The Research Of Nonlinear LTR Control Method For The Continuous Stirred Tank Reactors

At present, the continuous stirred tank reactors have been widely used in chemical industry production, which are the important equipment for a variety of complex chemical reaction. Its operation situation will directly affect the product quality and production. Since CSTR system is of high nonlinearity and the existed control methods are of poor robustness and real-time ability. Therefore, the research of control methods for CSTR has important practical significance. In a continuous stirred tank reactor, reaction temperature, one of the most important parameters of temperature in the reaction kettle will directly affect the product quality and quantity. If the excess heat of reaction is not timely removed, the safety of reaction maybe affected. Therefore, the control of reactor temperature is of great significance. Another important parameter is the reaction concentration, it directly shows the product quality in the reactor, so the real-time monitoring of reaction concentration is a very important task in the practical application. However, due to the limitation of actual conditions and measure conditions, the real-time detection of concentration is of high complexity, the cost is also expensive. To solve this problem, the system state observer is used to analyze the reaction concentration in real-time. This method has extremely important theory significance and practical value.With a large number of related literature about CSTR system state observer design and temperature control consulted, firstly the structure and working principle of CSTR is elaborated, the nonlinear mechanism model of CSTR and its derivation is given. The simulation experiment of its stability also has carried out.Secondly, since CSTR is of high nonlinearity and Kalman filter is used only in linear systems, so in this paper, the extended Kalman filter and scaled unscented Kalman filter are used respectively to estimate the concentration variable of nonlinear CSTR system. EKF use the Jacobian matrixes to linearize the system, UKF use the scaled symmetric sampling strategy to obtain the sigma points, the specific processes of the two algorithms are given, and the advantages and disadvantages of the two observers are analyzed. The simulation results verify the effectiveness of the two observers, and the accuracy and stability of two filters are also compared.Finally, the linear LTR method is improved, to apply it to the control of nonlinear systems. First of all, the method of variable gain LQR is used to the control of nonlinear systems, and the relevant certificates and simulation experiments are given; Then variable gain LTR control method on the basis of variable gain LQR is proposed for nonlinear CSTR systems, with extended Kalman filter as system state observer, the controller gain is obtained through LQG method, the integrator is introduced to ensure that the system output can track the set point, then the loop transfer recovery method is used to recovery the performance of the closed-loop system, then the design of the NLTR controller is completed; Simulation results verify the effectiveness, rapidity and robustness of the proposed controller. Through the comparison with terminal sliding mode control method, the superiority of the proposed method in this paper is proved.At last, a UKF based variable gain NLTR controller is proposed because that the UKF has higher accuracy and stability than EKF. The influence of weighting matrix coefficients is analyzed. Through the comparison with EKF based variable gain NLTR control method, the improvement of the proposed method is proved. The performance index sensitivity function also proved that the proposed method good recovery performance.