Application Research On Fuzzy Neural Network In Co-scheduling Decision-making Of Wastewater Treatment Plant

Biological-ecological combined process had combined the common advantages ofbiological treatment and ecological treatment, which can able to realize biologicalsegment’s flexible and efficient features and eco-section’s low energy andenvironmentally friendly features. Nowadays the scheduling of combined process ofbiological and ecological is mostly operated by humans; those may not guarantee thelowest energy consumption. As neural network has features of self-learning andself-organizing, and fuzzy logic has characteristics of transparent and logical, this paperproposed a fuzzy neural network intelligent scheduling algorithm, which can ensure thecombined process to achieve optimal scheduling.The subject combined fuzzy logic inference, artificial neural network and computertechnology, forming a co-scheduling expert system for the sewage treatment plant.Based on some relevant parameters, such as influent flow rate, temperature, and theinput water quality, at the same time,combined biological and ecological unit processcharacteristics，this system can collaborative schedule those two stage process torationally allocate pollutant load. Mainly the following aspects were studied:①Established up the energy consumption function for the small town wastewatertreatment plant. According to the kinetic theory of the sewage, this paper establishedenergy consumption model for small town sewage treatment plant, then according to theenergy consumption model calculated how much processing pollutants load thecorresponding biological segments should to bear.②Selected structure model for fuzzy neural network. According to theparticularity of this sewage treatment plant，this paper adopted the T-S structure, thoughanalysis and research,this paper chose COD, temperature, flow and PO43-as thenetwork input, Its membership functions are Gaussian function and every functiondivided into5fuzzy sets，the output this network is the biological treatment sectionborne pollution load.③Chose suitable fuzzy network training methods. This paper had chosen BPalgorithm to training first piece of this fuzzy neural network and chosen hybridalgorithm to training the second piece. Simulation results had showed that the hybridalgorithm of fuzzy neural network whose prediction error could converge to0.010012and iterative convergence was about120steps, better than BP algorithm. So this fuzzy neural network has features of small error and fast convergence, could completely beused in scheduling biological/ecological combined process of sewage treatment④Realized the system functions. Based on the function of fuzzy neural network,developed an expert system which used to obtain the knowledge, realize fuzzyreasoning, and the give the appropriate scheduling decisions, and in order to increaseuser confidence in the result the explain function will give the reasoning processThe results showed that create a fuzzy neural network can be better used to solvethe question of load distribution between biological and ecological processes, and canpredicted the water quality accurately. Therefore, combined self-learning withself-organizing functions of neural network with advantages of fuzzy reasoning in fuzzysystems to format the fuzzy neural network system, which had a general value ofapplication in intelligent scheduling of sewage treatment plant...