Search

The extensive deployment of camera-based IoT devices in our society is heightening the vulnerability of citizens’ sensitive information and individual data privacy. In this context, thermal imaging techniques become essential for data desensitization, entailing the elimination of sensitive data to safeguard individual privacy. Meanwhile, thermal imaging techniques can also play a important role in industry by considering the industrial environment with low resolution, high noise and unclear objects’ features. Moreover, existing works often process the entire video as a single entity, which results in suboptimal robustness by overlooking individual actions occurring at different times. In this paper, we propose a lightweight algorithm for action recognition in thermal infrared videos using human skeletons to address this. Our approach includes YOLOv7-tiny for target detection, Alphapose for pose estimation, dynamic skeleton modeling, and Graph Convolutional Networks (GCN) for spatial-temporal feature extraction in action prediction. To overcome detection and pose challenges, we created OQ35-human and OQ35-keypoint datasets for training. Besides, the proposed model enhances robustness by using visible spectrum data for GCN training. Furthermore, we introduce the two-stream shift Graph Convolutional Network to improve the action recognition accuracy. Our experimental results on the custom thermal infrared action dataset (InfAR-skeleton) demonstrate Top-1 accuracy of 88.06% and Top-5 accuracy of 98.28%. On the filtered kinetics-skeleton dataset, the algorithm achieves Top-1 accuracy of 55.26% and Top-5 accuracy of 83.98%. Thermal Infrared Action Recognition ensures the protection of individual privacy while meeting the requirements of action recognition. [ABSTRACT FROM AUTHOR]

In order to improve the performance of sliding mode observer in detecting the state of induction motor, a state detection method based on particle swarm optimization (PSO)-backstepping (BS)-sliding mode observer (SMO) is proposed in this paper. In this method, the controller is constructed and the parameters of the control rate are optimized, so that the tracking accuracy and robustness of the new observer are improved relative to conventional observer, exponential observer, PI and PID. Firstly, the state equation of the induction motor under stator and rotor winding fault and stator current sensor fault is established. Secondly, the new sliding mode observer is designed using the backstepping method based on the new reaching law. Then, the new fitness function and PSO is used to optimize the parameters of the new sliding mode observer. Finally, the simulation comparison experiment of stator current state detection is carried out under the simulated fault condition of induction motor. The feasibility of the method is verified by comparing the state tracking situation and the state detection error. The comparative experimental results show that the method has less jitter, stronger robustness, and higher state tracking accuracy when detecting stator current states under different faults. [ABSTRACT FROM AUTHOR]

In order to accurately track the state of the doubly fed induction generator (DFIG) for wind power systems and implement its fault detection, an event-triggered sliding mode observer based on particle swarm optimization (PSO-ET-SMO) is proposed in this paper. First, the sliding mode observer (SMO) is designed according to the rotor current state space equation of the DFIG. Then, a new event-triggered mechanism (NETM) is proposed for the designed SMO, and the constructed fitness function is optimized by using Particle Swarm Optimization (PSO) algorithm, so as to obtain the optimal parameters of the SMO. Finally, to verify the effectiveness of the proposed method in this paper, the fault detection of the DFIG for wind power systems is performed by using the sequence of residuals between the rotor current output values and the sliding mode observations. The innovation of this work is that a NETM is designed by combining sliding mode reaching law, and this mechanism is introduced into the design of SMO. The simulation results show that the proposed method not only reduces sliding mode chattering, but also improves the tracking effect of SMO and detects faults accurately. At the same time, due to the action of the NETM, the transmission burden between the SMO and the DFIG is also reduced. [ABSTRACT FROM AUTHOR]

Traditional fault recognition algorithms can only identify the known classes of fault in wind turbine systems(WTs). These faults have already appearedin the WTs, thus, the fault recognition algorithm can identify the classesof them. However, if a new classfault didnot happen in the WTs before, the traditional fault recognition algorithm can only identify it as known class of fault, which results in an incorrect identification. Toaddress this problem, a new class fault recognition method based on semi-supervised deep learning(SDL-NCFR)is proposed. Firstly, multiple WTs signals are used as input and features are extracted by convolutional autoencoder network; secondly, the initialization model is built with compressed features as input to the classifier and error feature map as input to the detector; finally, the detector will put new class fault instances into the buffer. When the buffer overflows, the algorithm starts to update, thus achieving the purpose of identifying new class faults. The experimental results show that the average accuracy of the initialized model could reach more than 98%. The accuracy of updated model could still reach 89.39%, and the detection rate could reach 99.50%, the recall reached 88.76%, the precision reached 92.03%, and the F1 score reached 90.36% respectively. The experimental results show that the proposed algorithm can effectively solve the problem of identifying new class faults in WTs, and the accuracy is much higher than that of traditional detection methods. [ABSTRACT FROM AUTHOR]

The majority of recently demonstrated Deep-Learning Side-Channel Analysis (DLSCA) use neural networks trained on a segment of traces containing operations only related to the target subkey. However, when the size of the training set is limited, as in this paper with only 5K power traces, the deep learning (DL) model cannot effectively learn the internal features of the data due to insufficient training data. In this paper, we propose a cross-subkey training approach that acts as a trace augmentation. We train deep-learning models not only on a segment of traces containing the SBox operation of the target subkey of AES-128 but also on segments for other 15 subkeys. Experimental results show that the accuracy of the subkey combination training model is 28.20 % higher than that of the individual subkey training model on traces captured in the microcontroller implementation of the STM32F3 with AES-128. And validation is performed on two additional publicly available datasets. At the same time, the number of traces that need to be captured when the model is trained is greatly reduced, demonstrating the effectiveness and practicality of the method. [ABSTRACT FROM AUTHOR]

In this paper, a five-dimensional hyperchaotic system for secure communication based on MCU is proposed. First, a five-dimensional hyperchaotic system is constructed and the features such as equilibrium stability and Lyapunov exponents are analyzed. The states of this system with different parameters are studied, including periodic, hyperchaotic and chaotic state. Next, the complexity of system is analyzed. By analyzing the complexity of each dimension, the correctness of system selection is verified and the dimension with the highest complexity is selected for communication encryption. Then, the hardware circuit is designed. The encryption and decryption of square wave signal are completed through simulation which realizing the secure communication. Finally, the accuracy of system and the algorithm of realizing chaos of MCU are adjusted, and the system after adjustment is in a chaotic state verified by 0–1 test, the chaos is successfully generated by MCU. Using MCU can reduce the hardware cost of signal encryption greatly, and apply five-dimensional hyperchaotic system to secure communication can not only improve the security but also enhance the anti-interference ability in the process of signal transmission. [ABSTRACT FROM AUTHOR]

Buck converter is a kind of converter device with high efficiency, wide adjustable output voltage, small loss, small size and light weight. Its circuit has nonlinearity and will exhibit abundant nonlinear phenomena with the change of circuit parameters, resulting in unstable output voltage and being susceptible to load and external disturbances. Therefore, a control method based on Extreme Learning Machine (ELM) combined with a proportionalintegral-derivative (PID) controller is proposed in this paper and used for output voltage control of buck converter. Firstly, the mathematical model of buck converter in Continuous Conduction Mode (CCM) is established by the state space averaging method. Then the PID tuning algorithm is designed in combination with ELM, and the stability analysis of the model is carried out. Finally, the simulation experiment is carried out under different disturbances. By comparing with the open-loop control strategy, the effectiveness of the proposed ELM-PID control strategy is verified, indicating that the proposed method can achieve the stability of output voltage and good dynamic response. [ABSTRACT FROM AUTHOR]

This paper proposes a novel seven-dimensional fractional-order chaotic system applied to secure communication. First, a seven-dimensional fractional order system is constructed by selecting an arbitrary fractional order q. Then, the chaotic and non-chaotic dynamic behaviors of the system are analyzed by 0–1 Test, and the correctness is verified by phase diagram comparison and complexity analysis. Then, the corresponding fractional-order chaotic circuit is designed, and the phase image which is basically consistent with the experimental simulation results is obtained. Finally, the seven-dimensional fractional-order chaotic system is used to complete the signal encryption and decryption test to achieve secure communication. This indicates that the seven-dimensional fractional-order chaotic system proposed to this paper has more complex dynamic characteristics and can be used for secure communication. [ABSTRACT FROM AUTHOR]

Electrical Parking Brake (EPB) has been popularly used in passenger cars over the past ten years. With the help of the several kinds of sensors mounted in driveline for the total traction force estimation, EPB can obtain well performance on drive-off assistance and automatic parking brake. Furthermore, its AUTOHOLD function can realize automatic parking brake and ease the driver. However, given that the higher cost and complexity of this traction force estimation method based on the driveline sensors and its slower response resulted by applying maximum parking force for safety parking while driving off, a novel automatic parking brake system without these transmission system sensors, such as clutch position sensor for the cars with manual transmission, is proposed in this paper, including its control scheme and application test. Firstly, the indirect judgement method of the appropriate moment to release the parking brake, which is based on the car pitch moment when it drives off, is introduced according to the force analysis when the car is ready to go. Then a pragmatic mass estimation method for proper brake force calculation is proposed for improving the drive-off performance. In addition, for the convenience and drivability of skillful driver, as well as the system reliability, a mechanical redundant design to reserve the conventional handbrake lever is also described. Finally, various simulations based on CarSim software and road tests are performed to validate its effectiveness. [ABSTRACT FROM AUTHOR]

Although electric vehicle fully exhibits its comparative merits of energy conservation and environmental friendliness, further improvement of its traction energy efficiency lacks comprehensive investigations in the past. In this paper, the effect of the torque vectoring on traction energy conservation during cornering for a rear-wheel-independent-drive electric vehicle is investigated. Firstly, turning resistance coefficient and energy conservation mechanism of torque vectoring are derived from the single track dynamic model. Next, an optimal torque vectoring control strategy based on genetic algorithm is proposed, with the consideration of the influence of the operation-point change of the in-wheel motors, to find out the best torque vectoring ratio offline. Finally, various simulation tests are conducted to validate the energy conservation effect after Simulink modelling. The results verify that though the optimization of the operating region of the motors is the main part for tractive energy conservation, the contribution of torque vectoring itself can reach up to 1.7% in some typical cases. [ABSTRACT FROM AUTHOR]

Pursuing more efficient running and operation has been the orientation of today's automotive powertrain and chassis technologies development. As for the brake system, increasing the braking efficiency or reducing the power consumption of actuation is an effective method to improve the energy-saving performance of the car. In this paper, a self-energizing wedge disc brake actuated by hydraulic power, that could be used to reduce both the actuation force and the energy required for braking is presented and compared in terms of the practical issues on braking efficiency, heat recession and self-locking. The selected critical parameters and their optimum value range are determined based on the static force analysis. Then the simulations based on AMESim/Simulink are implemented to compare the proposed wedge brake with traditional disc brake, and to investigate the influence of the wedge angle and the actuating angle on the braking efficiency performance. The simulation results validate the self-reinforcement and energy saving performance of this hydraulic wedge disc brake well. [ABSTRACT FROM AUTHOR]

Developing fault detection is very important for improving the equipment reliability and saving energy consumption. As compared with neural networks, extreme learning machine (ELM) is based on statistical learning theory, which has advantages of better classification ability and generalization performance. This paper presents a novel approach for fault detection and diagnosis based on Firefly-Chaos Algorithm and Extreme Learning Machine. The experiment result indicates this proposed method is effective for Analog Circuit fault detection and diagnosis and the model generalization ability is favorable. [ABSTRACT FROM AUTHOR]