| The performance of the controller determines whether the industrial equipment can low consumption,efficient and safe operation.In modern control engineering,there are many methods to improve the performance of the controller,including conventional PID controller tuning techniques,and advanced control methods such as optimal control,predictive control.The implementation of such methods generally depends on an accurate process object model.However,it is very difficult to describe the dynamic characteristics of process objects and establish accurate models because of many variables,complex internal mechanisms and coupling.Therefore,there are challenges in designing effective controllers and maintaining good performance over time.At the same time,thanks to the rapid development of computer and measurement technology,data generated in industrial processes can be easily obtained and stored.These data are usually contains the information related to production process and state of the controller.The methods of using them to directly design the controller or tune the controller parameters and improve the control performance of the system have received extensive attention.Virtual reference Feedback tuning(VRFT)is a "one-time" data-driven control performance optimization method.The paper has carried out in-depth research and expansion from the aspects of controller parameter optimization precision,online performance optimization and multivariable system control optimization based on the traditional research of VRFT.The specific contents include:(1)Aiming at the decline of control performance of online system,a VRFT control optimization method based on the minimum variance(MV)index is proposed.Firstly,the principle of MV performance assessment is described in detail,including the design of MV control rate,the establishment of time series model and the calculation of MV performance index.Secondly,the online performance optimization method is designed,including MV performance assessment module and VRFT optimization module.The expected threshold of the system is specified.If the MV index value of the actual system is lower than the expected threshold,the system control performance is improved based on VRFT.Otherwise,there is no need to optimize the controller.Finally,the performance monitoring and optimization of the actual water tank level control system is realized based on the method.(2)Aiming at the problem of inaccurate controller parameter tuning due to mismatch of VRFT reference model,a VRFT control optimization method based on IAE benchmark is proposed.Firstly,by analyzing the relationship between VRFT and IMC design process,a single variable VRFT reference model structure is designed.Secondly,the pending parameters of the reference model are determined based on IAE benchmark,and then the optimal VRFTPID control performance is designed.Finally,based on this method,the performance of the temperature control system in the beer fermentation process is optimized,and based on the dynamic performance index,the scheme is compared with the IMC-PID method to verify the superiority of the scheme.(3)Aiming at the problem that the controller parameters of multivariable control system are difficult to tuned,a data-based multi-variable VRFT control performance optimization method is proposed in the paper.Firstly,the principle of multi-variable VRFT is analyzed,including the definition of the control target and the setting of the controller parameters,and the multi-variable VRFT filter is designed by using Parseval to improve the setting accuracy of system controller parameters.Secondly,the multivariable system control performance assessment index is designed by using the square error integral to monitor the system operation state.Finally,the proposed method was used to control the Wood-Berry system,and based on the dynamic performance index,the scheme is compared with a variety of multivariable system control strategies to verify the superiority of the scheme. |
