An improved multi-sensor D-S rule for conflict reassignment of failure rate of set

Jiale Qiao, Jindong Zhang, Yuze Wang

Abstract: In order to realize high-precision sensor sensing system for unmanned vehicle, a method based on D–S (Dempster–Shafer) evidence theory was presented. D–S evidence theory is a method of evidence processing. However, because of the limitation of multiplication rule, it cannot deal with the evidence of high conflict caused by the failure of a sensor in the system. In the existing D–S theory framework, the prior probability is used to measure the sensor’s failure rate. The correlation matrix is obtained by the mass distribution of evidence, and the interval classification is carried out accordingly. Bayes formula is used to adjust the prior probability of the sensor by synthesizing the risk distribution function of different intervals, and the calculation rule of correction report using convolution rule is proposed. Compared to existing methods, it can reduce the conflict degree and information entropy of the system, so that a reasonable decision can be made.