The topic of water pollution is constantly being raised. In September, 20 14, some media exposed the phenomenon of illegal sewage discharge by enterprises in Tengger Desert at the border between Inner Mongolia and Ningxia, which has caused pollution to the surrounding environment, thus arousing people's extensive concern about the seriousness of water pollution and awakening people's awareness of environmental protection. Sewage treatment is widely used in construction, agriculture, transportation, energy, petrochemical, environmental protection, urban landscape, medical care, catering and other fields. Activated sludge process is a common sewage treatment method. It is reported that more than 95% of urban sewage and more than 35% of industrial wastewater are treated by activated sludge process. Aeration tank is the heart of all activated sludge processes, and its function is to stir the mixed liquid, so that the mud and water can fully contact and provide oxygen for microorganisms. One way of stirring is to fully mix the muddy water entering the aeration tank at the same time and keep it outside the outflow tank, instead of mixing it with the mixed liquid phase already in the tank to avoid short circuit. The long strip aeration tank is used to ensure that the muddy water entering the tank at the same time is discharged from the tank at the same time, so that the wastewater entering the tank at the same time has the same aeration time. Chemical oxygen demand (COD) is the most important indicator of sewage treatment, which is used to indicate the amount of residual organic pollutants in effluent or purified water. A sewage treatment plant adopted the activated sludge method, and recorded the main parameter values of the aeration tank section in the first half of 20 14, including influent, influent COD, dissolved oxygen, blower pipe pressure, activated sludge concentration, blower outlet valve opening, blower inlet valve opening, oxidation reduction point, outlet COD, etc. Typical parameters such as COD are recorded once every half hour; Other parameters such as flow, opening, etc. Every 10 minute. See appendix 1 and appendix 2 for specific data. Problem 1: delay estimation. Analyze the delay relationship between the parameters in the original data and the effluent COD, and output the results. Question 2: With reference to the results of the previous question, a nonlinear prediction model of COD for each parameter is established. Question 3: According to the results of the first two questions, how to adjust the parameters to reduce the COD value to 35. If the COD value has reached 35, there is no need to adjust other parameters.