Predictive Control Of DO Concentration Based On FA-BP Neural Network With HDE Algorithm

At present,the urban sewage treatment has become urgent problem to be solved and the method of activated sludge process is the most widely used.As sewage treatment system has several characteristics,like nonlinearity,hysteresis,parametric complexity,coupling between parameters,etc,lead to that modeling and control of sewage treatment process became more difficult.Nevertheless,the level of modeling and control of sewage treatment process at home is low.Thus,the research on modeling and control of sewage treatment process is of great importance.As dissolved oxygen is one of the key parameter index,the more in-depth control and study is of particularly important.In this paper,dissolved oxygen concentration as a controlled variable used in following studies.Foremost,sewage treatment process is analyzed.On the basis of activated sludge model 1,describes internal connection between reaction processes and concentrations of each component.According to the material balance relation and characteristics of sewage treatment system,builds a simplified mathematical model for control.Aiming at the problems of slow convergence speed and easy to fall into local optimum when traditional BP neural network is used as prediction model.In this paper,Firefly Algorithm is used to optimized selection of initial weights of neural networks.Optimized neural networks have higher accuracy than before in model identification,which is proved by simulation.Next,aiming at the concentration of dissolved oxygen,put forward the predictive control of neural networks based on HDE hybrid optimization algorithm.The method combines the attraction mechanism in Firefly algorithm and the mutation,crossover and selection mechanism in differential evolution algorithms,which can effectively solve the problem that the optimal objective performance function is difficult to solve in rolling optimization of predictive control.Standard function testing suggests that the fusion algorithm has better optimization performance.Ultimate,using FA-BP neural networks as predict control model of dissolved oxygen,introducing HDE algorithm into rolling optimization link,prove that the method used in this paper has higher accuracy,better tracking performance and stronger jamming ability.