Research On Intelligent Decision Making Of Membrane Fouling In Wastewater Treatment Process

Membrane bioreactor wastewater treatment technology is a kind of wastewater treatment technology which combines activated sludge process and membrane separation technology.With the development of membrane technology,membrane bioreactors have been widely used in municipal wastewater treatment.However,in the process of membrane bioreactor wastewater treatment,membrane fouling is inevitable,which will increase the operation cost,reduce the life of the membrane bioreactor,or even break down the whole wastewater treatment process,and hinder the smooth operation of the membrane bioreactor.In order to reduce the incidence of membrane fouling and ensure the long-term stable operation of sewage treatment process,it is very important to make accurate decision on membrane fouling.Therefore,the design of an effective decision-making method for membrane fouling has high research significance and application value for alleviating membrane fouling and ensuring effluent water quality.In order to realize the accurate decision of membrane fouling,an intelligent decision method of membrane fouling is proposed.First,a knowledge-based fuzzy broad learning algorithm was proposed to solve the problem of insufficient membrane fouling data.Second,in order to estimate the state of membrane fouling,a recognition method based on fuzzy broad learning was proposed to realize the accurate recognition of membrane fouling.Then,an interval type 2 fuzzy neural network based on data and knowledge was proposed to realize the precise decision of membrane fouling.Finally,in order to realize real-time monitoring and decision-making of membrane fouling,an intelligent decision-making system for membrane fouling was designed to provide operational suggestions for membrane fouling,so as to ensure the stable operation of membrane bioreactor wastewater treatment process.The achievements of this paper are showed as below:1.Research on knowledge-based fuzzy broad learning algorithm.To solve the problem of insufficient data in membrane fouling process,a knowledge-based fuzzy broad learning algorithm was proposed.First,the existing knowledge is extracted and expressed in the form of fuzzy rules.Second,a knowledge-based fuzzy broad learning algorithm is designed,and a hybrid fuzzy neural system is formed by replacing the feature nodes of the broad learning system with a fuzzy subsystem.Finally,pseudoinverse ridge regression parameter optimization algorithm is used to improve the accuracy of knowledge-based fuzzy broad learning algorithm.Experimental results show that the proposed knowledge-based fuzzy broad learning method has good performance.2.Research on identification method of membrane fouling.To solve the problem that it is difficult to realize the accurate identification of multiple indicators in membrane fouling process,a knowledge-based fuzzy broad learning method for membrane fouling identification was proposed.First,the existing knowledge was extracted from the humanistic category of membrane pollution in the form of fuzzy rules to improve the shortcoming of insufficient data.Second,a knowledge-based fuzzy broad learning algorithm is designed,and knowledge modeling and data-driven method are adopted to improve the learning performance of the algorithm.Finally,a knowledge-based fuzzy broad learning method for membrane fouling identification was designed to achieve accurate classification of membrane fouling types.The experimental results show that the proposed knowledge-based fuzzy broad learning method has a good recognition effect.3.Research on decision making method of membrane fouling.To solve the problem of low precision of membrane fouling decision,an interval type 2 fuzzy neural network based on data-knowledge was proposed.First,combining with the existing experience and identification results in the membrane bioreactor wastewater treatment plant,the knowledge base of membrane fouling decision was established in the form of fuzzy rules,which made up the shortcoming of insufficient data.Second,a knowledge reconstruction mechanism is proposed to complete the knowledge reconstruction by balancing the matching accuracy and diversity.Finally,an interval type two fuzzy neural network model based on data-knowledge is proposed,and the parameters of the membership function layer are designed by using fuzzy rules,and the transfer gradient descent algorithm is designed to adjust the parameters of the network model.The experimental results show that the proposed method can realize the accurate decision of membrane fouling.4.Research on intelligent decision system of membrane fouling.In order to realize the practical application of intelligent decision technology for membrane fouling,an intelligent decision system for membrane fouling was designed.First,a data acquisition and transmission platform for membrane fouling was established to realize real-time acquisition and transmission of process variables.Second,through the prediction of key process variables to complete the enhanced prediction of membrane fouling,according to the comprehensive analysis of process variables to realize the identification of membrane fouling,according to the data-knowledge information to provide decision support for the abnormal membrane bioreactor wastewater treatment process.Finally,the intelligent decision system of membrane pollution was applied to a sewage treatment plant in Beijing,and the developed intelligent decision system was tested in practice.The experimental results show that the intelligent decision system can reduce the occurrence of membrane fouling and provide a safe guarantee for the membrane process.