An uncertain information fusion method based on possibility theory in multisource detection systems

In multisource detection systems, the information representing detection characteristics is uncertain, also the relationship between detection characteristics and detection performance has uncertainty, meanwhile the influence degrees of each characteristic on the performance are also different. In order to represent and process these uncertain information effectively, this paper proposes an uncertain information fusion method based on possibility theory. Firstly, possibility distributions of a series of characteristic parameters about sensor information are constructed, and effective detection characteristics are extracted through the similarity measurement according to the similarity based on distances. The uncertain relationship between detection characteristics and detection performance is quantified, and a certainty calculation method of possibility distributions is proposed based on the combination of holistic and local characteristics. According to the above certainty, the weights of each characteristic are obtained to achieve the fusion of possibility distributions. Finally, the metallic and nonmetallic adhesive structure is taken as a case, and this case shows that this method not only considers the differences of the various characteristics contributed on adhesive performance, but also can achieve the fusion of possibility distributions under different situations. It is a more reasonable fusion method and fits the reality well in practice, also it provides new method and new idea for the fusion of possibility distributions.