故障分类的改进证据理论信息融合技术研究

Research on the Improved Evidence Theory Information Fusion Technology for Fault Classification

为了实现Dempster合成规则在奇异证据间信息融合后可信度分配的合理赋值,提出了一种基于改进证据理论的信息融合新方法。通过引入可信度因子,对原始证据进行修正,再由证据理论的合成规则进行信息融合,并以汽轮机组作为研究对象进行分析验证。实例表明,该方法可以在原始证据出现高度冲突的情况下得到较好的符合性判定结论,并避免了几种代表性的改进方法在合成一致证据时所存在的发散现象,从而验证了该故障分类方法的有效性。

In order to realize reasonable assignment of credibility assignation of Dempster synthetic rules after information fusion between singular evidence, the new method for information fusion method based on improved evidence theory is proposed. Through introducing credibility factor, the original evidence is corrected; then the information fusion is carried out by the synthetic rule of the evidence rule, and with the turbine as the research object, analysis verification is conducted. The practical examples show that under high conflict condition of the original evidence, this method still obtains better conclusion of compliance determination, and avoids the divergence phenomenon existing in consistent evidence synthesis of several representative improved methods, thus the effectiveness of this method of fault classification is verified.