Model-Free Controller Design For Chemical Process

It is not a trivial task to model most chemical processes and design the corresponding model-based control systems since the processes are multivariable,time-delay,and nonlinear in nature.At the same time,many process data are routinely measured and automatically recorded during chemical operation,while the amount of online information is usually rather low.To overcome the above problems,the model-free controller design methods for chemical process are studied in this thesis,in which the Just-in-time Learning?JITL?and Enhanced Virtual Reference Feedback Tuning?EVRFT?methods are directly used in adaptive Proportional-Integral-Derivative?PID?controller design based on the process data,without resorting to the identification of the process model.The main research work of this thesis are summarized as follows:?1?For the single-input and single-output?SISO?nonlinear chemical process,a new data-based model free adaptive PID control system design method is proposed by using JITL.Firstly,the reference database is built by using the open-loop data and closed-loop reference model.Then,using the adaptive nature and good predictive capability of JITL,the PID control system parameters are obtained from JITL by choosing the relevant dataset from the reference database without process model.As different closed-loop reference models will generate different reference databases and then affect the performance of the control system,the first-order virtual reference model T₁ and second virtual reference model T₂ are used respectively and the corresponding controller performances are compared.Furthermore,the PID controller parameters are determined by the different relevant data selection criterion of different JITL,therefore,the proposed model-free adaptive PID controller performances using two JITL method?E-JITL,Co-JITL?are compared.The numerical simulation results of two examples about SISO nonlinear chemical processes are shown in the below:The proposed model-free adaptive PID control system using JITL?including E-JITL and Co-JITL?have strong robutness,disturbance rejection capability and stability,and give better performances than those obtained by the VRFT-based PID.For the proposed model-free adaptive PID control system using JITL,the PID control system is designed with second-order virtual reference model T₂ gives better performances than those obtained by the first-order virtual reference model T₁.Both the two proposed PID control systems using E-JITL and Co-JITL have achieved good control performances,but the proposed PID control system using E-JITL gives a little better performances than that using Co-JITL.?2?For the multi-input and multi-output?MIMO?nonlinear chemical process,using decoupled control system,a new model-free adaptive PID control system design method is proposed by using JITL technique directly.The proposed method is similar to the above PID control system design for SISO process system.The numerical simulation results of the two MIMO nonlinear chemical processes examples are demostrated as follows:The proposed model-free adaptive PID control system using Co-JITL for multivariable process system have strong robutness,disturbance rejection capability and stability,and also gives better performances than those obtained by the VRFT-based PID control system,while the performances of the proposed PID control system using E-JITL for multivariable process system is bad.Therefore,it is conclude that the proposed model-free adaptive PID control system using Co-JITL is better than that using E-JITL.?3?For the MIMO nonlinear chemical process,using the decouple control system,a model-free adaptive PID control system design method for multivariable process system is proposed based on the EVRFT framework from process data.In the proposed method,a second-order virtual reference model is employed.Furthermore,to enhanced the algorithm performance of proposed method,the parameter in the virtual reference model is also updated as process dynamics change.The numerical simulation results of two MIMO nonlinear chemical processes example are illustrated in the following:EVRFT-based model-free adaptive PID control system for multivariable process system have some robutness,disturbance rejection capability and stability,and also gives better responses than those obtained by the VRFT-based PID control system for multivariable process system,but its performances are not better than those obtained by the proposed model free adaptive PID control system using Co-JITL for multivariable process system.