先验概率和代价函数均模糊时基于贝叶斯最小风险准则的分布式决策融合

当先验概率和代价函数均为梯形模糊数时，在贝叶斯最小风险准则意义下，研究了在融合中心对多个独立传感器的决策进行最优融合的问题，给出了四种决策融合算法，通过仿真和比较这四种融合算法的结果，找到了一种最适用于这种场合的最优决策融合算法。结果表明，在先验概率和代价函数均为梯形模糊情况下所导出的最优决策融合规则是各检测器决策的加权和与一门限之比较，权重是各检测器检测概率和虚警概率的函数，门限除与最优融合准则

Optimal Distributed Decision Fusion with Fuzzy a priori Probabilities and Fuzzy Cost Functions Based on Minimum Bayesian Risk Criterion

When the a priori prohabilities and imt functions are fuzzy, the optimal decision fusion in the sense 0f minimum Bayesian risk at the fusion center is considered. The fusion center receives decisions from various distributed sensors andfour optimal decision fusion schemes at the fusion center are derived. It is discovered that the optimal decision fusion rule is aweighted sum of local decisions in this case, the weights are functions of the prohability of detection and the probaility of falsealarm of the detector, and that the threshold depends not noly on the fuzzy a priori probabilities and cost functions but also onthe criterion used for defuzzfying fuzzy sets. Through the simulation, an optimal decision fusion scheme which is most suitablefor fuzzy a priori probabilities and cost functions with trapezoidal membership functions is found.