Multi-source Information Fusion Technology And Its Application

Information technology is an important role of social construction, and the multi-source information fusion is a key part of information technology. Based on the background and the help of the printing system project,this article came into being.The information fusion technology is divided into two parts:hardware and software. The hardware is mainly in the structure & function of information fusion and the software is mainly in the algorithms of the information fusion. Based on the theory, the information fusion model structures are compared and analysised.After that, a traditional Kalman filter,Bayes probability theory approach and D-S evidence theory in information fusion application are reseached. The reseach of the hardware and software structure provide a reliable theoretical basis for the subsequent fusion of the solution in a complex environment,which is shown as follows:the distributed sensor including'and','or'and 'voting fusion', comparing the advantages and disadvantages between the three under the matlab environment; the wavelet and fractal methods are used for solving the acquisition of singular signal, which are proved that this singular signal detection for both methods have good results from the experiments;for the singular signal detection recovery problems, we use the forward and backward linear prediction method, to restore the original purpose of the data; for the nonlinear and nonstationary complex environments, the neural fuzzy inference system model theory method which is a kind of the modern signal processing, can accurately describe the system characteristics.During the integrated project of printing in high-precision control and measurement system, the data fusion-related knowledge are been into practice, which is mainly in:1. Pressure sensor affected by environmental temperature, which is used the LM algorithm of BP neural Networks integrated the temperature information and pressure information to reduce the influence of the temperature characteristic;2. With the help of the neuro-fuzzy inference knowledge of the theory,the temperature and pressure are merged, which can precisely control the distance of the motor forward to ensure high quality print.