To address the difficulties in stochastic distribution control teaching, including high theoretical abstraction, the gap between simulation and real industrial constraints, and limited experimental resources, a pure physical experimental system based on a vertical disc grinding process was developed. The system integrates platform construction, data acquisition, modeling, and control into a unified framework. Centered on a self-developed vertical disc mill, it incorporates precise feeding, disc gap adjustment, and speed regulation, while real-time particle size distribution data are obtained through automatic intermittent sampling and a dry laser particle size analyzer. Radial basis function expansion and inverse integration are used to parameterize the output probability density function. Combined with iterative learning and subspace identification, a linear prediction model describing the dynamic relationship between manipulated variables and weight vectors is established. A constrained controller is further designed to achieve target distribution tracking under input constraints and disturbances. Experimental results show that the system is stable and interactive, and can effectively support experimental teaching in stochastic distribution control and related courses.

The problem of memoryless feedback stabilization for a class of linear systems with unknown input delays is studied. For a class of unstable linear systems whose open-loop poles are located on the imaginary axis, a finite-dimensional memoryless feedback control scheme is proposed when the input delay is unknown. By designing a memoryless truncated predictor feedback (TPF) controller, the implementation problems brought by traditional infinite-dimensional control schemes are avoided. When the unknown input delay is in a time interval, the global asymptotic stability of the closed-loop system is guaranteed. An explicit expression for the variation range of the unknown input delay is given. Finally, the designed memoryless truncated predictive feedback controller is applied to the delayed dual oscillator system and anti-pitching control of high-speed catamaran, and its effectiveness is verified through MATLAB.

For the calciner for collaborative waste disposal in a cement firing system, a method of modeling and predictive control based on the MISO Hammerstein model was studied. The model takes the outlet temperature of the calciner as the output, the coal feeding amount and waste flow rate as the inputs. The nonlinear part of the model is described by polynomials, and the linear part is described by the ARMAX model. The parameters of the linear and nonlinear parts in the model are identified by combining overparameterization and singular value decomposition methods. The control strategy is divided into two steps. First, the intermediate variables are solved by generalized predictive control algorithm. Then control variables are solved by the inverse of the nonlinear model. The results of the simulation experiment show that the calciner outlet temperature model based on the MISO Hammerstein model has a good fitting effect, and two-step predictive control makes outlet temperature of the calciner well tracking the set value and has a certain anti-interference ability.

To improve the calibration efficiency and accuracy of adaptive optics systems in complex environments, a control algorithm for adaptive optics systems that combines stochastic parallel gradient descent algorithm and convolutional neural network is proposed. The algorithm classifies wavefront aberrations through convolutional neural networks, optimizes initial estimation, reduces iteration times, and accelerates convergence speed. Experimental results show that compared with the original stochastic parallel gradient descent algorithm, the mixing efficiency of the new algorithm is improved from 0.80 to 0.90, the number of iterations is reduced from 280 to 140, the bit error rate is reduced to 10−10 after 158 iterations, and the root-mean-square value is reduced from 0.2~1.2 to 0~0.6. The aberration correction efficiency and accuracy of the adaptive optical system under atmospheric turbulence are effectively improved. The study combines innovatively stochastic parallel gradient descent algorithm with convolutional neural network and introduces Dropout mechanism to enhance the generalization of the model, providing a fast and effective new method for wavefront correction in adaptive optics systems.

In order to improve the accuracy of surface defect detection for metallic parts in Metallic production lines, a defect detection method based on an improved Faster R-CNN is proposed. Firstly, we introduced more receptive fields in the feature extraction stage, thereby allowing finer feature extraction. For the region proposal phase, a cascaded RPN network and adaptive convolution structure were employed to gradually refine the positioning of the proposal box. For the interest region feature extraction, the feature information from all output layers of the FPN was aggregated to fully utilize the defect feature information obtained, enhancing the network's detection capability for small and irregularly shaped defects. Finally, experimental results showed that the detection accuracy of our method outperformed that of the state-of-the-art object detection networks, and our method hence is of high value to be ultilized to detect surface defects such as corrosion, scratches, scuffs, and built-up edge of metallic parts.

In industrial control, PID controllers are widely used due to their simple structure and convenient operation. More than 95% of industrial controllers adopt PID control. The parameters PK,IK,DK directly affect the control performance. Traditional PID tuning methods suffer from long cycles, high difficulty, and poor stability, which make it difficult to meet the requirements of industrial applications. To solve these problems, a PID parameter self-tuning method based on the improved Harris hawks optimization (IHHO) algorithm is proposed. By introducing adaptive inertia weight, strengthening sine-cosine optimization operator, nonlinear escape energy factor, and Tent chaotic mapping, the problems of slow convergence and easy falling into local optimum in the original Harris hawks optimization (HHO) algorithm are improved. Tests show that IHHO is superior to many intelligent algorithms in optimization accuracy and convergence performance. When applied to DC motor speed regulation, MATLAB simulation results demonstrate that PID tuning based on IHHO can significantly improve the response speed and stability of the system, making the speed output stable. The method can be widely used for PID parameter self-tuning.

Research on product quality prediction for intermittent processes with limited data and nonlinear characteristics. In multi-source domain adaptation, the difference in the amount of data in the source and target domains leads to data imbalance, while there is nonlinearity in the process, which further leads to the inability to model accurately. To address this issue, the SMOTE method is first employed to eliminate the adverse effects of data imbalance on the model’s prediction results. Then, a nonlinear MDAJYPLS algorithm based on RBF networks is proposed to capture the nonlinear characteristics and make full use of the information from multiple similar source domains to assist the target domain process modeling. Further, a batch process quality prediction based on non-linear multi-source domain adaptive JYPLS via SMOTE is proposed, which can improve the prediction accuracy of the model. Finally, the effectiveness of the proposed method is verified by simulation experiments of penicillin fermentation process quality prediction.

Due to the randomness of wind energy, wind farms will abandon wind and maintain stability and cause serious impact on the grid, and there is a lack of communication between wind turbines. In this regard, a dual-mode distributed economic model predictive control strategy is proposed. Firstly, based on the analysis of the characteristics of the wind farm system with energy storage, a one-machine-one-storage distributed energy storage wind farm system model is established. Then, a dual-mode distributed economic model predictive control strategy is designed to propose the global economic optimization control objective and realize the collaborative control of subsystems. Finally, the simulation shows that the dual-mode distributed economic model predictive control strategy can effectively improves the system closed-loop stability and overall economy, and realizes the mutual communication between subsystems and reduces the loss. Its control effect is obviously better than that of centralized and decentralized economic model prediction control.

The Stewart platform is widely used in various motion simulators, where pose error determines the fidelity of the simulator. Therefore, for the posture error caused by the hinge gap error during the movement of the platform, an error compensation scheme based on optimization algorithm is designed. Firstly, the kinematic inverse solution is used to find the elongation of the driving rod under the trajectory of the positioning attitude. Secondly, the vector method is used to construct the pose error model, and the relationship between pose error and the driving rod is studied according to the pose error model, and the pose error is designed as the objective function with the driving rod as the variable. On this basis, an improved generalized normal distribution optimization (IGNDO) algorithm integrating Levi flight and chaotic mapping is proposed to find the optimal value of the objective function and the optimal solution of the driving rod. Finally, the optimal solution is substituted into the error model to output a new pose error, and compared with the theoretical error, which verifies the effectiveness of the algorithm. Meanwhile, comparing the IGNDO algorithm with the existing algorithm, it is found that the pose error can be reduced by 1-2 magnitude.

A quasi-reset control algorithm is proposed for second-order multi-agent systems. By designing a class of continuous nonlinear functions to replace the “reset” behavior of the conventional reset algorithm at the zero point, the continuity of the control algorithm is ensured while the transient performance of the system response is improved. By constructing a suitable Lyapunov-Krasovskii functional, the velocity of the agents can achieve average consensus is demonstrated, and further ensures the the worst-case transient performance of the system response. And the boundedness of the integration term is proved. Finally, the system responses under different algorithms are compared through simulation, which verifying the correctness of the proposed theory.
Key words: Second-order multi-agent system; average consensus; quasi-reset control; Lyapunov method;

A Funnel tracking control and sliding mode synchronization control method based on finite time extended state observer (FTESO) is designed for dual motor drive servo systems with nonlinear dead zones. The nonlinear friction, clearance, and external disturbances of the system are considered as the total disturbance of the system. FTESO is designed to estimate them and the estimated values are embedded in the controller to compensate for them. By designing the Funnel function to constrain the tracking error and convert it into a new error, it is introduced into dynamic surface control to ensure that the transient and steady-state responses of the tracking error in the system are well limited within the given Funnel boundary. On this basis, a sliding mode synchronous control strategy based on mean deviation coupling strategy is designed to ensure the synchronous operation of the dual motors. Finally, the effectiveness of the proposed control algorithm is verified through simulation.

For the problem that the structure design and the fuzzy PID speed control strategy of omni- directional mobile robot with independent drive caster-wheel have not been verified in the physical platform, the motion mechanism and control system of omni-directional mobile robot with independent drive caster- wheel is designed. Firstly, the integrated hub motor suspension is designed by using hub motor and slip ring, and the mechanical structure of the three-wheel structure is set up. Secondly, the robot control system is designed with industrial programmable logic controller (PLC) as the main controller. Three motors are controlled synchronously by CANopen communication protocol to realize the omni-directional movement of the robot. Finally, a three-parallel fuzzy PID control strategy is proposed based on the kinematics model, and the correctness of the modeling and control algorithm are verified by Simulink-Adams co-simulation. The rules of controller tuning are summarized, and the fuzzy rule base is established; the physical test is carried out based on the joint simulation experience and the built physical platform. The results show that the mechanical structure of the physical platform is stable and the control algorithm is effective.

To address the problems of insufficient accuracy, slow convergence, and susceptibility to external disturbances in manipulator visual servo systems, a composite control scheme combining continuous terminal sliding mode control (CTSMC) and an extended state observer (ESO) is proposed. First, a position-based kinematic model of the manipulator visual servo system is established by considering system uncertainties and possible external disturbances. Then, the ESO is employed to estimate the lumped disturbances and model uncertainties, and the estimated total disturbance is fed forward to the feedback controller for compensation. Based on the visual servo model, a continuous terminal sliding mode surface is designed, and a finite-time visual servo control law is developed. Furthermore, the stability of the whole system is analyzed using Lyapunov theory. Simulation and experimental results on the Diana manipulator show that, compared with traditional sliding mode control methods, the proposed method achieves finite-time convergence and better visual servo tracking performance.

Autonomous Underwater Vehicle (AUV) clustering is an important trend for future underwater intelligent warfare. In order to fully exploit the overall combat advantages of AUV clusters, an underwater adversarial environment model is established using performance indicators such as target value destruction, cluster loss cost, and relative path length. An improved wolf pack algorithm is used in combination with integer encoding to solve the target allocation problem. To address the problem of the wolf pack algorithm being prone to local optima, an effective generation distance concept is introduced in the wolf pack update mechanism to improve global optimization ability of the algorithm. Simulation results show that the proposed method has good adaptability and stability for different adversarial situations, as well as strong robustness, and can significantly improve convergence speed.

The fatigue test of aircraft structure plays a key role in the development of new aircraft. The structural fatigue testing system is a multi-component and multi-channel system. It is crucial to obtain the dynamic characteristics of the fatigue testing system before its operation for the smooth conduct of the test. Therefore, mathematical models of each component of the test system are established and coupled to form a semi-physical simulation model of the test system. A real-time computing platform is built, and through offline and online real-time simulation, the dynamic characteristics of the test system are analyzed. It is concluded that the displacement generated by the test article under loading is the external disturbance of the system; The coupling effect between the multi-channel of the test system mainly affects the phase characteristics of the output signal, with a phase difference of 170°. The rationality and effectiveness of the proposed method are verified through cross validation of offline and online real-time simulations.

For the issues that current coal blending in power plant bunkers, which relies on manual experience, is non-optimal and fails to deeply consider the optimization of mill operation combinations, a dynamic process of progressive coal blending for bunkers and mills based on a cascade optimization strategy is proposed. Firstly, to address the suboptimal nature of traditional manual experience-based bunker coal blending, an automated primary coal blending optimization model for bunkers is established, significantly improving blending efficiency and reducing costs. Then, to further tackle the drawback that bunker coal blending does not consider mill operation combination optimization, a secondary optimization coal blending method is proposed. This method selects reasonable mill operation combinations and optimal output points based on an optimization model, thereby optimizing the simple functional relationship between coal feed rate and load within the DCS system. Simultaneously, based on the practical need for mill vibration prevention, a linear fuzzy programming optimization model with flexible constraints and its solution method are established. Finally, experiments are conducted to validate the proposed method. The results demonstrate that the cascade optimization model-based strategy for the dynamic process of bunker-mill progressive coal blending and combustion further optimizes mill combinations and output to adapt to variable loads on the basis of primary coal blending.

For the large amount of band data in hyperspectral remote sensing images and the difficulty in distinguishing and classifying conventional features, a method for hyperspectral image classification based on a semi-supervised subsidiary classifier-generative adversarial network (SSC-GAN) is proposed. Firstly, the discriminator is developed into a semi-supervised multi-classifier, and train the GAN generator in cascade with the discriminator. Then, to expand the data, the discriminator pulls higher-level features from the generated images, and the auxiliary classification labels are coordinated with the generated samples. Finally, a convolutional neural network replaces the discriminator and generator in the GAN, and the dynamic stacking optimization of the two upgraded networks is carried out. The improved algorithm model achieves the best classification accuracy compared to some classical hyperspectral remote sensing image feature extraction and classification approaches, and has more advantages in classification ability and robustness.

In order to solve the problem of difficult feature extraction and low diagnostic accuracy of bearing fault in noisy environment, a bearing fault diagnosis method combining improved golden jackal optimization (GJO) algorithm, variational mode decomposition (VMD) and support vector machine (SVM) is proposed. Firstly, to deal with the problem that GJO algorithm is easy to fall into local optimum, Tent chaotic map, nonlinear decreasing parameter, reverse learning strategy and adaptive weight are introduced to improve it. Secondly, for the problem that the fault signal is easy to be submerged by noise and difficult to extract features, VMD is used to decompose the signal after singular spectrum analysis (SSA) noise reduction to suppress the noise, and then t-distribution-random neighborhood embedding (t-SNE) is used to screen the extracted features and select useful features. Finally, for the problem of SVM parameter setting, the improved GJO optimization is used to solve and construct the diagnosis model. Compared with the diagnostic model optimized by other algorithms under the same experimental conditions, the results show that the diagnostic accuracy of the model is higher than that of the compared model.

The existing research results have problems such as large gap between the scheduling model and the actual problem, insufficient research on the essential characteristics of the problem, and insufficient systematic research on the optimization algorithm, and there is no literature on modeling or optimization for the prototype vehicle testing problem. There is no literature on modeling or optimization of the test problems in the experimental stage, which makes it difficult to effectively serve production management. This paper starts from the production reality, investigates the test scheduling problem in depth, establishes mathematical models, systematically discusses its essential characteristics and optimization methods, proposes a hybrid intelligent scheduling method for test vehicles in the development and testing stage of new models, and carries out simulations and industrial applications. The results show that the method can enrich and deepen the existing optimal scheduling theory and method; and can directly serve the automotive industry and promote its production management level and market competitiveness.

It’s important to optimize full multiple length shearing in bar process line for improving yield and production efficiency. Bar rolled piece total length prediction is the key technology for full multiple length shearing optimization. Traditional methods predict the total length using mechanism models and single factor variables, which results in large error or late prediction. The data-driven prediction model realized early and accurate total length prediction of rolled pieces by employing multiple factor variables and the generalized linear regression algorithm. Firstly, sample length measuring data of each rolling mill stand with high frequency. Then, the rolling length of each mill stand and the total rolling length of the final mill stand are calculated. Finally, train model and rolling predict the total length of the rolled piece. Using high speed bar process data from a certain steel plant for testing, the maximum absolute percentage error of prediction is 0.130 7%. Compared with traditional methods, the prediction accuracy is improved over 6 times, which is an effective technical support for full multiple length shearing optimization.

In order to accurately grasp the real-time healthy operation status of high-speed doubly fed grid connected units and maintain them in time, a fuzzy evaluation model for the healthy operation status of wind turbines based on Game theory combination weighting is proposed. Firstly, the subjective and objective weights of the evaluation indicators are calculated through the analytic hierarchy process and CRITIC method. In order to eliminate the one-sided nature of single weighting, the subjective and objective weights are linearly combined with the idea of Game theory and their proportion is determined. Secondly, establish an evaluation index deterioration membership function based on the ridge distribution function. Finally, the Game theory combined weighting fuzzy evaluation method is used to build an assessment model for the healthy operation status of wind turbines, and the healthy operation status of each layer of wind turbines is evaluated from bottom to top. Taking the high-speed doubly fed grid connected unit of a wind farm in Xinjiang as an example, the application analysis of the evaluation model shows that the model can detect deterioration trends before faults occur and accurately determine the degree of component deterioration.