Forecasts Based On Least Squares Support Vector Machine Control Ph Neutralization Process

Continuously Stirred Tank Reactor (CSTR) is one of the core equipment of the polymers production, and it is also the main energy-consuming equipment. It is widely used in the Industrial production, such as, reagents, medicines, food, and synthetic materials. In the actual production of industrial process, it is a nonlinear chemical reactor. Among them, the main process parameters contain concentration, temperature, flow, pressure, pH value and so on. However, for the continuous stirred tank reactor, the quality of their products is mainly affected by the impact of these process parameters. In order to greatly improve product quality and yield, to achieve the premise of reliable, stable, and energy-saving, design online control algorithm to ensure that the materials production process technology curve to meet the ideal requirements in the industrial, that is essential. However, due to the control object CSTR with highly nonlinear, control algorithm design complexity, and led to building the predictive model difficulty.Faced with the above problems, based on strong nonlinear function fitting ability of the least squares support vector machine (LS-SVM), this paper discusses identification approach for the nonlinear system inverse model used of LS-SVM. Combining LS-SVM with the traditional PID inverse system method, proposed predictive control algorithm based on LS-SVM/PID composite control inverse system. This method is applied in the pH neutralization process of CSTR, verified the effectiveness of the algorithm. Among them, the main research work of this paper as follows:1) Using nonlinear function fitting ability of SVM and LS-SVM, identified the inverse model in pH neutralization process of the CSTR, and verified the characteristics of the LS-SVM is a good recognition results and high recognition speed.2) Introduced the concept of inverse system, Series inverse model based on LS-SVM and the original nonlinear system, with the PID control, to component pseudo-linear composite system. The method used in pH neutralization process of CSTR, simulate in the pseudo-linear system. To expect verification the method has good tracking performance and robustness.3) Brief the basic principles and characteristics of predictive control, make pseudo-linear composite system which established in the second and third chapters as a controlled object, combined with the predictive control theory. Raise the predictive control theory used in pseudo-linear composite system, which based on LS-SVM/PID composite inverse system. This method used in simulation validation of the CSTR pH neutralization process. The control effect is better.