Constrained Generalized Predictive Control Based On Genetic Algorithm And Nonlinear Programming Algorithm

The Prediction control is a kind of new computerized control algorithm developed in the1970s. Over the past30years, predictive control theory and practice development has achieved fruitful results, and create a unique roll optimizing the uncertainty of the system and stable robust design theory system. Generalized Predictive Control (GPC) based on adaptive control and predictive control research developed control methods. The algorithm is a advanced control algorithm, which is robust system that can effectively, overcome the lag and be applied to the open-loop unstable non minimum phase system. However, the traditional GPC algorithm does not consider the restrictions of the industrial process of with control action. If traditional generalized predictive control has constraints, it is bound to be difficult about control quantity of the solution, and affect the performance of the generalized predictive control. Usually, constrained generalized predictive controller of the solution is a optimization problem, it is difficult to solve, and huge computation. Therefore, other ways of optimizing is applied to solve problems of constrained generalized predictive control,The Genetic Algorithm based on a random search and mimics the process of natural evolution and genetics and is optimized solution of dealing with the complex optimization problems. The Genetic Algorithm is not restricted by factors such as nature of the problems or optimization guidelines form, its objective function fitness guided by fitness and probability, it is effective to deal with optimization problems with constraints. Nonlinear programming is an important branch of optimization theory and method,mainly studies extreme value problems and constrained extreme-value problems of theory and algorithms. But the GA also has its limitations such as premature convergence, weak local search ability. And the population size impact on genetic algorithm Optimal Performance of genetic algorithm. Small population size may give rise to a single species, group evolution process end quickly, and when population size is large values, it would result in burden some calculation and affect the computational efficiency.In this thesis the constrained generalized predictive control based on genetic algorithm is proposed, but this algorithms application may cause problems. Genetic algorithms have its own limitations, therefore, genetic algorithm combined with nonlinear programming algorithm make up for a lack of genetic algorithm, Using the strong search ability of genetic algorithm and the stronger local search ability of nonlinear programming algorithm, combining the two algorithm’s advantage are used in circular optimization process of GPC, and the optimal control rule is obtained. Finally, simulation results validate the applicability and validity of combinatorial algorithm.under constraint generalized predictive control, improves the accuracy and speed for optimizing and can get good performance.Finally, The constrained generalized predictive control based on genetic algorithm and nonlinear programming algorithm is put forward to implement mould level control with constraints, the simulation results demonstrate the effectiveness of the algorithm.