The Study Of Intelligent Early Warning Method And Application For Membrane Fouling Based On Data And Knowledge

Membrane bioreactor(MBR)is a new kind of wastewater treatment process that combines membrane separation technology and biological reaction principle.It has many advantages such as good solid-liquid separation,low sludge yield,good effluent quality,small footprint and so on.It has been used in the process of urban wastewater treatment successfully.However,membrane fouling is a bottleneck problem of MBR wastewater treatment process,which will reduce the quality of effluent water,increase the energy consumption of production,even lead to the collapse of wastewater treatment process,and seriously restrict the promotion and application of MBR.Therefore,how to achieve accurate identification of membrane fouling for reducing the incidence of membrane fouling has become a key issue for researchers at home and abroad,which has the high theoretical and practical application values.For the problem of safe operation of MBR wastewater treatment process,an intelligent early warning method based on data and knowledge is proposed.Firstly,through analyzing deeply the membrane fouling mechanism,the membrane pollution prediction model based on self-organizing deep belief network(SDBN)is constructed to realize the accurate prediction of membrane water permeability.Secondly,through analyzing the fault variables affecting membrane pollution,an intelligent decision-making model based on data and knowledge is established to achieve effective identification of membrane fouling processes and provide decision support for operators.Then,based on the the predicted value of water permeability and the intelligent decision-making results,an early warning evaluation model was established to achieve the accurate identification of membrane fouling status and factors.Finally,the intelligent early warning system for membrane fouling is developed and applied to the real wastewater treatment plant to realize online intelligent warning of membrane fouling and reduce the incidence of membrane fouling.The achievements of this paper are showed as below:1.The study of the soft-computing method for water permeability.A soft-computing method of water permeability based on SDBN is proposed to solve the the problem that the water permeability is difficult to detect online.First,a soft-computing model of water permeability based on SDBN is established to achieve online prediction of water permeability.Second,a self-organizing mechanism based on information relevance strategy is designed to achieve self-organization of this prediction model,which improves the adaptive ability of the soft-computing model.Finally,the accuracy of the soft-computing model is improved by using the unsupervised and supervised parameter optimization algorithm.The experimental results show that the soft-computing model based on SDBN has better prediction accuracy.2.The study of the intelligent decision-making method for membrane fouling.In order to solve the problem that it is difficult to identify the membrane fouling process effectively,an intelligent decision-making method based on data and knowledge is developed.First,the input and output information of this decision-making model are determined by mining the correlation between membrane fouling factors and decision-making suggestions.Second,the intelligent decision-making model of membrane pollution based on SDBN and Softmax classifier is built to identify the degree of membrane fouling degree of membrane fouling.Finally,the self-organizing mechanism is used to dynamically optimize the model structure and parameters,which improves the decision making accuracy.The experimental results show that the intelligent decision-making method based on data and knowledge can effectively identify the membrane fouling process.3.The study of the intelligent early warning method for membrane fouling.In order to solve the the proplem that MBR wastewater treatment process is difficult to operate stably,an early warning method using principal component analysis-independent component analysis(PCA-ICA)method is proposed.First,an early warning evaluation model of membrane fouling is established based on the predictive value of water permeability and the the intelligent decision-making results for membrane fouling.Second,the contributions of process variables to the early warning evaluation model are analyzed in MBR wastewater treatment to determin the main factors of membrane fouling.Finally,using the operational data of the main factors for membrane fouling,an early warning model of MBR wastewater treatment process is designed to achieve effective warning of membrane fouling.Experimental results show that the early warning method for membrane fouling based on PCA-ICA method can identify membrane fouling and diagnose its pollution factors accurately to reduce the false alarm rate and missing alarm rate of membrane fouling.4.The development of the intelligent early warning system for membrane fouling.In order to realize the practical application of the intelligent early warning technology for membrane pollution,this paper designed an intelligent early warning system,which integrates the intelligent prediction module,the intelligent decision-making module and the intelligent early warning module for membrane fouling.The system has many advantages,including process variables monitoring,water permeability prediction,membrane fouling warning and so on.The design of the intelligent early warning system for membrane fouling is applied to real wastewater treatment plants in Beijing.Experimental results demonstrate that the early warning system can effectively improve the operating efficiency of MBR wastewater treatment.