An Improved Bayesian Classifier Based Basic Probability Assignment Construction Method

Yuan Jin-sh · Electrical Measurement & Instrumentation · 2014

Due to the problem of the basic probability assignment(BPA)function produced by D-S evidence theory being adopted in the transformer fault diagnosis, an improved Bayesian classifier based method has been proposed in this paper, which will confirm the suitable interval according to the different attribute values so as to solve the problem that the basic probability assignment is invariable in the individual areas. Experimental results varify the effectiveness and accuracy.

Read the paper · More papers on PaperTik