Credibility Evaluation Method Based on BP Network for Large-Scale Complex System Simulation
Huang Ke-di · Jisuanji fangzhen · 2006
Huge numbers of data and great influences from surrounding and structure variations are characteristics of credibility evaluation for LSCS simulation; and the concurrent processing and self-learning capabilities ANN technology provided meet the requirements perfectly. Firstly, the foundation of applying ANN technique to credibility evaluation was established by building up evaluation model and comparing it with ANN algorithm, and the evaluating process was presented accordingly. Then, the BP-based credibility evaluation method was researched from designing network structure, choosing and pre-processing training sample, and analyzing learning algorithm. And finally, the application of the method was illustrated by an Air-Ground battle simulation. Results show that the method not only has preferable precision, but also is simple and adaptive in operation.