Tracking Control Method For Wastewater Treatment Based On Fuzzy Neural Network

Wastewater treatment process(WWTP)is a highly nonlinear industrial control process which includes many complex biochemical reactions.And thus it has obvious characteristics of numerous variables,large time-varying,large time delay and strong coupling.At the same time,because of the characteristics of the wastewater treatment process,the precise mathematical model is difficult to establish,and the control of the wastewater treatment process faces multiple difficulties and is challenging.Moreover,compared with other countries,the wastewater treatment industry in china started late,and the control technology of WWTP is still relatively backward,which makes the control accuracy of the sewage treatment relatively low,and it is difficult to achieve a higher quality of the effluent.Therefore,according to the characteristics of the wastewater treatment process,the control methods applicable to the water treatment process are studied,and the effective control of the wastewater treatment process is realized.The development of the water treatment industry is of great significance and far-reaching.In this paper,the current research results are carefully summarized and analyzed.Aiming at the precision model of the wastewater treatment is difficult to be established and the problem of low control accuracy,in this paper,a process control method for wastewater treatment based on fuzzy RBF neural network is proposed to achieve an effective wastewater treatment process control.The main work of this paper includes the following contents:1.In this paper,the properties of activated sludge process are deeply analyzed.At the same time,the BSM1(Benchmark Simulation Model No.1)benchmark simulation model,which is jointly proposed by the European Union of scientific and technological cooperation(COST)and the International Water Association(IWA),is studied.The structural characteristics of the biochemical reaction tank and the two sedimentation tank in the BSM1 model are analyzed in detail.Finally,the simulation model is established and verified in the Matlab environment,which proves the correctness and reliability of the model,and provides a simulation platform for the control strategy and the optimization method.2.In the WWTP,the dissolved oxygen concentration is one of the important factors that affect the effect of wastewater treatment.However,the traditional control method has low accuracy and difficulty in controlling dissolved oxygen concentration.Based on this,a fuzzy neural network control method is proposed and designed,which combines fuzzy control with neural network.The simulation results are presented,showing that this control method can better track and control the dissolved oxygen concentration,and has good control accuracy and control performance.3.Because of the characteristics of many variables and serious interference in the process of wastewater treatment,a self-organizing fuzzy neural network controller based on the activation strength and the importance of the neuron is proposed and designed in this paper.The self organizing method takes into account the characteristics of the actual biological neural network and the contribution of the neuron to the network output,and uses a method of deletion based on the importance of neurons and the method of increasing the activation intensity based on the neuron.In the deleting phase,a method based on importance attenuation is adopted.Although this strategy fails to delete the "redundant" neurons in time,it can guarantee the stability of the neural network and reduce the possibility of error deletion of important neurons.At the same time,gradient descent algorithm is used to adjust and learn the network,optimize network parameters and improve network performance.The simulation experiment shows that the self-organizing controller can adjust the structure of the controller flexibly according to the environment change in different environment conditions,and ensure the control effect of the concentration of dissolved oxygen and the concentration of nitrate nitrogen.4.In view of the large amount of data collected in the wastewater treatment process and the complex relationship between the data,the configuration software has a weak data storage and application capacity,thus a database-based wastewater treatment process detection system is designed.The key work is to establish the data communication between the database and the configuration,and provide a certain reference for future data transmission work.