城市污水处理水质模拟与预测研究环境科学与工程专业论文.docxVIP

Abstract1he Abstract 1he Artificial Neural Network(ANNl is a developping newest subject that based on neural science．It is a mathematic model of theorization of human brain nerve systems．With its concurrency and highly nonlinerity,it can carry through complicated logical operation．AlSO it can carry out the nonliner relations between inputs and outputs by utilizing the recollection and learning of nerve cell in network． Along with industrial contr01．the process controlin wastewater treatment has gained application at large and rapidly development．Especially,the theoris and applications of ANN and fuzzy neural network not only provide theory basis to opening and complicated wastewater treatment system，but also enable control the wastewater treatment process． This paper is performed based on pilot field of national 863 key project：study on technics of urban wastewater treatment and reuse and engineering demonstration． AU the data in this P印er are from the real．time supervision and the results of water characteristics analysis．The study results in this paper iS basic study of A NN soft．measure for water characteristics of wastewater treatment demonstrating engineering，and pre-study of fuzzy control for wastewater treatment process． Water qualities of COD、SS、TP、NH3．N have been simulated and predicted that inlet and outlet of bio．chemical reactor and of four construction wetlands and of whole wastewater treatment process．After testing the structure of ANN and quantities of nerve cell many times，the best results has been obtained for each water index．It evaluates the advantage and shortcoming of the A NN structures of multi-input and multi—output(MIMO)and multi—input and sigle—output(MISO)for simulating water index，and point out that the MISO BP structure best than the MIM0 BP structure．The predicting results show that the best correlation coefficient is between 0．64 and 0．98． To obtain the application values of fuzzy neural network for simulating water index．the research