Multisensor information fusion algorithm and application combining D-S evidence theory and BP neural network

In modern medical science, several medical data signals are required to be collected together at the same time, such as, pulse, heart sound, respiratory sound, and so on. So that multisensor is necessary to gather multichannel signals, and to extract their respective feature, and finally produce the fused diagnosis results. This paper proposed a method of multisensor information fusion which combines BP neural network and D-S evidence theory. In this way, the weak real-time of mono-neural network, caused by the frequent iterative times, is perfected and the conduction of D-S evidence theory to the system is more accurate, in terms of applying large numbers of standard samples to neural network training. Finally our algorithm is proved to be practicable by experiments.