Development Of Software System For Effluent Quality Forecasting Based On Soft-Sensing

The thesis developed on an existing problem for forecasting the effluent quality parameters of urban sewage treatment factories, which are usually difficult to measure with conventional online apparatus, through applying soft-sensing technique. The goal of this research is to develop a suit of software system for realizing the forecast of effluent quality based on soft-sensing technique.Research methods applied in the thesis are as follows.The first method is to construct the soft-sensing models of effluent quality parameters with history data of a sewage treatment factory for years. Another method is to plant the models into the application system to develop the computer forecasting system for effluent quality parameters of sewage treatment factories.Two main research results of the thesis are as follows. (1) The obtainment of soft-sensing models for effluent quality parameters forecasting.Firstly, a sewage database is designed with history data of a sewage treatment factory for years. Secondly, many forecasting algorithms for Multiple Linear Regression and BP neural network are designed by using samples exported from the sewage database, At last, a kind of soft-sensing model of sample interpolation and multi-step memory for forecasting effluent quality parameters is presented. The model improved the forecasting of effluent quality parameters mostly.(2) The accomplishment of the software system for effluent quality parameters forecasting based on soft-sensing technique.The application of this system developed with Visual C++ while soft-sensing models designed with MATLAB and the sewage database designed with Access or SQL Server 2000. The thesis integrated the above three into one application system by applying MFC ODBC and MATLAB Engine techniques.In general, there are two innovations in the thesis. One is the hybrid programming technique for Visual C++ and MATLAB with MATLAB Engine, which is an inexpensive and time saving way and can be introduced into other systems. The other is the presentment of the soft-sensing model of sample interpolation and multi-step memory, which can also be introduced into other situations similar to sewage treatment.