20 June 2025, Volume 32 Issue 6
    

  • Select all
    |
  • WANG Yinsong, YAN Xin
    Control Engineering of China. 2025, 32(6): 961-968.
    Abstract ( )   Knowledge map   Save
    Because the components in the control system will gradually deteriorate in the process of use, the performance of the whole system will continue to decline in the long-term operation, and finally its life will be terminated because it cannot complete the expected control objectives. In order to accurately evaluate the random performance of the control system and manage its health, a performance evaluation method of the control system based on the improved Babbitt distance index is proposed, and the weight coefficient is introduced to reduce the interference of various noises in the data to the evaluation value. The kernel density estimation method is used to obtain the system performance evaluation value. Considering only the sensor deterioration, the new index is used as the performance evaluation criterion to determine the performance failure threshold and life threshold of the control system. The nonlinear Wiener process is used to model and predict its residual life distribution. The effectiveness of the proposed method is verified by the simulation case of three tank liquid level control system.
  • LI Fudong, JIANG Bin, YANG Yuequan, CHEN Xinyu, CAO Zhiqiang, JIANG Yuanlei
    Control Engineering of China. 2025, 32(6): 969-976.
    Abstract ( )   Knowledge map   Save
    For the problems of poor pin consistency and low assembly success rate during parallel assembly of in-line components with multiple peg-in-hole, a precision assembly system of in-line component multiple peg-in-hole is developed based on stable and accurate peg-in-hole visual positioning algorithm and high-efficiency robot vision-guided insertion technology. Firstly, offline calibration is to locate the axis and hole centre to determine the robot pose; secondly, combined with the computer aided design(CAD) model, a constrained random sample consensus (RANSAC) ellipse fitting algorithm is proposed to accurately locate the centres of shafts and holes. Then, a novel insertion guide algorithm is developed to guides the deployment by predicting the assembly clearance of multiple peg-in-hole through the anchor points and the rotation basis of the average angle deviation of the multiple peg-in-hole. The experimental result showed that the designed system had the advantages of high positioning and assembly accuracy, fast speed and high assembly success rate. The overall detection time of a single component is less than 100 ms, the final assembly error of a single axis is less than ±0.3 mm, and the assembly success rate is greater than 98%.
  • ZUO Mingxin, WEI Dong, XIONG Yaxuan, ZHAO Ruochen
    Control Engineering of China. 2025, 32(6): 977-986.
    Abstract ( )   Knowledge map   Save
    During the transmission of waste gas and other gases produced by natural gas, the pressure regulating system needs to regulate the gas from high pressure to low pressure. During this process, the pressure energy of the gas can be recovered by driving the generator through an expansion machine. A gas pressure energy recovery power generation control system is designed and developed, and a fuzzy PI controller is designed in the generator speed control loop to achieve adaptive adjustment of PI parameters; a model predictive controller is also designed for the generator torque control loop, and voltage vector calculation is performed based on the optimization objective of making the generator torque and magnetic chain stable to improve the efficiency of the control strategy and to improve the control performance and robustness through delay compensation. A gas pressure energy recovery experimental platform was built to simulate gas pressure changes by using a compressor to regulate air pressure, and based on the above strategy, a gas pressure energy recovery power generation controller was designed and developed using DSP28335. Experimental results show that the proposed control strategy improves the quality of power generation by significantly reducing the range of generator torque and speed fluctuations in the presence of user-side pressure fluctuations compared to the conventional direct torque control strategy.
  • QIAN Ziyuan, WANG Zuo
    Control Engineering of China. 2025, 32(6): 987-994.
    Abstract ( )   Knowledge map   Save
    Focused on the DC-DC buck converter system with mismatched disturbances, an improved adaptive reaching-law based discrete-time sliding mode controller (ADSMC) is proposed. To eliminate the effects of mismatched disturbances, a time-delay based observer is firstly applied to have accurate estimations of the mismatched disturbances. Then, the disturbance estimations are introduced into the design of a modified sliding surface, which realize the compensation of mismatched disturbances. Moreover, to alleviate the chattering problem and ensure anti-disturbance ability, a new adaptive discrete-time sliding mode reaching-law is designed and applied to the control of buck converters. A rigorous proof of the stability analysis of the closed-loop system is presented. Finally, the effectiveness of ADSMC is verified by both simulation and experimental results. Compared with existing discrete-time sliding mode controllers, the proposed ADSMC has achieved higher voltage tracking accuracy and stronger anti-disturbance ability.
  • TANG Hao, YANG Chenfang, CHENG Wenjuan, WANG Zhengfeng, SHI Mingguang
    Control Engineering of China. 2025, 32(6): 995-1007.
    Abstract ( )   Knowledge map   Save
    Various flexible resources of both source and load sides are gradually involved in power grid, which complicates the dynamic characteristics of power system. In order to improve the learning efficiency of dispatching optimization for power system, a learning optimization method is proposed by cross-dimensional transfer of dispatching knowledge matrix of source power system without flexible resources. Firstly, the similarity determination method of correlation features between source task and target task is given by using Euclidean-dynamic time warping distance. Then, the principal component analysis technique is introduced to establish the mapping relationship between similar states or similar actions of source task and target task. Thus, a reinforcement learning method based on the cross-dimensional transfer of knowledge matrix is proposed, which is used to solve the problem that the previous dispatching knowledge cannot be used directly due to the different state or action dimensions of source task and target task. Finally, the IEEE-300 bus system is taken as an example for simulation analysis. The results show that the proposed method can effectively make use of the previous dispatching knowledge of source tasks, and realize the rapid dispatching optimization of complex power system with flexible resources.
  • LIU Yongmin, XIAO Fengjiao, QIAO mengyuan, DENG Weihao, MA Haizhi
    Control Engineering of China. 2025, 32(6): 1008-1015.
    Abstract ( )   Knowledge map   Save
    Because the general graph neural network feature extraction module is designed as a fixed convolutional neural network (CNN), which leads to the existence of limited acceptance domain and easy to ignore the key feature information of the image when capturing the global feature information. In order to extract comprehensive and critical feature information, a novel CA-MFE algorithm is proposed: firstly, different convolutional kernels in the CNN are utilized to capture multi-scale local feature information, and then the channel and spatial attention mechanisms are processed in parallel according to the global feature extraction capability of the attention mechanism, in order to extract multi-dimensional global feature information. The performance of the new model is evaluated on mini-ImageNet and tiered-ImageNet datasets for a comprehensive and comprehensive performance evaluation. The classification accuracies are improved by 1.07% and 1.33% compared to the baseline model, respectively. Using the mini-ImageNet dataset in the 5-way 5-shot task, the classification accuracies are improved by 11.41%, 7.42%, and 5.38% compared to the GNN, TPN, and Dynamic models, respectively. The experimental results show that the CA-MFE model is significantly superior to the baseline model and several representative small-sample classification algorithms when dealing with small-sample classification data.
  • LIU Huanxiao, SHI Yilun, YANG Xiaofei, XIANG Zhengrong, WANG Ronghao
    Control Engineering of China. 2025, 32(6): 1016-1021.
    Abstract ( )   Knowledge map   Save
    Unmanned surface vehicles (USVs) often encounter dynamic obstacles. It is frequently used to solve dynamic obstacle collision avoidance by the dynamic windows approach (DWA). However, the velocity information of dynamic obstacles is not always considered. Due to insufficient time to turn, the USVs will collide with the dynamic obstacle. Therefore, the collision risk index and the velocity obstacle method are introduced to improve the DWA and realize local path planning. Simulation scenes of head-on and crossing situations are constructed, and the feasibility and advantages of the dynamic path planning method proposed are proved.
  • QI Chenman, MAO Zehui, ZHANG Gengwei
    Control Engineering of China. 2025, 32(6): 1022-1027.
    Abstract ( )   Knowledge map   Save
    Rolling bearings often need to work under harsh conditions. The normal operation of the bearing is directly related to the safety of production, so predicting the life of the rolling bearing to ensure that the rolling bearing can work safety is of great significance to avoid casualties and property losses. For the problem that the degradation features of rolling bearings with different failure signals are coupled with each other and the features with weak energy are not easy to extract, resulting in inaccurate prediction of the life of rolling bearings, BTCN with less time and memory consumption is proposed. By combining BLS and TCN, the feature extraction capability of TCN is used to enhance the identification and extraction of multi-failure location features, and the flat structure of width learning is used to reduce the complexity of the network model, thereby reducing calculation time. Through the experimental on the XJTU-SY rolling bearing data set, the results show that the algorithm can effectively solve the problem of rolling bearing life prediction under multiple failure modes, which provides a new idea for rolling bearing life prediction in the industry.
  • WANG Huimeng, GE Quanbo, WU Qingtao, ZHENG Ruijuan, ZHU Junlong
    Control Engineering of China. 2025, 32(6): 1030-1038.
    Abstract ( )   Knowledge map   Save
    An edge detection method applicable to images acquired by a visible camera mounted on an unmanned ship is proposed for a coordinated landing system consisting of an unmanned ship and an unmanned aircraft. Firstly, an optimization defogging algorithm based on sky region segmentation is designed to cope with the challenge of sky region color distortion in the dark-channel a priori defogging algorithm. Then, focusing on the problem of the small target of the UAV in the image, a strategy for selecting the rectangular region of the UAV part as the ROI region is introduced; on this basis, an optimized depth-guided filter is devised to address the difficulty of balancing noise removal and edge preservation with the Canny operator. Finally, a fast and accurate iterative method is used to select the optimal threshold. Through simulation experiments, the results confirm that the improved algorithm maintains the color realism of the sky region in the defogged image and effectively extracts the more complete and real edges of the UAV.
  • ZHOU Jianmin, XIA Xiaofeng, LI Jiahui
    Control Engineering of China. 2025, 32(6): 1039-1048.
    Abstract ( )   Knowledge map   Save
    In order to solve the problem that the fault data in the actual operation of rolling bearing is much less than the normal data, which affects the diagnosis rate of fault diagnosis model, a rolling bearing fault diagnosis method based on improved GAN (WGAN-GP) is proposed in the case of data imbalance. Firstly, CWT is used to transform the vibration signal set into two-dimensional image data set. Then the Wasserstein distance is used to replace the JS divergence of GAN, and the gradient penalty strategy is used to optimize the model in the WGAN weight clipping, so that the weight of the loss function of the generator is balanced in the interval, the fault data is generated automatically and the fault data set is expanded. Finally, unbalanced data sets and data enhancement contrast experiments are set up, the results show that the model diagnosis rate of WGAN-GP is improved by 2.29%, 1% and 2.85% respectively under the different unbalanced proportion experiments, and the diagnosis rate in the data enhancement contrast test is also higher than that of geometric transformation data enhancement and original data.
  • TANG Jiaxiang, HUANG Congzhi
    Control Engineering of China. 2025, 32(6): 1049-1057.
    Abstract ( )   Knowledge map   Save
    For the problem of early warning of thermal power unit equipment status detection, a long short-term memory (LSTM) multiple output regression model based on inovative optimizer named chaos weighted mean of vectors (CINFO) is proposed. Firstly, the Spearman correlation coefficient analysis method is used to screen out auxiliary variables with high correlation coefficients with the secondary fan bearing temperature and secondary fan bearing vibration, which reduces the dimension of the input data. Secondly, the optimal hyperparameters of the multi-output LSTM network are determined by CINFO, which improves the prediction accuracy of the neural network. Subsequently, the equipment failure threshold is determined according to the sequential probability ratio test (SPRT). Finally, the selected feature parameters are used as the input of the CINFO-LSTM network, and the sequential probability ratio test method is used to realize the fault early warning of the secondary fan. The feasiblity and effectiveness of the proposed approach is validated by the given extensive experimental results.
  • GE Zhenpeng, WANG Hongfeng
    Control Engineering of China. 2025, 32(6): 1058-1065.
    Abstract ( )   Knowledge map   Save
    As one of important factors in keeping high-quality production, preventive maintenance that refers to conduct an equipment maintenance in advance before failure is helpful to improve the working condition of equipment. In order to solve the coupling problem between flow shop production and preventive maintenance, an integrated scheduling model is developed with consideration of the step deterioration effect of equipment and the minimal reliability limit based incomplete preventive maintenance. A Q-learning based aquila optimizer (QL-AO) is proposed for solving the developed model, where Q-learning is used to guild AO to adjust the selection probabilities of four update ways, and local search methods as well as a diversity strategy are introduced into the proposed QL-AO algorithm. The validity of the proposed model and algorithm is verified by the comparative experiments in different scale examples.
  • SANG Hongqiang, LU Wei, LIU Fen, HUANG Fang
    Control Engineering of China. 2025, 32(6): 1065-1073.
    Abstract ( )   Knowledge map   Save
    In view of the difficulty and safety of existing minimally invasive surgical robot's preoperative adjustment from the internal resistance of the slave manipulator joint and its own gravity, a force-free control method based on flexible-joint dynamic model and Luenberger state observer is proposed. The dynamic model of the slave manipulator including friction and joint flexibility is established. The dynamic parameters can be identified by using least square method. To quickly and accurately obtain joint velocity and acceleration, a joint equivalent state space model of the flexible joint is established and a Luenberger observer is designed. The corresponding simulation and force-free adjustment experiment are carried out. The results show that the proposed force-free control method is effective and can compensate most of the resistance in the adjustment process, improve the efficiency of preoperative adjustment and ensure the safety of preoperative operation.
  • WANG Yufang, HUA Xiaolin, ZHANG Dianqing, YAO Binbin, CHEN Fan
    Control Engineering of China. 2025, 32(6): 1074-1085.
    Abstract ( )   Knowledge map   Save
    Consider the production of different workpieces using multiple molds and the operational requirements of workpieces between machines, the transportation time of the workpiece and the setup time of the mold are incorporated into the flexible job-shop scheduling model, and a multi-time constrained energy-efficient scheduling of flexible job-shop model with the goal of minimizing the maximum completion time and energy consumption is established, and a hybrid Jaya algorithm is proposed to solve the problem. Firstly, a hybrid initialization strategy is designed to improve the quality of the initial population and accelerate the convergence speed of the algorithm. Secondly, the Jaya optimization strategy is used to traverse all non-optimal individuals to improve the global search ability of the algorithm. Thirdly, in order to mine better solutions in the population, a variety of local search strategies are designed, and targeted searches are carried out based on individual characteristics, so as to expand the search ability of the algorithm. Finally, Ablation experiments were carried out on the improvement strategy through standard examples to verify the performance of the improvement strategy. Through test cases and production examples, the effectiveness of the hybrid Jaya algorithm is verified by comparing with other literature algorithms.
  • LV Yao, ZHAO Yuan, LIU Yefeng, SUN Weitang, ZHAO Kexue
    Control Engineering of China. 2025, 32(6): 1086-1091.
    Abstract ( )   Knowledge map   Save
    An optimization method for complex surface machining parameters based on Response Surface Methodology (RSM) is proposed. Firstly, a second-order model is established to characterize the influence of cutting speed, feed rate per tooth, and cutting depth on tool wear. Related three-dimensional response surfaces are generated. The effectiveness of the proposed model is verified through analysis of the normal probability plot of residuals, the plot of residuals versus predicted values, and the plot of predicted values versus actual values. Subsequently, to address the issues of tool wear and compensation in automated impeller machining, an experimental plan was designed based on the principles of Response Surface Methodology (RSM) to investigate the relationship between cutting parameters and tool wear. Experimental results demonstrate that the second-order model for tool length wear, developed using RSM, is accurate and effective. The order of influence of the cutting parameters on tool length wear is determined as: cutting speed > feed rate per tooth > cutting depth. The optimal cutting parameter scheme for the automated machining of a specific type of impeller is obtained through model analysis and calculation. Experimental verification is conducted.
  • LU Xun, CAO Yuqing, XIE Li
    Control Engineering of China. 2025, 32(6): 1092-1100.
    Abstract ( )   Knowledge map   Save
    A bias compensation based recursive least squares (BCRLS) identification algorithm for dual-rate Hammerstein output error systems with interferences of non-zero mean Gaussian noises is proposed. Firstly, by using the polynomial transformation technique, the target system is converted into a model that can be directly identified based on the dual-rate sampled data, and the recursive least squares (RLS) algorithm is utilized for its identification. Secondly, in order to effectively compensate the biased parameter estimates given by the RLS algorithm, based on the principle of bias compensation, parameters in the bias term are solved by introducing a non-singular matrix and an extended information vector, thus the BCRLS algorithm is derived. Finally, the numerical simulation experiment illustrates that the BCRLS algorithm can obtain unbiased parameter estimates for dual-rate Hammerstein output error systems, which has strong robustness and is less susceptible to the mean and variance changes of the noise.
  • CHENG Haokuan, XIAO Min
    Control Engineering of China. 2025, 32(6): 1101-1110.
    Abstract ( )   Knowledge map   Save
    In order to study the propagation mechanism of malicious viruses in cyber physical systems (CPS), a class of SIR (Susceptible Infected Recovered) model with bilinear incidence is improved. A more general SIR malicious virus model with time delays is proposed. The complex dynamical behavior is analyzed based on stability theory and Hopf bifurcation theorem by selecting the time delay as the bifurcation parameter, which leads to the condition of Hopf bifurcation. It is shown that the dynamical behavior of the system depends on the critical value of the bifurcation. In addition, the canonical type and central flow shape theorems are further applied to derive the formulae for the direction of Hopf bifurcation. Finally, the correctness of the theoretical analysis is verified by numerical simulation.
  • LIU Haodong, YU Jinfei, LIU Jiapeng, YU Huihui, YU Jinpeng
    Control Engineering of China. 2025, 32(6): 1111-1121.
    Abstract ( )   Knowledge map   Save
    For the control problem of induction motor system considering time-varying full state constraint, a finite-time position tracking control scheme based on command filter approximation technology is proposed. Firstly, the unknown nonlinear terms in the induction motor system are reconstructed by using the command filter and the fuzzy logic system, and the convex optimization technique is applied to construct the update law of the fuzzy logic system. Secondly, the finite-time control method is introduced to speed up the convergence of the system. Finally, the introduction of the time-varying barrier Lyapunov function ensures that the state quantities of the system are constrained within a predefined tight set. Simulation results show that the proposed method can achieve fast and effective tracking for the desired signal.
  • ZHAO Tianxing, ZHAO Zhenhua, YAN Hongtao, CAO Dong, ZU Jiakui
    Control Engineering of China. 2025, 32(6): 1121-1129.
    Abstract ( )   Knowledge map   Save
    The trajectory tracking issues of unmanned helicopters affected by actuator faults and multi-source disturbances is studied, and a full-loop continuous nonsingular terminal sliding mode control method is proposed. Firstly, the trajectory tracking problem is transformed into the commands tracking of position and attitude loops. Secondly, the high order sliding mode observer is designed to estimate the lumped disturbances. Finally, a composite continuous nonsingular terminal sliding mode controller is constructed based on the estimation of disturbances. Simulation results validate that the proposed method achieves high precision trajectory tracking even there are actuator faults and multi-source disturbances in the systems.
  • YUAN Liangliang, YAN Yan, YU Shuanghe, ZHAO Ying
    Control Engineering of China. 2025, 32(6): 1130-1136.
    Abstract ( )   Knowledge map   Save
    The fault-tolerant control problem of unmanned surface vehicle with uncertain parameters and external disturbances is studied. A barrier function based adaptive sliding mode fault-tolerant control method is adopted for trajectory tracking of unmanned surface vehicle. This method adopts an adaptive law to deal with lumped disturbance which including actuators fault, parameter uncertainty and external disturbance, according to the equivalent sliding mode control method and active fault-tolerant control scheme, an adaptive sliding mode fault-tolerant control scheme with switching adaptive law is designed. Unmanned surface vehicle can still have good tracking performance in the event of actuators fault, the addition of barrier function can ensure that the sliding variables converge and stabilize within the boundary which is unrelated to the disturbance upper bound, further improving the control accuracy and increasing the robustness of the system. The stability of the proposed scheme is proved by using the Lyapunov stability analysis criterion and verified by numerical simulation.
  • ZANG Xiangyi, WANG Cheng
    Control Engineering of China. 2025, 32(6): 1137-1144.
    Abstract ( )   Knowledge map   Save
    In order to reduce the energy consumption of vertical grinding and increase the production of ore powder, a process parameter optimization method based on genetic algorithm back propagation (GABP) neural network and strength Pareto variable adaptive chaotic differential evolution (SPVACDE) is proposed. To solve the problem that strength Pareto evolution algorithm 2 (SPEA2) is prone to fall into local optimum and unequal distribution of Pareto front in the optimization process, SPVACDE algorithm is proposed. SPVACDE algorithm introduces variable chaotic mapping strategy and adaptive crossover and mutation operators to SPEA2. Experimental results show that compared with the SPEA2 algorithm, SPVACDE algorithm has better convergence and distribution when solving ZDT test function, and can find better solutions when solving the optimization problem of vertical mill process parameters, and the Pareto front distribution is more uniform. Compared with the original process parameter design, the energy consumption of vertical mill is reduced by 11.47%, and the ore powder production is increased by 18.36%.
  • MO luying, ZHOU Chuan, GUO Jian, HAN Fei, SUN Yue
    Control Engineering of China. 2025, 32(6): 1145-1152.
    Abstract ( )   Knowledge map   Save
    For the problem of real-time and observation efficiency in the assignment of cooperative space observation tasks for distributed micro-nano satellite clusters in space scenarios, the traditional contract network to optimize the assignment results is extended. In the original bidding, bidding and evaluation, the main star released all the tasks at one time in the bidding process, and the subordinate star only selected the task with high comprehensive income to bid, and the main star only returned the winning result to the subordinate star, and returned the remaining tasks to the other subordinate stars. This process can save the time of bidding evaluation of the main star and improve the overall profit effectively by the selective bidding of the secondary star. Finally, simulation results show that the extended contract network algorithm proposed in this paper has a shorter allocation time and a higher overall income.