutional neural network is implemented to extract spiking features optimally, and a multi-scale structure is adopted to enhance the SNN’s ability to handle details. A channel attention module is then incorporated to re-calibrate the weights of four-dimensional spiking feature maps. Finally, effective spiking features are obtained according to the recognition decisions which are made. The experimental results showed that the proposed method improved the accuracy of defect recognition. The recognition time of a single image is only 0.0195 s on average. The overall performance of the proposed method is noticeably superior to current mainstream algorithms.

Research on Health Stage Division of Switch Machine Based on Bray-Curtis Distance and Fisher Optimal Segmentation Method

WU Xiaochun, WEN Xin

2023, 32(5): 955-962. doi: 10.23919/cje.2022.00.250

Abstract(333) HTML (163) PDF(46)

Abstract:
In order to reasonably and accurately evaluate the health status of the switch machine, a health stage division method of switch machine combining Bray-Curtis distance and Fisher optimal segmentation is proposed. First, the power curve of switch machine is divided into five sections, and eight time-domain characteristic parameters of each section are extracted. Second, the characteristic parameters with the largest correlation between fifteen dimensions and state of the switch machine are selected by using the Holder coefficient method as the input of Bray-Curtis distance algorithm, using Bray-Curtis distance to calculate health index (HI), which represents health state of switch machine. Finally, HI curve is divided by Fisher optimal segmentation method, and the optimal number of health stages of switch machine is determined to be three, and HI interval and threshold of each health stage are obtained. The effectiveness of this method is verified by 4382 sets of on-site switch machine data experiments. The experimental results show that the health index curve calculated by Bray-Curtis distance can accurately represent the health status of the switch machine. Compared with Frechet distance and European distance, this method has better performance in tendency, robustness, and runtime. Combining with Fisher optimal segmentation method, it can reasonably and effectively divide the health stage of the switch machine, providing some support for the on-site judgment of the health status of the switch machine.

A Risk Prediction Model Based on Crash History Data for Railway Trams

JI Wenjiang, YANG Jiangcheng, WANG Yichuan, ZHU Lei, QIU Yuan, HEI Xinhong

2023, 32(5): 963-971. doi: 10.23919/cje.2022.00.231

Abstract(436) HTML (222) PDF(43)

Abstract:
Risk prediction is an important task to ensuring the driving safety of railway trams. Although data-driven intelligent methods are proved to be effective for driving risk prediction, accuracy is still a top concern for the challenges of data quality which mainly represent as the unbalanced datasets. This study focuses on applying feature extraction and data augmentation methods to achieve effective risk prediction for railway trams, and proposes an approach based on a self-adaptive K-means clustering algorithm and the least squares deep convolution generative adversarial network (LS-DCGAN). The data preprocessing methods are proposed, which include the K-means algorithm to cluster the locations of trams and the extreme gradient boosting recursive feature elimination based feature selection algorithm to retain the key features. The LS-DCGAN model is designed for sparse sample expansion, aiming to address the sample category distribution imbalance problem. The experiments implemented with the public and real datasets show that the proposed approach can reach a high accuracy of 90.69%, which can greatly enhances the tram driving safety.

Vibration Signal-Based Fault Diagnosis of Railway Point Machines via Double-Scale CNN

CHEN Xiaohan, HU Xiaoxi, WEN Tao, CAO Yuan

2023, 32(5): 972-981. doi: 10.23919/cje.2022.00.229

Abstract(490) HTML (244) PDF(66)

Abstract:
In the railway transportation industry, fault diagnosis of railway point machines (RPMs) is vital. Because operational vibration signals can reflect the condition of various faults in mechanical devices, vibration sensing and monitoring and more importantly, vibration signal-based fault diagnosis for RPMs have attracted the attention of scholars and engineers. Most vibration signal-based fault-diagnosis methods for RPMs rely on data collected using high-sampling-rate sensors and manual feature extraction, hence are costly and insufficiently robust. To overcome these shortcomings, we propose a double-scale wide first-layer kernel convolutional neural network (DS-WCNN) for RPMs fault diagnosis using inexpensive and low-sampling-rate vibration sensors. The proposed wide first-layer kernels, which extract features from vibration observations, are particularly suitable for low-sampling-rate signals. Meanwhile, the proposed double-scale structure improves accuracy and noise suppression by combining two types of timescale features. Sufficient experiments, including noise addition and comparison, were conducted to demonstrate the robustness and accuracy of the proposed algorithm.

Sigma-Mixed Unscented Kalman Filter-Based Fault Detection for Traction Systems in High-Speed Trains

CHENG Chao, WANG Weijun, MENG Xiangxi, SHAO Haidong, CHEN Hongtian

2023, 32(5): 982-991. doi: 10.23919/cje.2022.00.154

Abstract(372) HTML (185) PDF(36)

Abstract:
Fault detection (FD) for traction systems is one of the active topics in the railway and academia because it is the initial step for the running reliability and safety of high-speed trains. Heterogeneity of data and complexity of systems have brought new challenges to the traditional FD methods. For addressing these challenges, this paper designs an FD algorithm based on the improved unscented Kalman filter (UKF) with consideration of performance degradation. It is derived by incorporating a degradation process into the state-space model. The network topology of traction systems is taken into consideration for improving the performance of state estimation. We first obtain the mixture distribution by the mixture of sigma points in UKF. Then, the Lévy process with jump points is introduced to construct the degradation model. Finally, the moving average interstate standard deviation (MAISD) is designed for detecting faults. Verifying the proposed methods via a traction systems in a certain type of trains obtains satisfactory results.

A Security Defense Method Against Eavesdroppers in the Communication-Based Train Control System

YANG Li, WEI Xiukun, WEN Chenglin

2023, 32(5): 992-1001. doi: 10.23919/cje.2022.00.248

Abstract(396) HTML (200) PDF(27)

Abstract:
The communication-based train control system is the safety guarantee for automatic train driving. Wireless communication brings network security risks to the communication-based train control system. The eavesdropping of the transmitted information by unauthorized third-party personnel will lead to the leakage of the estimated value of the system state, which will lead to major accidents. This paper focuses on solving the problem of defense against eavesdropping threats and proposes an eavesdropping defense architecture. This defense architecture includes a coding mechanism based on punishing eavesdroppers, an information upload trigger mechanism based on contribution, and a random information transmission strategy, and provides a guarantee for the privacy protection of information. This research makes three contributions. First, it is the first attempt to construct an information encoding mechanism with punishing eavesdroppers as the objective function; Second, for the first time, an information upload trigger mechanism based on contribution is proposed; Third, the strategy of random transmission of information is proposed. The proposed method in this paper is verified by taking the medium and low-speed maglev train as the object. The experimental results show that, compared with Gaussian noise and non-Gaussian noise mechanisms, the coding mechanism proposed in this paper can not only protect the security of information but also make the estimation error of eavesdroppers tend to be infinite. Using the state estimation error as a metric, the average growth rate of the state estimation error of the system using the trigger mechanism in this paper is less than 2% while improving the security of the system. The transmission strategy in this paper does not increase the system state estimation error while improving the security of the system.

Optimization on the Dynamic Train Coupling Process in High-Speed Railway

CHENG Fanglin, TANG Tao, SU Shuai, MENG Jun

2023, 32(5): 1002-1010. doi: 10.23919/cje.2022.00.189

Abstract(453) HTML (225) PDF(141)

Abstract:
This work focuses on the driving strategy optimization problem of a scenario in which two trains come from two branches under virtual coupling, aiming at going through the junction area efficiently. A distance-discrete optimal control model is constructed. The optimization objective is to maximize the trip time during which the two trains operate in coupled state. The line conditions, dynamic properties of the trains and the safety protection constraints are considered. The nonlinear constraints are converted into linear constraints with piecewise affine function and logical variables, and the proposed problem is converted into mixed integer linear programming (MILP) problem which can be solved by existing solvers such as Cplex. Four simulation experiments are conducted to verify the effectiveness of MILP. The dynamic programming (DP) algorithm is used as the benchmark algorithm in the case study. Compared with DP algorithm in small state space, MILP has better performance since it shortens the coupling time. Moreover, the improvement of line capacity of virtual coupling is 35.42% compared with the fixed blocking system.

Joint Optimization Communication and Computing Resource for LEO Satellites with Edge Computing

JIA Min, WU Jian, ZHANG Liang, GUO Qing

2023, 32(5): 1011-1021. doi: 10.23919/cje.2022.00.314

Abstract(241) HTML (121) PDF(31)

Abstract:
Low earth orbit (LEO) satellites with wide coverage can carry mobile edge computing (MEC) servers with computing power to form the LEO satellite edge computing system, providing computing services for ground users that cannot access the core network. This paper studies the joint optimization problem of communication and computing resource in the LEO satellite edge computing system to minimize the utility function value of the system. Due to the fact that, general optimization tools cannot effectively solve this problem, this paper proposes a deep learning-based bandwidth allocation algorithm. The bandwidth allocation schemes are generated through multiple parallel deep neural networks (DNNs). The utility function values of the system are calculated according to the derived optimal CPU cycle frequency and optimal user transmission power. The bandwidth allocation scheme corresponding to the optimal system utility function value is stored in the memory to further train and improve all DNNs. The simulation results show that the proposed algorithm can achieve good convergence effect and the algorithm proposed in this paper outperforms the other four comparison algorithms with low average time cost.

Scheduling Pattern of Time Triggered Ethernet Based on Reinforcement Learning

HE Feng, XIONG Li, ZHOU Xuan, LI Haoruo, XIONG Huagang

2023, 32(5): 1022-1035. doi: 10.23919/cje.2021.00.419

Abstract(387) HTML (191) PDF(58)

Abstract:
Time-triggered Ethernet (TTEthernet or TTE for short) is a deterministic and congestion-free network based on the Ethernet standard. It supports mix-critical real-time applications by providing different message classes. The time-triggered (TT) messages have strict end-to-end delay and accurate jitter requirement, and the rate-constrained (RC) messages have less determinism than TT messages but with bounded end-to-end delay requirement. Traditionally, the scheduling of TT messages makes it free of conflicts for the transmission on physical links, but ignoring RC messages scheduling, so it cannot guarantee the transmission of RC messages with a bounded delay. Therefore, the design of TT schedule becomes the key to TTE network applications within avionics environment. In this paper, we propose an algorithm called RLTS based on reinforcement learning and tree search, to optimize the end-to-end delays of both TT and RC messages. Besides, its computation speed is dozens of times faster than satisfied modularity theory (SMT) with asynchronous method for the calculation of the optimal scheduling table. In the case of a large network with more than 1000 TT and 1000 RC messages, the RLTS method can find a scheduling timetable in 10 seconds, and reduce the worst-case delay of RC messages averagely by 20% compared to the genetic algorithm. Meanwhile, our algorithm has a good generalization performance, in another word, it can quickly adjust itself to satisfy the scheduling with the similar performance as before. By using our method, the scheduling pattern of TTEthernet is further discussed. According to the experimental results, the uniformly distributed slots scheduling pattern, namely the porosity scheduling model which is usually recommended for TTE application, is not always suitable for general situations.

A Novel Blind Detection Algorithm Based on Spectrum Sharing and Coexistence for Machine-to-Machine Communication

ZHANG Yun, ZHOU Jing, LIU Rong, YU Shujuan, LI Binrui

2023, 32(5): 1036-1049. doi: 10.23919/cje.2021.00.244

Abstract(204) HTML (103) PDF(36)

Abstract:
This paper proposes a new scheme that allows decentralized machine-to-machine (M2M) communication to share spectrum with conventional user communication in an orthogonal frequency division multiplexing system. This scheme can effectively separate and recover mixed signals at the receiving end. It mainly uses the signal space cancellation-complex system Hopfield neural network (SSC-CSHNN) blind detection algorithm to reconstruct the complementary projection operator and the blind detection performance function to restore the M2M communication signals. In order to further improve the anti-interference performance of the system and accelerate the convergence of the algorithm, the double sigmoid idea is introduced, and the signal space cancellation-double sigmoid complex system Hopfield neural network (SSC-DSCSHNN) blind detection algorithm is proposed. The proposed blind detection algorithm improves the anti-interference ability and the convergence speed and prevents the Hopfield neural network from falling into the local optimal solution based on the successful separation and recovery of mixed signals. Compared with existing methods, the blind detection algorithm used in this paper can directly detect the transmitted signal without identifying the channel.

A Layout Method of Space-Based Pseudolite System Based on GDOP Geometry

YANG Xin, WANG Feixue, LIU Wenxiang, XIAO Wei, YE Xiaozhou

2023, 32(5): 1050-1058. doi: 10.23919/cje.2022.00.013

Abstract(277) HTML (139) PDF(31)

Abstract:
The pseudolite system can be used to provide positioning and timing service for users in a specific area. In order to provide better positioning and timing service, a good geometric configuration needs to be formed for the pseudolite system. For the problem of pseudolite system deployment, the average and mean square values of geometric dilution of precision (GDOP) are optimized in this paper for users in a target area. We proposed a space-based pseudolite deployment method based on GDOP geometry, starting from the minimum GDOP value for users in the central area. From the simulation results, it can be seen that, compared with the empirical method and the NSGA-II method, the method in this paper has the smallest average GDOP, a better robustness, and a higher positioning accuracy in the target area, through which a pseudolite deployment proposal can be quickly obtained.

New Construction of Quadriphase Golay Complementary Pairs

LI Guojun, ZENG Fanxin, YE Changrong

2023, 32(5): 1059-1065. doi: 10.23919/cje.2021.00.215

Abstract(442) HTML (220) PDF(32)

Abstract:
Based on an arbitrarily-chosen binary Golay complementary pair (BGCP) ( c , d ) of even length N, first of all, construct quadriphase sequences a and b of length N by weighting addition and difference of the aforementioned pair with different weights, respectively. Secondly, new quadriphase sequence u is given by interleaving three sequences d , a , and − c , and similarly, the sequence v is acquired from three sequences d , b , and c . Thus, the resultant pair ( u , v ) is the quadriphase Golay complementary pair (QGCP) of length 3N. The QGCPs play a fairly important role in communications, radar, and so on.

Towards Order-Preserving and Zero-Copy Communication on Shared Memory for Large Scale Simulation

LI Xiuhe, SHEN Yang, LIN Zhongwei, ZHAO Shunkai, SHI Qianqian, DAI Shaoqi

2023, 32(5): 1066-1076. doi: 10.23919/cje.2021.00.393

Abstract(482) HTML (243) PDF(21)

Abstract:
Parallel simulation generally needs efficient, reliable and order-preserving communication. In this article, a zero-copy, reliable and order-preserving intra-node message passing approach named ZeROshm is proposed. This mechanism partitions shared memory into segments assigned to processes for receiving messages. Each segment consists of two levels of index L1 and L2 that recordes the order of messages in the host segment, and the processes also read from and write to the segments directly according to the indexes, thereby eliminating allocating and copying buffers. As experimental results show, ZeROshm exhibits nearly equivalent performance to message passing interface (MPI) for small message and superior performance for large message. Specifically, ZeROshm costs less time by 43%, 40% and 55% respectively in pure communication, communication with contention and real PHOLD simulation within a single node. Additionally, in hybrid environment, the combination of ZeROshm and MPI also shorten the execution time of PHOLD simulation by about 42% compared to pure MPI.

Vector Memory-Access Shuffle Fused Instructions for FFT-Like Algorithms

LIU Sheng, YUAN Bo, GUO Yang, SUN Haiyan, JIANG Zekun

2023, 32(5): 1077-1088. doi: 10.23919/cje.2021.00.401

Abstract(654) HTML (315) PDF(23)

Abstract:
The shuffle operations are the bottleneck when mapping the FFT-like algorithms to the vector single instruction multiple data (SIMD) architectures. We propose six (three pairs) innovative vector memory-access shuffle fused instructions, which have been proved mathematically. Combined with the proposed modified binary-exchange method, the innovative instructions can efficiently address the bottleneck problem for decimation-in-frequency or decimation-in-time (DIF/DIT) radix-2/4 FFT-like algorithms, reach a performance improvement by 17.9%–111.2% and reduce the code size by 5.4%–39.8%. In addition, the proposed instructions fit some hybrid-radix FFTs and are suitable for the terms of the initial or result data placement for general algorithms. The software and hardware costs of the proposed instructions are moderate.

A Novel Adaptive InSAR Phase Filtering Method Based on Complexity Factors

XU Huaping, WANG Yuan, LI Chunsheng, ZENG Guobing, LI Shuo, LI Shuang, REN Chong

2023, 32(5): 1089-1105. doi: 10.23919/cje.2021.00.280

Abstract(794) HTML (394) PDF(66)

Abstract:
Phase filtering is an essential step in interferometric synthetic aperture radar (InSAR) imaging. For interferograms of complicated and changeable terrain, the increasing resolution of InSAR images makes it even more difficult. In this paper, a novel adaptive InSAR phase filtering method based on complexity factors is proposed. Firstly, three complexity factors based on the noise distribution and terrain slope information of the interferogram are selected. The complexity indicator composed of three complexity factors is used to guide the adaptive selection of the most suitable and effective filtering strategies for different areas. Then, the complexity scalar is calculated, which can guide the adaptive local fringe frequency estimation and adaptive parameters calculation in different filter methods. Finally, validations are performed on the simulated and real data. The performance comparison between the other three representative phase filtering method and the proposed method have validated the effectiveness and superiority of the proposed method.

Track-Oriented Marginal Poisson Multi-Bernoulli Mixture Filter for Extended Target Tracking

DU Haocui, XIE Weixin, LIU Zongxiang, LI Liangqun

2023, 32(5): 1106-1119. doi: 10.23919/cje.2021.00.194

Abstract(685) HTML (347) PDF(73)

Abstract:
In this paper, we derive and propose a track-oriented marginal Poisson multi-Bernoulli mixture (TO-MPMBM) filter to address the problem that the standard random finite set filters cannot build continuous trajectories for multiple extended targets. First, the Poisson point process model and the multi-Bernoulli mixture (MBM) model are used to establish the set of birth trajectories and the set of existing trajectories, respectively. Second, the proposed filter recursively propagates the marginal association distributions and the Poisson multi-Bernoulli mixture (PMBM) density over the set of alive trajectories. Finally, after pruning and merging process, the trajectories with existence probability greater than the given threshold are extracted as the estimated target trajectories. A comparison of the proposed filter with the existing trajectory filters in two classical scenarios confirms the validity and reliability of the TO-MPMBM filter.

An Adaptive Interactive Multiple-Model Algorithm Based on End-to-End Learning

ZHU Hongfeng, XIONG Wei, CUI Yaqi

2023, 32(5): 1120-1132. doi: 10.23919/cje.2021.00.442

Abstract(685) HTML (335) PDF(43)

Abstract:
The interactive multiple-model (IMM) is a popular choice for target tracking. However, to design transition probability matrices (TPMs) for IMMs is a considerable challenge with less prior knowledge, and the TPM is one of the fundamental factors influencing IMM performance. IMMs with inaccurate TPMs can make it difficult to monitor target maneuvers and bring poor tracking results. To address this challenge, we propose an adaptive IMM algorithm based on end-to-end learning. In our method, the neural network is utilized to estimate TPMs in real-time based on partial parameters of IMM in each time step, resulting in a generalized recurrent neural network. Through end-to-end learning in the tracking task, the dataset cost of the proposed algorithm is smaller and the generalizability is stronger. Simulation and automatic dependent surveillance-broadcast tracking experiment results show that the proposed algorithm has better tracking accuracy and robustness with less prior knowledge.

Microwave Tomographic Imaging of Anatomically Realistic Numerical Phantoms with Debye Dispersion for Breast Cancer Detection Using a Regularized Inverse Scattering Technique in the Time Domain

LIU Guangdong

2023, 32(5): 1133-1150. doi: 10.23919/cje.2021.00.343

Abstract(367) HTML (180) PDF(40)

Abstract:
In recent years, three similar versions of time-domain inverse scattering (TDIS) algorithms have been proposed for the successful estimation of the dispersive dielectric properties of several single-pole Debye media. For practical applications in common biomedical engineering, an improved TDIS algorithm is explicitly derived to provide a more versatile algorithm for the microwave tomographic imaging of biological tissues. Its three improvements are as follows. The number of poles for Debye models is extended from one to a positive integer W. The second improvement is the extension of unknowns from three to 2W+2 for each discretized cell. The third improvement is the adoption of the first-order Tikhonov regularization scheme. Based on the four classes of 2-dimensional anatomically realistic numerical phantoms with two-pole Debye dispersion from the University of Wisconsin Computational Electromagnetics Laboratory (UWCEM) database, the performance of the developed algorithm for the detection of a 3-mm-diameter tumor implanted in the four types of breast models was investigated for three scenarios. The obtained results preliminarily indicate that the modified technique is feasible and promising for the quantitative reconstruction of sparse breast tissues.

Transformer-Based Under-sampled Single-Pixel Imaging

TIAN Ye, FU Ying, ZHANG Jun

2023, 32(5): 1151-1159. doi: 10.23919/cje.2022.00.284

Abstract(794) HTML (394) PDF(118)

Abstract:
Single-pixel imaging, as an innovative imaging technique, has attracted much attention during the last decades. However, it is still a challenging task for single-pixel imaging to reconstruct high-quality images with fewer measurements. Recently, deep learning techniques have shown great potential in single-pixel imaging especially for under-sampling cases. Despite outperforming traditional model-based methods, the existing deep learning-based methods usually utilize fully convolutional networks to model the imaging process which have limitations in long-range dependencies capturing, leading to limited reconstruction performance. In this paper, we present a transformer-based single-pixel imaging method to realize high-quality image reconstruction in under-sampled situation. By taking advantage of self-attention mechanism, the proposed method is good at modeling the imaging process and directly reconstructs high-quality images from the measured one-dimensional light intensity sequence. Numerical simulations and real optical experiments demonstrate that the proposed method outperforms the state-of-the-art single-pixel imaging methods in terms of reconstruction performance and noise robustness.

The Novel Instance Segmentation Method Based on Multi-Level Features and Joint Attention

XU Bowen, LU Yinan, WU Tieru, GUO Xiaoxin

2023, 32(5): 1160-1168. doi: 10.23919/cje.2021.00.226

Abstract(265) HTML (134) PDF(25)

Abstract:
Instance segmentation is an important task in computer vision. In order to enhance the multi-level features expression ability of the segmentation networks, a novel module is proposed in this paper. Firstly, we design a weighted bi-directional feature fusion way by computing the weight distribution function of bi-directional feature pyramid network. Secondly, we propose a joint attention mechanism to effectively filter different levels of feature information by adopting serial and parallel ways to combine the channel attention and spatial attention modules. At the same time, the module uses dynamic convolution to stabilize the calculation speed while improve the 6.7% mean average precision of segmentation. The experiments on the COCO dataset demonstrate that the module can effectively improve the performance of the existing instance segmentation networks.

EAODroid: Android Malware Detection Based on Enhanced API Order

HUANG Lu, XUE Jingfeng, WANG Yong, QU Dacheng, CHEN Junbao, ZHANG Nan, ZHANG Li

2023, 32(5): 1169-1178. doi: 10.23919/cje.2021.00.451

Abstract(653) HTML (327) PDF(48)

Abstract:
The development of smart mobile devices brings convenience to people’s lives, but also provides a breeding ground for Android malware. The sharp increasing malware poses a disastrous threat to personal privacy in the information age. Based on the fact that malware heavily resorts to system application programming interfaces (APIs) to perform its malicious actions, there has been a variety of API-based detection methods. Most of them do not consider the relationship between APIs. We contribute a new approach based on the enhanced API order for Android malware detection, named EAODroid, which learns the similarity of system APIs from a large number of API sequences and groups similar APIs into clusters. The extracted API clusters are further used to enhance the original API calls executed by an app to characterize behaviors and perform classification. We perform multi-dimensional experiments to evaluate EAODroid on three datasets with ground truth. We compare with many state-of-the-art works, showing that EAODroid achieves effective performance in Android malware detection.

2023 Vol. 32, No. 5

Links

Chinese Journal of Electronics