Mathematic Model and Optimal Control Method Based on Hybrid Intelligent System

A mathematic model based on RBF neural network and genetic algorithm for multivariable optimal control with the lowest operational cost by limiting total substrate discharge in activated sludge process are discussed in this paper. It shows that the RBF neural network has preferable convergence for modeling the process. Genetic algorithm is an effective searching method to resolve the optimal problem in this case. Based on satisfying the requirements of precision, binary coding is used to express units, and 20 bits of binary digits express DO, Q_w separately. According to the adaptive degree of units, which can be operated genetically through genetic operator, superior units can be saved, inferior ones are eliminated, and a group of new units can be obtained. The optimization strategy made up of RBF neural network and genetic algorithms is adopted. After achieving the discharge standard of biochemical oxygen demand, the control rule for variables to make operation cost be least is found.