Abstract: The current decision fusion algorithms of distributed system isn’t fit to the practical application, because these algorithms reckon without alterable target and need the prior probability of the testing-target of given system. This paper develops a new method for self-adaptive decision fusion based on Bayesian, which estimates target by accumulation and volatility strategy of pheromone as ant-colony. Test result shows that detection performance possesses good convergence property with increasing feedback stage, and is suitable for variable systems by changing self-adaptively correlative parameter.

Key words: pheromone, adaptive, decision fusion

摘要： 现有的分布式系统决策融合算法一般不考虑可变目标，且需已知测试目标的先验概率或假设系统虚警率，与实际应用相差较大。该文在贝叶斯融合算法的基础上模拟蚁群利用信息素的积累与挥发结合择优策略估计目标，从而实现自适应决策融合。试验结果表明，检测性能随反馈步数增加具有较好的收敛性，可根据检测对象自适应修改相关参数，适用于可变系统。

关键词: 信息素, 自适应, 决策融合

CLC Number: 

• TP274

• TP212.6