Intelligent Detection Method Of Sludge Bulking Based On Self-Organizing Type-2 Fuzzy Neural Network

The activated sludge process(ASP)is a kind of wastwater purification method,which decomposes and oxidizes biodegradable organic substances in wastwater by domesticated microorganism groups.The ASP,which has the advantages of simple structure,convenient operation and management,and high processing efficiency,has been widely used in urban wastewater treatment process(WWTP).As a common and frequently occurring process problem in activated sludge process,sludge bulking has become the bottleneck of operation management and control of WWTP,which restricts the development of urban wastewater treatment.Sludge bulking can easily lead to sludge loss and sludge sedimentation performance degradation,resulting in excessive effluent quality,even causing a decline in wastewater treatment capacity and operation system collapse.Due to the complicated mechanism and changeable causes of sludge bulking,the traditional identification method is difficult to meet the precision and stability requirements of WWTP.It is very important to develop an accurate and efficient detection method for sludge bulking.In order to realize the detection of sludge bulking,an intelligent detection method based on self-organizing type-2 fuzzy neural network is proposed.First of all,the formation mechanism and influencing factors of sludge bulking were studied.Combined with the working conditions of urban WWTP,the type of fault of sludge bulking and the main water quality variables were obtained.Secondly,the self-organizing type-2 fuzzy neural network based on the intensity of information transmission algorithm was designed.The structural adjustment and parameter optimization improve the approximation performance of the network.Then,a SVI soft-computing model based on self-organizing fuzzy neural network is established.The online prediction of sludge bulking is realized by constructing SVI soft-computing model based on self-organizing type-2 fuzzy neural network.Finally,the intelligent detection method based on target-related identification algorithm is proposed.The intelligent detection of sludge bulking types is realized by target-related identification algorithm.The works in the paper are as follows:1.The characteristic analysis of sludge bulking.The types of sludge bulking fault and main water quality parameters affecting sludge bulking were obtained.Analysis the mechanism of sludge bulking and the actual operating conditions in WWTP,5 types of sludge bulking fault are obtained:(1)Low oxygen concentration fault(2)Nutrient deficiency fault(3)Low sludge loading fault(4)Low temperature fault(5)Low potential of hydrogen fault.Meanwhile,according to the key water quality parameters of the sludge settlement process,the main water quality parameters affecting the sludge volume index(SVI)are obtained: biological oxygen demand(BOD),chemical oxygen demand(COD),dissolved oxygen(DO),total nitrogen(TN),total phosphorus(TP),mixed liquid suspended solids concentration(MLSS),organic load rate(F/M),temperature(T)and(PH).2.The study of the self-organizing type-2 fuzzy neural network.In this paper,a self-organizing mechanism of type-2 fuzzy neural network based on the intensity of information transmission algorithm is designed to improve the model approximation effect by dynamically optimizing network structure and parameters.Firstly,through the analysis of the contribution degree of the independent components of the firing layer neurons,the growth or deletion conditions are judged,the structural adjustment of the type-2 fuzzy neural network is realized,and the flexibility and adaptability of the network are enhanced.Secondly,an adaptive second-order algorithm is developed to optimize the parameters of fuzzy neural network and improve the accuracy of network operation.Finally,the self-organizing type-2 fuzzy neural network is applied to nonlinear system modeling,and the experimental results show that it has a simple structure and high prediction accuracy.3.The SVI soft-computing model based on self-organizing type-2 fuzzy neural network.A SVI soft-computing model based on self-organizing type-2 fuzzy neural network is established.The correlation variables are selected by partial least squares(PLS)algorithm.Among them,the 8 variables with the largest correlation are used as soft-computing model inputs,and the SVI values are used as model outputs to construct the SVI soft-computing model based on self-organizing type-2 fuzzy neural networks.The model structure and parameters are calibrated by the operation data in the actual wastewater treatment plant,and the validity of the soft-computing model is tested.The experimental results show that SVI soft soft-computing model based on self-organizing fuzzy neural network can effectively predict SVI.4.Research on intelligent detection method of sludge bulking.An intelligent detection method for sludge bulking based on target-related identification algorithm is proposed.Firstly,the residual size of the predicted values of the self-organizing type-2 fuzzy neural network soft-computing model SVI and the actual values,can be used to determine whether the sludge bulking occurs.Subsequently,when the sludge bulking occurs,the target-related identification algorithm is used to identify the fault type and realize the intelligent detection of sludge bulking.Finally,the effectiveness of the intelligent detection method is verified by experiments.The experimental results show that the intelligent detection method of sludge bulking can accurately detect sludge bulking and identify fault type.