Research of Diagnosis method Based on Optimized D-S

In fault diagnostic, because of the imperfection of the test node, unpreciseness of the measured value and unknowable structure of system, people have to go on fault uncertain diagnosis. Among them, evidence theory is important for it´s advantage in uncertain expression and focus. In evidence theory, the key is to get the basic probability value. With incorrect basic probability value, the ability of fault diagnosis will be reduced. Now, they are always presented subjectively by experts, but the given values and the facts aren´t really match. This paper puts forward a way of optimizing them to objectively solve the key problem of basic probability value. First, this paper explicates the optimization background of problem, then presents a set of target function and limited function. Solution of these function are found through particle swarm optimization algorithm. In simulated experiments, the result shows that: the way in this paper to find solution of basic probability value heightens the credibility of the fault diagnosis.