基于最大后验概率的K/N规则研究

Research on the K/N Rule Based on Maximum Posteriori Probability

现有K/N规则大都未考虑判决代价,未提供一般式,且从数学意义上不符合假设检验的最优准则。本文基于最大后验概率准则,使用贝叶斯理论,考虑判决代价,推导出K/N融合规则的一般式。数值试验结果表明本文的算法较同类算法性能更佳,适用性更广。同时本文还研究了漏检率,误检率之间的关系和组成传感器的选择规则。

Most of current K/N fusion rules don＇t consider the adjudging cost and don＇t provide universal formula, and don＇t fit to the optimization rule of the hypothetic test theory on the mathematics significance. This paper educes the universal formula of the K/N fusion rule based on Maximum Posteriori probability, using Bayesian theory and considering the adjudging cost. The result of data examination indicates that the proposed arithmetic is better and more universal than other similar arithmetic. In addition, the relationship between the leaking-check rate and the error-check rate, and the selection rules of the sensors are investigated in the paper.