The CyberKnife system relies on chest markers for body surface feature extraction, yet it suffers from incomplete information retrieval and key data loss. Moreover, respiratory-induced structural coupling leads to significant regional differences in chest-abdominal motion, where extracting features from low-correlation areas degrades model accuracy. To address this issue, we propose a significance-based feature extraction method. Firstly, a significance evaluation function is constructed to quantify regional importance using three attribute values. Kernel principal component analysis (KPCA) is then applied to reduce these three attributes to a single significance value, facilitating the selection of high-correlation regions. Secondly, the target regions undergo voxelization using Octomap, and locally linear embedding (LLE) is employed to extract low-dimensional feature vectors. These vectors serve as the foundation for constructing a high-precision correlation model between body surface and tumor motion. Experimental results demonstrate that, compared to conventional marker-based methods, the proposed approach significantly reduces correlation error while enhancing model robustness and prediction accuracy.

For solving the problem of nonlinear system estimation with unknown but bounded noise, a state-estimation algorithm combines zonotopic Kalman gain prediction and parallelotope space observation strip constraint filtering are proposed. Utilizing the central difference expansion in a one-dimension context and the linearization error boundary space is calculated based on convex difference programming, feasible set and linearization error boundary demonstrated by zonotope, then get next time feasible set using Kalman gain state estimation. In the update step, the observation strip is considered as intersection by multiple strips, a new feasible set is obtained by the intersection of predicative set and strip in turn. The proposed algorithm avoids dimension increasing of feasible set, reduces conservation of state estimation space and proves its feasibility and effectiveness in the application of simulation applications.

For the problems of many inflection points and large load swing caused by the traditional manual control tower crane transporting goods, a new intelligent path planning algorithm of artificial fish swarm tower crane is proposed. According to the working environment of the tower crane, a three-dimensional map environment model is established to simulate the complex building environment with many obstacles. Combined with the operation characteristics of the crane in the building site, the traditional artificial fish swarm algorithm (AFSA) is improved. Using adaptive strategy to let the fish in the state changing in the process of optimization, timely adjust their mobile step length and the field of vision, and based on survival competition mechanism to improve the random behavior of artificial fish, to a certain extent, improved the searching capability of the algorithm, using the cubic spline data interpolation curve to get more suitable for smooth obstacle avoidance path of tower crane. The simulation results show that the improved algorithm can find an optimal obstacle avoidance path for the tower crane in the complex building environment with many obstacles.

In order to improve the detection accuracy of surface defects of industrial products, an Surface defect detection method based on improved YOLOv5 is proposed. Firstly, data enhancement is carried out by mixup, mosaic and traditional methods, and the residual unit is modified on the basis of YOLOv5 to reduce the floating-point calculation of the model. Secondly, squeeze and exception (SE) attention mechanism is inserted into the end of the feature extraction layer and the head of the neck to remove useless background interference in the feature map and improve the efficiency of feature extraction. Finally, a contextual transformer (CoT) module is inserted at the end of the neck to improve the average detection accuracy. And the improved shape-intersection over union non-maximum suppression (SIoU-NMS) is used to eliminate duplicate target boxes. The experimental results show that the average detection accuracy of the proposed algorithm is 81.2% and 79.7% on the new material floor defect data set and bottled Baijiu defect data set, which is 3.8% and 4.6% higher than the YOLOv5 network model as the baseline, and is superior to other typical target detection algorithms. It shows the accuracy of the algorithm in identifying and classifying the surface defects of industrial products, and can better complete the quality inspection process of industrial products.

A disturbance observer-based composite controller is designed to solve the trajectory tracking control problem of underactuated surface vessel, which suffers the external environmental disturbance and has bounded input constraint. Firstly, a fast finite-time disturbance observer (FTDO) is designed and its effectiveness is demonstrated by Lyapunov stability analysis. Compared with the existing FTDO, the new observer enables the estimation error to quickly converge to zero even if the initial error is far from the origin. Then, based on the fast FTDO, a new trajectory tracking composite controller is designed by using the backstepping method, which realizes that the trajectory tracking error of the surface ship control system tends to be in the neighborhood of an arbitrarily small radius of the zero point within a limited time. Finally, the designed controller is compared with the existing controllers through simulation, which verifies the effectiveness of the proposed method.

The current coal mill outlet temperature setting value in power plants has the disadvantage of not taking into account the characteristics of the changing coal quality parameters of blended coal and the uncertain influence of the pulverizing process. In order to further enhance the rational use of energy, a study on feedforward and feedback compensation modeling of the outlet temperature set value of coal mill in power plants is proposed. Firstly, to solve the problem that the characteristics of coal quality parameters of mixed coal change in the coal mill outlet temperature control system are not considered, we propose to form a feed-forward modeling of the mill outlet temperature set value by online parameter monitoring of volatile conten and moisture of mixed coal. Then, due to the existence of uncertain factors such as environment, coal type change or equipment aging in the pulverizing process, the traditional PI control accuracy is reduced. A feedback fuzzy model with triggering mechanism and discrete compensation is established based on expert domain knowledge, and solve the drawback that the actual output temperature is difficult to converge to the temperature setpoint. Finally, the proposed method is validated experimentally. The results show the feedforward and feedback modeling and control can ensure safe production in mixed coal parameters and pulverization optimization process, and enhance the rational utilization of energy.

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.

Because the industrial cyber-physical system has an amount of high-dimensional data, and there are huge differences in the amount of various types of data, making it is difficult to accurately detect rare samples for an attack detection system. To enhancing the model’s ability of detecting rare samples, a multi-agent adversarial training mechanism is designed on the basis of deep reinforcement learning to dynamically increase the training times of rare samples, and also a priority training mechanism is adopted to realize the priority learning of hard-to-detect samples. To verify the performance of the proposed attack detection model, a real data set produced with Modbus communication protocol is used for testing. The results show that, compared with the traditional attack detection methods, the designed attack detection model can significantly improve the attack detection ability of rare samples.

Introducing a novel method of self-adjusting sliding mode control using a recurrent neural network for guiding the tilt-quad-rotor UAV in both helicopter and fixed wing modes with varying tilt angles. Firstly, the dynamic model of the UAV is separated into two distinct subsystems: actuated and underactuated. Considering the parametric uncertainties and external disturbances of UAV, the equivalent control law obtained by sliding mode control approach cannot be applied to UAV. Thus, a recurrent neural network (RNN) controller is used to mimic the equivalent control. Subsequently, a new switching controller is devised to reduce control chattering and maintain system stability. Utilizing Lyapunov stability theory, both subsystems are capable of reaching the sliding surface. Finally, simulation results confirm the efficacy of the suggested method.

In order to achieve accurate temperature control in the air-drying area of the hard capsule production line, the problems of large inertia and large time delay are solved. A temperature control system using Smith-fuzzy PID controller is proposed. Firstly, according to the experience and actual situation of temperature control related to the production line, a mathematical model of the temperature control system is constructed. Secondly, the PID temperature controller is designed based on the mathematical model, the fuzzy self-adjustment of the three parameters of PID is carried out by using the fuzzy control algorithm, and then the Smith estimation method is introduced for delay compensation, and the Smith-fuzzy PID control temperature control system is designed. Finally, a simulation comparison of the three PID controllers is carried out in MATLAB-Simulink. The results show that the Smith-fuzzy PID control system has the advantages of basically zero overshoot, fast adjustment time and better stability. Combined with the experiment, the temperature of the air-drying zone is controlled within the range of (40±0.5) ℃ in the whole process, which further proved the feasibility and effectiveness of the temperature control method.

As a typical urban renewable resource, magnetic storage media need to be destroyed in order to ensure privacy and prevent leakage before resource utilization. At present, the industry mostly uses the highest magnetic field intensity for degaussing, resulting in a waste of resources. This is not in line with the current national “two-carbon” strategy. In addition, there is a lack of theoretical exploration on how to conduct customized demagnetization. To solve this problem, a magnetic field numerical modeling and analysis method for customized demagnetization of magnetic hard disk is proposed. First, the magnetic field intensity of multi-layer and multi-turn rectangular coil commonly is calculated theoretically. Then, the magnetic field is numerically modeled and analyzed based on COMSOL. After verifying the consistency between the model and theoretical calculation, the relationship between current density, coil height, package size and magnetic field strength is analyzed. Finally, verification is carried out based on the degaussing positive prototype. The research results verify that the customized degaussing a

To address the no-uniform distribution of alumina concentration in large-scale aluminum reduction cells, a feeding strategy based on adaptive distributed subspace prediction control is proposed. Firstly, the aluminum reduction cell is divided into multiple interconnected subsystems according to the spatial distribution of feeding ports, where adaptive distributed subspace predictive models for the alumina concentration of each subsystem are directly constructed from online data. Secondly, a residual-driven model switching mechanism is designed to trigger parameter updates only when model deviations exceed thresholds, thereby balancing computational efficiency and model accuracy. Finally, a distributed predictive controller based on Nash optimality theory is developed to achieve global coordinated feeding optimization. Simulation tests using real production data from an aluminum plant demonstrated that the proposed method significantly improved the spatiotemporal uniformity of alumina concentration in large-scale aluminum reduction cells.

In collaborative filtering algorithms based on formal concept analysis, the construction of concept lattices typically requires substantial computational resources, limiting their application to large-scale datasets. To address this challenge, the introduction of pseudo-strong concepts significantly reduces resource consumption. Additionally, by partitioning the “nearest neighbor” sets and integrating impact factors with similarity measures, the recommendation performance can be further enhanced. A collaborative filtering recommendation algorithm based on user pseudo-strong concepts is proposed. The algorithm first partitions the “nearest neighbor” user groups using the extent sets of user pseudo-strong concepts and defines corresponding similarity calculation methods for different types of users. Subsequently, it computes prediction ratings by integrating mean squared error impact factors and similarity measures. Experimental results on the MovieLens-100K dataset demonstrate that, compared to traditional user-based collaborative filtering algorithms, the proposed algorithm achieves significant improvements in evaluation metrics such as recall and 1F score, validating its effectiveness and practicality.

This problem of the platoon control of connected vehicles on curved roads is investigated. Firstly, a novel vehicle dynamics model is established considering curved road. Secondly, a time-delay switched tracking error system is constructed for platoon, which considers the impact of packet dropouts and actuator delay. Then, combined with the switching system theory and Lyapunov stability theory, the existence condition of the controller is obtained to design the feedback controller, which can guarantee the robust stability and string stability for the vehicle platoon control system. Finally, the effectiveness of the proposed algorithm is verified by simulation.

For the estimation accuracy of rotor speed and position and system chattering in sensorless control of permanent magnet synchronous motor (PMSM), an improved adaptive sliding mode observer (ASMO) method based on pre-filtered dual phase-locked loop (P-DPLL) is proposed. Firstly, the continuous sigmod function with nonlinear properties is selected as the switching function, then the Lyapunov function is constructed, and a detailed stability condition is obtained. On this basis, the back electromotive force error is estimated by combining the adaptive algorithm and the sliding mode current observer. Finally, P-DPLL is designed to extract the accurate rotor position information of PMSM under positive and negative acceleration and deceleration conditions. The experimental results show that the improved ASMO method based on P-DPLL can effectively improve the accuracy of rotor speed and position estimation in PMSM sensorless control and reduce system chattering.After the motor is disturbed, the average recovery time is reduced by 0.02 s, and the average position error is reduced by 6.7 %.

In order to save network resources and improve the control performance of complex systems, for nonlinear system model of rotating inverted pendulum, a funnel tracking control scheme under event triggered mechanism is proposed. Firstly, by utilizing funnel control method, normalization function and barrier function related to tracking error are constructed. Secondly, a funnel control law is designed based on backstepping. Finally, an event triggered funnel controller is designed, and a new event triggered mechanism is proposed. By introducing a power function, the triggering threshold related to the power function of funnel control law is designed, which can save control resources while ensuring system performance. Theoretical analysis shows that: ① all signals in the closed-loop system are bounded; ② The controller ensures the inverted pendulum rod to track a reference signal; ③ For 0t∀≥, the tracking error always evolves within a prescribed performance funnel, and Zeno phenomenon is excluded. Finally, the practicality and effectiveness of the proposed scheme is verified by simulation.

The detection of the installation orientation of the Mprop is a key issue in the quality inspection of the oil pump. For the problem of the classical object heading detection algorithm based on orientation object detection, which has low accuracy and limited detection angle, a full-angle object heading detection algorithm named YOLOX-FA (YOLOX-full angle) is proposed innovatively. Firstly, a six parameters notation method is designed. By improving the decoupled head structure of the YOLOX algorithm, all six parameters can be output. Then the target position, width, height, orientation, and category can be detected simultaneously. Secondly, the convolutional block attention module (CBAM) embedded in the front end of the decoupled head can enhance the model's learning ability for the target’s key features and improve recognition accuracy. Finally, by introducing an angle deviation correction coefficient, full-angle detection of oil pump’s Mprop module is completed with the improved rotation-robust intersection over union (RIoU) loss function. The experimental results show that YOLOX-FA has higher detection accuracy than the OHDet and can be used to achieve the full-angle high precision heading detection of oil pump’s Mprop module. The detection accuracy rate reaches 94.51%.

Pose graph optimization (PGO) is one of the most important back-end optimization techniques in simultaneous localization and mapping (SLAM). It aims to search for solutions with global optimality. To verify the global optimality of a given candidate solution, a dual based fast 3D pose graph verification technique is proposed. Firstly, the original PGO problem formula to establish an equivalent optimization problem with lower complexity is revised. Secondly, the corresponding Lagrange duality problem, and evaluated the quality of the candidate solution by the properties of the duality problem is derived. Finally, experimental evaluation on simulated and real-world SLAM datasets shows that the proposed algorithm has the advantages of scalability and low computational cost, and can effectively verify optimality.

Addressing the premature convergence, rapid convergence speed, and challenges in balancing global and local search capabilities, as well as achieving high optimization accuracy in the gravitational search algorithm, a jump-gravitational search algorithm (JGSA) with a random update velocity mechanism to escape local optima is proposed. Initially, the algorithm’s population quality is enhanced using inverse populations. Subsequently, the learning strategy from the particle swarm optimization (PSO) is introduced to balance global and local search capabilities. Finally, a random update velocity is designed to facilitate escaping local optima when trapped. The JGSA is validated through simulations on six benchmark functions and its application in scheduling for steelmaking-continuous casting production. Comparative analysis with other algorithms demonstrates that the JGSA performs superior optimization in seeking optimal scheduling plans.

For the limitation of existing control methods for obtaining intermediate states, a practical control method (stateless control) is proposed for suspension system by combining the linear quadratic regulator with the Luneberger observer. It directly establishes the relationship between current control input with historical control inputs and outputs. Using a quarter-car suspension system model as an example, the practical control performance of the stateless method is investigated. It incorporates sensor noise, modeled according to sensor precision, and accounts for suspension system parameter uncertainties. The results demonstrate that under both excitation and random excitation, the stateless control method effectively balances the three ride comfort performance indices of the suspension system, achieving excellent overall control performance. Crucially, the stateless control method only requires historical sensor outputs and control inputs to realize optimal control effectiveness even with limited sensors, thus significantly simplifying the control system design and offering significant practical value.

For the problem of multi-AUV underwater cooperative task assignment, the auction algorithm is used to partially improve the selection process of bee colony optimization, so that the result of task assignment is closer to the global optimum. On the modelling issue, take the AUV energy reserve into account. In the calculation, the hierarchical task allocation mode is adopted to reduce the data dimension in the calculation. In the first-level target allocation, the auction algorithm is used for accurate one-to-one allocation. In the selection process of the second stage, the improved bee colony algorithm is used for iterative optimization, and a contraction factor is introduced into the fitness to make the final result of task assignment closer to the Pareto optimal state. The experimental results show that the scheme has a fast convergence speed and good stability.

Periodic soot blowing is a common practice to improve thermal efficiency and to ensure safe operation of the utility boiler. Taking a 75 t/h circulating fluidized bed boiler as an example, a method to calculate the ash accumulation factor of low temperature convection heat exchangers is proposed. The calculated ash accumulation factor together with the data provided by the DCS are used as input variables to build the boiler thermal efficiency estimation model based on the XGBoost. The boiler thermal efficiency estimation model is used to calculate the impact of the soot blowing period on thermal efficiency. The simulations demonstrate that the maximum benefit is followed with the soot blowing period of 8 hours. The approach has the potential for developing intelligent soot blowing strategies.

For the current error problems caused by model mismatch, parameter perturbation and external disturbance in the internal model control of permanent magnet synchronous motor, an adaptive Kalman filter based on internal model control is designed to realize the compensation control of the current loop. In the case of current loop decoupling, an internal model controller is established, and the system disturbance caused by model mismatch, parameter perturbation and external disturbance and the feedback current of the internal model controller are used as state variables to construct the state equation of the system, and the extended state model is established. The adaptive Kalman filter under the model, on the basis of reducing the computational complexity, observes the system disturbance and provides real-time disturbance compensation for the internal model control. The simulation and experimental results verify that the proposed algorithm can realize the current when the model is mismatched. The compensator can effectively reduce the current ripple, reduce the steady-state error of the current, and improve the robustness of the system.

A cooperative output-feedback secure control scheme based on observers is proposed for connected and autonomous vehicle systems with intermittent denial-of-service (DoS) attacks on communication. Firstly, the dynamic model of the longitudinal connected and autonomous vehicle system is analyzed, and the feedback is linearized to obtain the linear dynamic equations. Secondly, by using the common Lyapunov function, a secure control scheme is designed to make the cooperative tracking error asymptotically stable. Finally, for maximizing the duration of DoS attacks, appropriate parameters are selected to design an optimization algorithm, in order to ensure the safe operation of the connected and autonomous vehicle system. The experiment simulated a networked vehicle system consisting of 4 followers and 1 leader, and the simulation results verified the effectiveness of the proposed method. The experiment is conducted by simulating a connected and autonomous vehicle system consisting of 4 followers and 1 leader. The simulation results verify the effectiveness of the proposed method.

In order to improve the dynamic performance of asynchronous motor control system, a joint control strategy based on internal model control and fuzzy sliding mode control is proposed. Firstly, the internal model controller with two degrees of freedom is improved by fractional order function, and the anti-disturbance performance and tracking performance of the controller are separated. Then, considering that the motor is sensitive to external disturbance at low speed, a load disturbance observer based on hyperbolic tangent function and Fal function is designed to estimate the load disturbance in real time, and the estimated value is compensated to the sliding mode controller. Finally, in order to further overcome the chattering problem of the sliding mode controller at low speed, the non-singular fast terminal sliding mode surface is improved, and the parameter of the exponential reaching law is selected in real time by the fuzzy control principle. The simulation results show that the proposed control strategy can improve the tracking performance and anti-disturbance performance of the asynchronous motor at low speed.

To solve the problem that it is difficult for dual robots to achieve reasonable task allocation and cooperative operation, a method based on workload balance mechanism and master-slave cooperative ant colony optimization algorithm is proposed to accomplish task allocation and cooperative operation of dual robots. Firstly, an unbalanced task allocation model is established based on the task point set, and the path planning algorithm is iterated in the task assignment stage to balance the workload of the two robots. Then, the master-slave cooperative ant colony optimization algorithm is used to solve the multi-objective cooperative operation optimization model that avoids interference among robots and keeps the workload minimum. Finally, the experiment is carried out in combination with the reinforcement binding scene, the experimental results show that the proposed method can achieve reasonable task allocation between the two robots, reduce the workload difference between the two robots, make them efficiently complete the reinforcement binding operation, and effectively avoid the robot interference in the operation process.

To maintain the stability of continuous stochastic systems driven by the Lévy process, a discrete control method is proposed. Firstly, a sliding mode controller is designed, which ensures that the system can maintain mean square exponential stability under Lévy noise interference by designing the control law reasonably. Secondly, the continuous controller is discretized to meet the requirements of actual digital control systems. Thirdly, through deduction, it is concluded that the second-order moment of the difference between
the system state under the action of a discrete controller and the system state under the action of a continuous controller is bounded, indicating that the discretization process does not significantly increase the instability of the system. Finally, the stability of the closed-loop system under the action of a discrete controller is demonstrated through theoretical analysis. The experimental results show that the discretized controller can still keep the system stable.

In the case of limited network communication resources and actuator saturation, it is difficult to ensure trajectory tracking, communication maintenance and collision avoidance for the nonholonomic multi- robot formation. Therefore, an adaptive dynamic event-triggered formation control method is proposed. Firstly, the fuzzy logic systems are employed to approximate the unknown functions in robot dynamics, and a fuzzy adaptive control scheme is proposed based on the backstepping method and dynamic surface technology. Secondly, the barrier function and the prescribed performance approach is used to realize the communication maintenance and collision avoidance between multiple robots, and the dynamic event-triggered mechanism is designed with saturated actuator to reduce the execution number of the controllers and further reduce the communication burden from controllers to actuators. Finally, simulation results demonstrate the effectiveness of the proposed method.

For the steady-state operation problem of continuous industrial processes, a method of selecting the optimal measurement combination of global self-optimizing control based on constraint processing is proposed. The relationship between constraint and disturbance is deduced by establishing a linearized model of constraint. In order to select a subset of measurement variables, a penalty term is added to the global self-optimizing control optimization problem, and the optimal balance between the steady-state loss and the number of measurement variables was achieved by sparse column elimination of measurement variables that had little influence on the system control. On this basis, the interior point method of the penalty function is introduced to deal with the change of active constraints. Because the interference usually causes the change of active constraints in the actual industrial process, it is more realistic to consider the change of active constraints in the selection of global self-optimizing control measurement combination. The simulation results of the evaporation process verify the effectiveness of the proposed method in constraint treatment and measurement combination selection.

To solve the problem of three-dimensional path planning of the unmanned aerial vehicle (UAV) in complex environment, a compound algorithm is proposed by integrating the traditional particle swarm optimization (PSO) algorithm and the grey wolf optimization (GWO) algorithm, called the PSO-GWO compound algorithm. Firstly, nonlinear control parameters and weighted adaptive individual location update strategy are used to balance the global search capability and local search capability of the algorithm. Secondly, the random guidance strategy is used to increase the diversity of the solutions. Finally, B-spline curve is used to smooth the generated flight path to make the path more suitable for the UAV. The experimental results show that the PSO-GWO compound algorithm can generate a safe and feasible path, and its performance is significantly better than that of the GWO and other improved GWO algorithms.

There are problems in the optimization process of pelican optimization algorithm, such as the reduction of population diversity, the decrease of convergence speed, and the tendency to fall into local optimum. To solve these problems, multiple strategies are integrated to improve pelican optimization algorithm, and the improved pelican optimization algorithm (IPOA) is proposed. Firstly, the pelican population is initialized by using the tent chaos map and refracted opposition-based learning strategy, which not only increases the population diversity, but also lays the foundation for improving the optimization ability of the algorithm. Then, a nonlinear inertia weight factor is introduced at the stage when pelicans approach their prey to improve the convergence speed of the algorithm. Finally, the leader strategy of salp swarm algorithm is introduced to coordinate the global search ability and local optimization ability of the algorithm. The improvement effect of the single improvement strategy is tested and IPOA is compared with 9 other optimization algorithms in the experiment. The experimental results prove the effectiveness of each improvement strategy and the superiority and robustness of IPOA.

A grabbing and recycling method with visual guidance for the vehicle-mounted rotor unmanned aerial vehicle (UAV) is proposed, which provides a feasible solution for the recycle of the rotor UAV. Firstly, an improved object detection is proposed to ensure the real-time and accurate detection of UAVs on edge devices. Then, ZED2i camera is used to obtain the three-dimensional coordinates of the center point of the UAV in the point cloud data, and hand-eye calibration method is used to obtain the coordinate system transformation relationship between the camera and the end of the robotic arm. Finally, combined with the self-posture information measured by the camera sensor, the posture of the end of the robot arm is adjusted to adapt to the grabbing. The experimental results show that the proposed method can realize the real-time detection and accurate grabbing and recycling of UAVs.

For the problem of low distributed fault diagnosis accuracy due to the heterogeneous data of wind turbine bearings and small sample, a fault diagnosis method for wind turbine bearings based on weighted federated distillation (WFD) is proposed. Firstly, an improved weighted federated parameter update method is proposed. Cosine similarity is calculated for the wind turbine client to update parameters, thereby achieving personalized federated learning. Then, a weighted federated distillation method based on convolutional neural networks is proposed. Knowledge distillation is conducted between the teacher network and the student network. Finally, the loss function is minimized by using mini-batch stochastic gradient descent to classify the types of bearing faults. The proposed method effectively learns knowledge of other clients through federated distillation in case of smalls, and improved weighted federation method for parameter update can effectively reduce feature difference caused by data heterogeneity. Through simulation, compared with other methods, the proposed method has higher accuracy and better generalization performance.

Taking a typical cold rolling production system as the research background, the batch production planning problem for high-value-added O5 products is studied. Due to the extremely high surface quality requirements of these products, the associated setup costs and inventory costs in the production process are significantly high. In light of this, a mixed-integer programming model is formulated to describe the batch production planning problem for O5 products, aiming to minimize setup costs and inventory costs with full consideration of practical technological constraints. Additionally, an effective Lagrangian relaxation approach is proposed to solve the problem. A numerical experiment composed of 400 instances is designed based on actual production data. Computational results prove that high quality solutions could be found in a reasonable time.

A proportional integral differential (PID) controller parameter optimization method based on the subspace model is proposed to improve the performance of the PID controller. Firstly, the explicit expression of the controller performance with respect to the PID controller parameters is deduced by using the subspace matrix equation. Then, the subspace matrices corresponding to the process model and the random disturbance model are identified by using the closed-loop data with set point excitation, and the estimated dynamic matrix is directly applied in the calculation of the best performance to obtain the optimal controller parameter values. Finally, the effectiveness of the proposed method is verified by a numerical simulation and an industrial example.

The sound signals of belt conveyors contain a large amount of operating status information, so the acquisition of sound signals is crucial for the diagnosis of belt conveyor faults. The existing wavelet threshold denoising algorithms can not meet the requirement of extracting weak sound signals in strong noise background. Therefore, the conventional wavelet threshold denoising algorithm is improved, and a continuous low-error wavelet threshold function with adaptive wavelet threshold is proposed. The adaptive wavelet threshold is a segmentation function about the number of decomposition layers, which is inversely proportional to the number of decomposition layers, and can better adapt to the feature that the noise coefficient decreases with the increase of the decomposition layers when wavelet decomposition is performed. The experimental results show that the improved algorithm has stronger denoising ability and more accurate reconstruction of the original signal compared with hard threshold denoising algorithm, soft threshold denoising algorithm and wavelet modulus maximum denoising algorithm.

The basis for predictive control of air conditioning terminal systems in data centers is the multi-step prediction of cabinet inlet temperature. To improve the precision and the portability of the prediction model, a nonlinear Takagi-Sugeno (T-S) fuzzy method is proposed to construct the temperature prediction model of data center server rooms. Firstly, the computational fluid dynamics (CFD) model of a server room is developed by using CFD numerical simulation method, and a data acquisition strategy is designed to capture the complete dynamic characteristics of the system. Then, to solve the problem that the fuzzy C-mean clustering algorithm is prone to local optimum, the improved beetle antennae search algorithm is proposed to optimize the identification of the forepart structure of the T-S fuzzy model. Finally, the cubature Kalman filter is adopted for the posterior part parameter identification and online correction of the T-S fuzzy model. The experimental results show that the T-S fuzzy model constructed by this method has higher computational efficiency and prediction accuracy than that of the conventional T-S fuzzy model, and the requirement of model portability can be met through the update of the posterior part parameters.

In order to improve the intelligent decision-making ability of the multi-agent system, a multi-agent reinforcement learning algorithm based on prioritized value network is proposed, the disadvantages of experience replay of multi-agent reinforcement learning and the problems of emphasizing action value and ignoring state value in agent decision-making are solved. Firstly, the algorithm introduces a preferential experience replay mechanism to reuse experience according to importance weights, which solves the problem of experience reuse through random sampling. Secondly, the value advantage network is introduced into the value network of the agent to compare the information of state value and action advantage, which makes the agent learn the dominant action fast. The experimental results of multiple collaborative scenarios show that the algorithm can improve the learning and cooperation quality of the multi-agent system, so that the agent can make decisions faster and better, and complete the given task.

Ideological and political education reform is a new concept and practice for colleges and universities in China to implement the fundamental task of cultivating morality and cultivating people. The improvement of ideological and political education ability of teachers of professional courses is an important focus and key to the ideological and political education construction of college courses. Therefore, we analyze the dilemma faced by the improvement of ideological and political education ability of teachers of automation courses, put forward the ideas of improving ideological and political education ability of teachers of automation professional courses. In addition, taking the Department of System Science of the School of Mathematics of Southeast University as an example, the practice of improving ideological and political education ability of teachers of automation professional courses is expounded.

In order to improve the joint flexibility of the upper limb exoskeleton robot, an upper limb flexible exoskeleton robot with modular series elastic actuators and Bowden cable is proposed. To reduce the nonlinear friction, unknown external disturbance and model uncertainty caused by Bowden cable, an adaptive composite controller based on radial basis function (RBF) neural network is proposed. The disturbance observer and RBF neural network adaptive controller are used to estimate and compensate the disturbances, and the sliding mode controller is used to implement the tracking control of the upper limb flexible exoskeleton robot. In addition, the stability of the controller is proved by Lyapunov theory. The simulation results show that the proposed controller has better disturbance compensation capability, higher tracking control accuracy and robustness compared with the conventioal proportional integral differential (PID) controller and sliding mode controller, and can realize the precise tracking control of the upper limb flexible exoskeleton robot.