Parameter Online Self-tuning Of Model Free Controller

With the progress of the times and the development of science and technology,the actual production process is becoming more and more complicated,so many complex features such as multi input and multi output,strong coupling,strong nonlinearity,multi working condition,variable load,time variation and so on are increasingly presented,It makes the precise mathematical models of controlled objects are difficult to obtain,and in the meanwhile,It brings unprecedented challenges to traditional control method based on mathematical model.Model-free control algorithm is a new type of data-driven control method,it only relies on the input-output data measured in real time to analyze and design the controller,and it does not rely on any mathematical model information of the controlled system,the impact of unmodeled dynamics is avoided,so this algorithm has a good prospect of application.However,the determination of model-free controller parameters requires someempirical knowledge,and the regulation process is time-consuming,it limits the further application of model-free controller.At present,there are only a few related researches on the field of model-free controller parameters tuning,and there is still no systematic parameter tuning method.In order to promote the practical application of the model free controller in theindustrial process,in this paper,the effect of model-free controller parameters on control effectiveness is analyzed in depth,the penalty factor and step factor of the control input estimation algorithm are selected as the parameters to be tuned,so combining with the strong learning ability of BP neural network,three new methods for on-line self tuning of controller parameters are proposed and implemented,a system theory and method for on-line self-tuning of model free controller parameters are formed:(1)For the SISO system,a novel on-line self-tuning method for SISO model freecontroller parameters was proposed and implemented.The simulation experiment of a typical time-varying nonlinear SISO system shows that the proposed method can significantly improve the control precision and stability.(2)For the MIMO system,a novel on-line self-tuning method for MIMO model free controller parameters was proposed and implemented.The simulation experiment of a typical time-varying nonlinear MIMO system shows that the proposed method can significantly improve the control precision and stability.(3)For the MIMO system,a novel decoupling control method of MIMO system based on SISO model free controller was proposed and implemented.First,according to the coupling characteristics and the tendency characteristics of the MIMO system,it is decomposed into multiple coupled SISO systems,then the online decoupling of multiple SISO systems is implemented synchronously and on-line self-tuning of model-free controller parameters of SISO system.The simulation experiment of a typical time-varying nonlinear MIMO system shows that the proposed method can significantly improve the control precision and stability.