Multi-step Predictive Control Research Of Networked System Via Piecewise Lyapunov Functions

Predictive control which is an effective and widely used strategy in the field of process industry has become a recognized method for dealing with complex constrained and multi-variable problems.However,the conventional predictive control algorithm applied in the process industry cannot deal with the problems of packet loss and quantization errors when the sensor signals and the controller signals are transmitted in the network environment.The industrial process using the conventional predictive control algorithm is likely to adversely affect the control quality of the system.Due to the good stability and anti-interference ability of multi-step control,the predicted value can be approximated more accurately,therefore,multi-step predictive control is introduced to overcome a series of bad effects of single-step predictive control and achieve long-term optimal control.In this thesis,the networked control system is taken as the research object and the multi-step predictive control is taken as the basic scheme,the goal is to solve the typical problems of quantization errors and packet loss in network control systems,and systematically study the multi-step predictive control method under network environment based on Lyapunov stability theory and the segment Lyapunov method.The specific research contents of this thesis include the following aspects:1.The networked predictive control with packet loss and quantization errors is studied based on segmented Lyapunov function fuzzy model.The Markov process is used to model the process of packet loss,and the sector bound method is used to describe quantization errors.Combining with the T-S fuzzy model and the extended segmentation Lyapunov function,less conservative and more stable predictive control method can be obtained.Using the linear matrix inequality(LMI)tool,the optimization problem of the minimum-maximum model is transformed into the solution minimization performance index function in the infinite time horizon.Taking the DC servo control system in industry as an example,the effectiveness of the proposed method is verified.2.The networked multi-step predictive control of T-S fuzzy model with packet loss and quantization errors is studied.Packet loss is modeled by using a Markov process.In order to overcome a series of bad effects in single-step prediction andachieve long-term optimal control,multi-step predictive control is adapted.According to the multi-step predictive model theory,the computational cost of the model is a geometric multiple of the single-step prediction model,so the predictive model is more conservative.The closed-loop multi-step control strategy is applied to enlarge the feasible region of the controller and improve the system control performance,.The effectiveness of the algorithm which can guarantee the asymptotic stability of the closed-loop uncertain network control system can be verified through the DC servo control system in the industry.3.The multi-step predictive control of T-S fuzzy model based on piecewise Lyapunov function is studied.Firstly,the data quantization process in the system is described by using the quantizer,and the mathematical model of the linear time-varying network control system in the data quantification and packet loss network environment is established.Then,combining with the T-S fuzzy model and introducing the piecewise Lyapunov function.the stability conditions can be obtained.Using the multi-step prediction model to obtain better control performance and a larger feasible region of system.The requirements for constrained multi-step predictive control guarantee stability are met.Finally,the system simulation verifies the effectiveness of the predictive control algorithm.