System Prediction and Control Model Based on Data Simulation and its Application

A new model is introduced in this paper to construct the input-output relation in the prediction and control problem of non-analytic systems. The historical input-output data of general system is de-noised with wavelet transformation and SVM, and the input-output variables which can reflect the features of the system are determined with correlation analysis and sensitivity analysis. With the historical information of these variables, the input-output relationship of singular systems is constructed through system and parameter identifications with differential simulation and neural networks. Based on this method, a software/hardware module with prediction and optimal control functions is able to be designed which can realize the functions through computers or some other electronic products and can be used in relative engineering fields. This method is not only beneficial to the data proceeding problems on geology, water quality, meteorology, environment and so on, but also can promote the development of information science itself.