Research On Mbr Membrane Fouling Prediction Based On Improved Sparrow Search Algorithm

Membrane Bioreactor(MBR),as a new type of wastewater treatment process,has the advantages of good solid-liquid separation,low sludge production,simple operation and strong treatment capacity,which has become a water treatment technology vigorously developed in the field of wastewater treatment.Currently,membrane bioreactor is in a period of rapid development,in the process of membrane water treatment,membrane pollution problems cannot be ignored.Membrane pollution can cause a decline in effluent quality,increase energy consumption and other problems,which seriously restricts the promotion and application of MBR.Therefore,a targeted approach to membrane pollution prediction,to ensure that the membrane bioreactor can be operated stably with low energy consumption is the key to promote the widespread application of MBR.In this thesis,the membrane pollution prediction of MBR system is studied based on the improved sparrow search algorithm combined with neural network method.The main research contents are as follows:(1)The concept of membrane bioreactor,its technical advantages and its development and application at domestic and abroad are analyzed from existing theories;the mechanism of membrane contamination in MBR systems,its classification,influencing factors and its research status are reviewed to lay the foundation for subsequent membrane pollution prediction.(2)This thesis takes the(Sparrow search algorithm)SSA as the core and proposes a multi-strategy mechanism to improve the standard Sparrow search algorithm for the shortcomings of the SSA such as easy fall into local optimum and slow convergence speed,which limit its performance with the combined membrane pollution model of neural network.Firstly,the Tent chaos mapping was introduced to improve the uniformity of the initial population distribution of sparrows;secondly,the positive cosine optimization algorithm and the adaptive step size strategy were used to improve the update formulae of discoverer and follower positions to enhance the global exploration ability and optimization accuracy of the algorithm;finally,the Improved Sparrow Search Algorithm(ISSA)was tested in 10 tests.The performance of ISSA was verified on 10 test functions.The experimental results show that ISSA has greatly improved in terms of accuracy,stability and convergence speed,laying a foundation for the establishment of membrane pollution prediction models.(3)During the operation of MBR systems,there are characteristics such as large amount of data and many relevant parameters,which make the selection of auxiliary variables difficult and the prediction model of membrane pollution difficult to establish accurately.Therefore,a membrane pollution prediction model based on ISSA combined with BP Neural Network(BPNN)was proposed as the main technical tool.Firstly,Principal Component Analysis(PCA)was used to reduce the dimensionality of the membrane pollution data and select the auxiliary variables for the model;secondly,ISSA was used to optimize the key parameters of the BP network to obtain better network parameters;Finally,an ISSA-BP membrane pollution prediction model was established to predict the membrane flux,which is an important indicator of membrane pollution.The simulation results show that the prediction accuracy of the proposed membrane pollution prediction model reaches 97.94%,which is much higher than that of the traditional BP neural network model,and achieves the purpose of accurate prediction of membrane pollution.(4)The membrane flux can measure the degree of membrane pollution,but it is not enough to fully portray the state of membrane pollution,and the interference of environmental noise with randomness can affect the reliability of membrane pollution data.Therefore,based on the previous MBR singleparameter prediction model,a multi-parameter prediction model based on Wavelet Threshold Denoising(WTD)and ISSA combined with Long ShortTerm Memory Network(LSTM)method is proposed.The WTD was used to remove the noise from the original data,and then the Spearman correlation was used to analyses the correlation between the membrane pollution characteristics,the Trans-Membrane Pressure(TMP)and membrane flux were selected as the characteristics reflecting the degree of membrane pollution.The final multi-parameter prediction model of membrane pollution based on ISSA-LSTM was developed.The experimental results show that the proposed multi-parameter prediction model has good prediction effect and high accuracy,its can describe the pollution status of membrane modules in MBR more comprehensively.