Constructing Basic Probability Assignment Based on the Slope Correlation Degree

DS evidence theory can be applied to deal with multi-sensor data and get reliable and accurate decision result of system’s target or object due to its good measurement and reasonable description of system’s uncertainty. In practical application field, DS evidence theory is achieved by two steps, one is the acquisition of the basic probability assignment function (BPA), and the other is the synthesis of multiple evidences that is obtained from multi-sensor data. The existing relevant literatures mainly solve the combination problem of conflict evidences, while lack of systematic research on the acquisition of BPA. In this paper, we focus on the discussion and innovation of the acquisition of BPA. BPA in DS evidence theory is the basic unit to describe the uncertainty of the system and evidences, which fundamentally determines its superiority with respect to probability theory and Bayesian reasoning. Getting a reasonable and effective BPA is the premise of the application of DS evidence theory. Traditional BPA acquisition methods have the limitations like absent priori knowledge and low acquisition accuracy. Aiming to solve the technical difficulties of traditional methods, we put forward a new BPA acquisition algorithm based on the slope correlative degree. The novel algorithm proposed in this paper successfully realizes the reliable descriptions of system’s uncertainty, which is conducive to product the accurate BPAs and the precise fusion results. Experimental results and analyses demonstrate that the innovative algorithm can not only produce reasonable BPAs, which can better fit the sensors’ monitoring data, but also expand the gap between the support degree of different propositions in BPA, which is more beneficial to the synthesis of evidences and the decision fusion of the system. Thus, the novel BPA acquisition algorithm has great application value and research significance.