Abstract:In order to address the attitude control problem for hypersonic morphing vehicle affected by parameter per-turbation and external disturbances,a data-driven model predictive control(MPC)approach is developed based on the com-posite observer in this paper.To tackle nonlinearities in aircraft modeling and control,this study introduces a data-driven modeling method based on the Koopman operator theory.An autoencoder neural network is employed to approximate the optimal lifting function of the Koopman operator,enabling the extraction of a linear nominal model on finite-dimensional lifting space using extended dynamic mode decomposition(EDMD).Additionally,to mitigate the adverse effects of dis-turbances,a synchronous estimation scheme is proposed to construct a composite observer for hypersonic vehicles,which consists of a Luenberger-type observer based on Koopman lifting model and a synchronous disturbance observer.Sub-sequently,a MPC controller is designed based on the linear nominal model and disturbance estimates.The recursive feasibility and the input-output stability of the controller are proven.Finally,simulation results validate the effectiveness and feasibility of the proposed controller.

Abstract:Under the big data era,the effective discovery of typical process routes can provide more accurate and sufficient process information for the retrieval process of computer aided process planning(CAPP),thus improving the quality of process planning.However,the existing approaches cannot be applied directly and are difficult to quantify the process information because they ignore the structural complexity of the networked process routes.In addition,most of the existing researches have overlooked the clustering effectiveness in the discovery of typical process routes,revealing a gap in effective algorithm design for this challenge.To address these shortcomings,this paper proposes an automated discovery method for typical networked process routes based on the multi-dimensional process information fusion.For the similarity measure,different quantification methods for the four types of process information are designed based on the information requirement analysis.Then,the proposed method integrates these findings into a comprehensive similarity using principal component analysis(PCA).Besides,considering the clustering effectiveness,the fire hawk optimizer(FHO)is introduced into the original affinity propagation(OAP)clustering algorithm to optimize its reference degree and damping coefficient,so that a balance between the clustering results and the soft constraints can be achieved.This way,typical networked process routes that are more in line with the practical requirements can be found.Simulation experiments validate that the proposed similarity measure for networked process routes can effectively distinguish various similarity cases with higher sensitivity.Meanwhile,the introduced FHO can enhance the clustering performance of AP,in which FHO-IAP shows the best clustering effect.

Abstract:Aiming at the safety problems caused by servomotor degradation in 3D stage wire system,considering the ex-isting degradation models with low accuracy,conservative definition of traditional life and low efficiency of life extension,this paper proposes a system-level remaining useful life prediction method based on the composite multi-stage degradation modeling,and carries out the research of adaptive variable sampling rate autonomous maintenance strategy based on the prediction results.Firstly,starting from practical engineering considerations,a three-level autonomous maintenance frame-work is constructed,which includes the basic control layer,the health assessment layer and the autonomous maintenance layer;secondly,considering the variability of the degradation stages and the influence of random shocks,a composite multi-stage degradation model of Wiener+Poisson is established,and an expectation-maximization parameter estimation algorithm is adopted,which integrates the parameter estimation process with the cumulative sum and change-point detec-tion algorithms,and adopts a more appropriate evaluation of the last exit time definition of the control system life is used to obtain the analytical solution of the remaining useful life of the system under the composite multi-stage degradation model;Finally,based on the prediction results,a DMC-PID life extension control strategy based on adaptive variable sampling rate is proposed to improve the maintenance efficiency and effectiveness.Simulation experiments verify the effectiveness of the proposed method.

Abstract:Studying the dynamic flexible job-shop scheduling problem(DFJSP)is of great significance for effectively controlling production processes and enhancing enterprises' economic profitability.Existing studies mostly only consider machine selection flexibility and dynamic production scenarios;however,a more complex issue of flexible process sequence exists in practical production.Specifically,in scenarios like customized equipment manufacturing,job processing opera-tions can be divided into several groups according to process requirements.There are no mandatory processing sequence constraints apply to operations within the same group,whereas strict processing sequence constraints must be followed between different groups.The introduction of this flexible process sequence significantly increases the complexity of the problem's search space,leading to a substantial extension of solution time.Therefore,by integrating the practical produc-tion needs of customized equipment manufacturing workshops,this paper comprehensively considers machine selection flexibility,flexible process sequence,and the dynamic characteristic of new job arrivals,and proposes the dynamic multi-flexible job-shop scheduling problem(DMFJSP)with the optimization objective of minimizing the average flow time.To solve this problem,a genetic programming(GP)method based on individual simplification is proposed which analyzes the structural complexity of individuals by introducing dynamic terminal node frequency,and sets a penalty function during the fitness evaluation stage to guide population evolution—ultimately simplifying individual structures and reducing GP training time.To verify the effectiveness of the proposed method,tests are conducted under three different production sce-narios.The results demonstrate that the method can significantly shorten computation time while ensuring solution quality;when addressing larger-scale and more complex problems,it exhibits fast convergence and can obtain satisfactory solutions within a reasonable time frame.

Abstract:This article investigates the robust current tracking control problem of three-phase grid-connected inverters with LCL filter under external disturbance by a dynamic state feedback control method.First,this paper constructs an internal model to learn the information of the states and input of the grid-connected inverter under steady state.Second,by utilizing the internal model principle,the paper turns the tracking control problem into the robust stabilization control problem based on some appropriate coordinate transformations.Then,The paper designs a dynamics state feedback control law to deal with this robust stabilization problem,and thus the solution of the robust current tracking control problem of three-phase grid-connected inverters can be obtained.This control method can ensure the asymptotic stability of the closed-loop system.Finally,the paper illustrates the effectiveness of the proposed control approach through several groups of simulations,and compares it with the feedforward control method to verify the robustness of the proposed control method to uncertain parameters.

Abstract:In order to realize a high-precision damping force tracking of magnetorheological damper(MRD),a discrete-time linear quadratic tracking(DLQT)control strategy based on Koopman operators is proposed.Aiming at the hysteresis nonlinearity of MRD,a nonlinear recurrent neural network(RNN)model of MRD is established.Koopman operators the-ory and extended dynamic mode decomposition(EDMD)algorithm are used to obtain a high-dimensional model of MRD.A discrete-time linear quadratic tracking controller is designed by using the high-dimensional linear model.The expected signals of different frequencies are tracked through simulation experiments,which verifies the effectiveness of the pro-posed scheme.Furthermore,physics experiments are conducted on a 2-degree-of-freedom quarter suspension experimental system equipped with a MRD,which show that the strategy can achieve high-precision tracking of signals.

Abstract:This paper studies the prescribed performance consensus tracking control problem for nonlinear multi-agent systems(MASs).Unlike most existing nonlinear MAS models,this approach considers unknown external bounded dis-turbances affecting individual agent states,with the MAS states being unmeasurable directly and the nonlinear functions completely unknown.An error transformation function is introduced to convert the nonlinear MAS with predefined tracking error constraints into an unconstrained nonlinear system exhibiting desired performance characteristics.A novel adaptive weighting radial basis function neural network(AW-RBFNN)system is proposed to address unknown nonlinear functions in the MAS model.Additionally,a state observer is employed to estimate unmeasurable state variables,and a control law is designed based on the AW-RBFNN system and state observer.Through Lyapunov stability theory and prescribed per-formance stability analysis,it is demonstrated that the consensus tracking error converges to a predefined region while all closed-loop signals remain uniformly ultimately bounded,enabling the nonlinear MAS to achieve prescribed-performance-satisfying tracking control.The effectiveness of the prescribed performance collaborative tracking control method based on AW-RBFNN is verified by comparing with the multi-dimensional Taylor net(MTN)based method through an numerical simulation example and tracking control examples modeled of nonlinear agents as actual mechanical systems.

Abstract:In the traditional current-voltage double closed-loop strategy,the sliding mode controller has a strong depen-dence on the system model parameters,which leads to problems such as reduced robustness and sluggish dynamic response in the current inner loop controller of the active power filter.To address this,this paper proposes a double closed-loop sliding mode control strategy based on radial basis function(RBF)neural networks to improve the dynamic response speed and robustness of the compensation current.The inner loop of this control strategy adopts a RBF neural network global fast terminal sliding mode controller,while the outer loop uses a linear sliding mode controller.The RBF neural network reduces the dependence on the model by online approximation of unknown terms,and the global fast terminal sliding mode controller is used to enhance the system's convergence.Experimental results show that the proposed control strat-egy enables the single-phase three-level active power filter to exhibit superior current tracking performance and stronger robustness under both steady-state and dynamic operating conditions.

Abstract:To overcome the strong time-dependency of conventional intermittent communication mechanism and achieve the rapid convergence requirement in multi-agent systems,this paper proposes a finite-time region-division in-termittent consensus approach.By introducing two boundary functions with finite-time convergence characteristics,the non-negative real domain is partitioned into three sub-regions:Working,resting,and buffer zones.The communication activation among agents is governed by the relationship between the Lyapunov function trajectory and preset sub-regions.A distributed region-division intermittent communication protocol with finite-time convergence is designed,along with an estimation framework for dwell time.Numerical simulations validate the effectiveness of the theoretical results.

Abstract:This paper studies cooperative robust parallel operation of multiple actuators over an undirected communi-cation graph.The plant is modeled as an uncertain linear system,and the actuators are linear and identical.Based on the internal model principle,a distributed dynamic output feedback control law is proposed to achieve both robust output regulation of the closed-loop system and plant input sharing among the actuators.A practical example of five motors co-operatively driving an uncertain shaft under an external load torque is presented to show the effectiveness of the proposed control law.

Abstract:Permanent magnet synchronous motors(PMSM)are widely used in industrial applications due to their high efficiency and favorable dynamic characteristics.However,its control performance is often limited by modeling uncer-tainties,stochastic disturbances,and input saturation.To address these challenges,this paper proposes an adaptive control strategy based on radial basis function neural network(RBFNN).A stochastic PMSM model incorporating modeling errors and stochastic disturbances is constructed,while input saturation is handled through a saturation function.The RBFNN is employed to approximate unknown nonlinearities online,and adaptive laws are designed for parameter adjustment.A non-recursive tracking differentiator is introduced to avoid the"complexity explosion"problem in conventional backstepping,and a compensation mechanism is further developed to mitigate filtering and saturation errors.Based on Lyapunov stability theory for stochastic systems,it is rigorously proven that all system errors are probabilistically uniformly ultimately bound-ed.Numerical simulations and semi-physical experiments on dSPACE platform validate the effectiveness of the proposed control strategy,demonstrating robust performance under input constraints.

Abstract:Considering the limited communication resources in the state monitoring process of renewable energy mi-crogrid system,a distributed blocking moving horizon estimation algorithm based on Round-Robin protocol is proposed.Building upon distributed moving horizon estimation,the Round-Robin protocol is introduced to reduce the communication burden of the sensor network.Under this protocol,each sensor node transmits its measurement component successively and periodically,avoiding data congestion and fully utilizing the limited bandwidth resources.Moreover,the disturbance sequence in the estimation window is designed using the block concept,which reduces the number of optimization variables and the amount of online computation.By analyzing the feasibility and convergence of the algorithm under the maximum block length,the sufficient conditions are established to guarantee the existence of an equivalent solution to the optimization problem of the algorithm,and the results are extended to the case of arbitrary block of the disturbance sequence.Simulation results show that the proposed algorithm can estimate the states of renewable energy microgrid system effectively.

Abstract:To address the issue of potential spoofing attacks on GPS signals in connected vehicles,this paper proposes an attack detection strategy based on an adaptive distributed Kalman filtering(DKF).Firstly,an uncertain dynamical model of a connected vehicle platooning system subject to the GPS spoofing attack is derived,and then an adaptive DKF algorithm is proposed to solve the problem of unknown noise statistical properties of the connected vehicles in driving.Secondly,a chi-square detection method based on the state residuals of the DKF is designed to detect GPS spoofing attacks.Finally,the effectiveness of the proposed detection method is verified through experimental simulations.

Abstract:PID controller is widely used in the field of industrial control,the selection of its parameters is over-dependent on manual experience,the efficiency is low and the process is complicated.In recent years,deep reinforcement learning has been successfully applied in many fields because of its ability to self-learn from complex environments.In this paper,a PID parameter optimization method based on twin delayed deep deterministic policy gradient(TD3)algorithm of double replay buffer is proposed,and the parameters of PID controller are optimized by deep reinforcement learning.In the whole optimization process,the control problem is regarded as a sequence decision process.The optimization process of PID parameters is transformed into the updating process of the weights of the agent's network by designing the state space,action space and the network structure of the agent.At the same time,to solve the problem of low exploration efficiency in the early stage of TD3 algorithm training,the double experience replay buffer mechanism is added on the basis of TD3 algorithm to improve the efficiency of the early stage of algorithm training.Finally,simulations are performed on the second-order system and first order plus delay time system,and compared with the PID parameter optimization method based on particle swarm optimization(PSO)algorithm.The experimental results show that the PID parameters optimized by the proposed algorithm have better control performance than the PSO algorithm.

Abstract:This paper investigates formation control problem of higher-order multi-agent systems with both unknown states and mismatched disturbances.Using disturbances estimation(compensation)method and sliding-mode control,out-put feedback based active anti-disturbance formation control algorithms are proposed.When the agents states are known,generalized proportional integral observer is designed for each agent.Combining disturbances compensation with sliding mode control,state feedback based active anti-disturbance formation controllers are developed.The developed distributed controllers guarantee that the multi-agent systems achieve formation asymptotically.When the agents states are unknown,both the unknown states and the disturbances are estimated by constructing extended state observer.Combined with state feedback formation control design,output feedback based active anti-disturbance formation control algorithms are given.Under the output feedback based distributed control algorithms,the multi-agent systems realize formation asymptotically.Simulation results validate the effectiveness of the proposed formation control methods.

Abstract:A spatial adaptive fully-saturated repetitive learning control method is proposed for rotating motor systems that perform spatial repetitive tasks.The spatial differential operator is introduced to transform the controlled system from the time domain to the spatial domain.By utilizing the spatial periodic repetitive operation characteristics of rotating motor systems,a spatial adaptive fully-saturated repetitive learning controller is designed to achieve high-precision tracking of desired trajectory for the angular velocity of rotating motor.The fully-saturated repetitive learning law is constructed to estimate and compensate for system non-parametric uncertainties with spatial periodic characteristics,and the estimated value can be limited within the specified bounds.Finally,the error convergence is analyzed through the Lyapunov stability theory,and simulation results are provided to verify the effectiveness of the proposed method.

Abstract:This paper is concerned with the problem of delay-dependent frequency stability of isolated island micro-grid systems.Firstly,a virtual inertial control component is put in place to compensate for the insufficient inertia of the system caused by the increase in renewable energy.Then,a novel mathematical model for delay-dependent load fre-quency control of isolated island microgrids is constructed by using an advanced model reconstruction technique.Based on this,a Lyapunov-Krasovskii functional is built that includes extensive time delay information and fully incorporates model reconstruction technique,and a generalized free-matrix-based integral inequality and a linearization condition of quadratic function inequality are used to handle the functional's derivatives.Thus,a delay-dependent stability criterion in linear matrix inequalities is obtained.Finally,the effectiveness of the proposed method is demonstrated through simula-tion experiments and data analysis,which confirms its lower conservatism and the significant improvement in the dynamic performance of the microgrid systems due to the introduction of the virtual inertial control component.

Abstract:This paper proposes a prescribed-time controller for a class of nonlinear systems with asymmetric output constraints and time-varying parameters.Compared with the traditional constraint control in which the designer cannot preset the convergence time and the assumption that the initial value satisfies the constraints,the proposed controller can not only regulate the system state to zero within a prescribed-time while satisfying the constraints,but also prevent the potential singularity issue of controller when the system initial value violates the constraints.Firstly,the construction of virtual constraint boundaries eliminates the requirement for initial states to satisfy constraint conditions,while guarantee-ing that states violating constraints can recover to constraint satisfaction within a prescribed-time.Secondly,an equivalent transformation method is designed to achieve synchronous convergence of system states and transformation function under asymmetric constraints.Unlike existing prescribed-time control approaches that require uncertain nonlinearities to meet linear growth conditions or unknown constant parameters,the studied nonlinear system in this paper incorporates uncer-tain time-varying parameters.By employing backstepping techniques based on the congelation of variables method and prescribed-time regulation function,a prescribed-time controller capable of handling the asymmetric output constraints and time-varying parameters is designed.Finally,the effectiveness of the proposed method is demonstrated through two numerical simulations.

Abstract:For nonlinear multi-agent systems with unknown disturbances,a cooperative learning control algorithm consisted of a memory-based event-triggered mechanism and privacy preservation is proposed.Firstly,to prevent the leakage of the key state and control information in the interaction between intelligent agents,a time masking function is constructed to provide the privacy protection for the entire system,thereby enhancing encryption capabilities while reducing the impact on system performance.Then,the memory-based event-triggered mechanism is applied within a backstepping framework,and the secure control objectives are achieved by leveraging the mechanism's ability to store historical data.Meanwhile,historical data is used to set threshold conditions,avoiding false signals from affecting communication and ensuring the system's transient performance.Lastly,the proposed error-based cooperative learning adaptive rate does not require prior assumptions about the finite nature of neighbor weights,thereby reducing conservatism.The event-triggered control scheme is verified through theoretical analysis and simulation results.

Abstract:Underactuated systems have been widely studied and applied due to their unique performance advantages,but their complex dynamic characteristics can lead to challenges in control.Constrained-following control method can effectively address the control problems of underactuated systems.To reduce communication costs and enhance control performance,a novel event-triggered mechanism is proposed for this method.Firstly,the desired trajectories of the system are designed as servo constraints,and the analytical solution of servo control input is derived using constraint-following control theory.Secondly,an event-triggered mechanism with adaptive gain is introduced:By comparing the state deviation compensation control input with the deviation of the real-time/target control inputs,the mechanism determines permissible time instants for control signal updates.Subsequently,adaptive gain is employed to select optimal moments for triggering control signal updates.This mechanism dynamically adjusts the update frequency through adaptive gain to optimize com-munication efficiency.The stability of this control strategy is analyzed using the Lyapunov method,and it is demonstrated that the proposed approach effectively avoids Zeno phenomenon.Finally,simulations on an underactuated planar vertical take-off and landing(PVTOL)aircraft verify the effectiveness of the proposed method.Results indicate that the event-triggered mechanism not only achieves a balance between communication load and system performance but also exhibits strong adaptability to disturbances.