基于改进蚁群算法的多值属性系统故障诊断策略

Fault diagnosis strategy of multi-valued attribute system based on improved ant colony algorithm

针对传统蚁群算法难以精准解决多值属性系统(multi-valued attribute system,MVAS)诊断策略的问题,在改进蚁群算法的基础上,提出一种改进蚁群算法的测试序列寻优(ANT-TS)算法以搜索MVAS的故障测试序列.首先,引入多值D矩阵和五元组完成诊断策略的公式化处理;然后,为实现ANT-TS算法与MVAS诊断策略的融合,重新表述蚁群算法、设置状态转移规则、设定信息素初始化及更新的方式;最后,通过实例说明算法的实现过程,运用随机仿真实验验证其正确性和稳定性.结果表明:与传统蚁群算法相比,ANT-TS算法的运行过程与诊断策略的一致,且其参数和循环次数少、期望测试费用低、运行速度快;与传统的MV-IG算法和多值Rollout算法相比,ANT-TS算法能获得费用较少的测试序列.

Aiming at the problem that the traditional ant colony optimization(ACO)algorithm cannot solve the diagnosis strategy for multi-valued attribute systems(MVAS)accurately,based on the improvement of the ACO algorithm,the ANT clony optimization-test sequence(ANT-TS)algorithm is proposed to search the fault test sequence for MVAS.Firstly,multi-valued D matrix and five-tuple are introduced to complete the formulation of the diagnosis strategy.Then,the ant colony algorithm is reformulated,where the state transition rule and initialization and update of pheromone for the algorithm are set to combine the diagnosis strategy of MVAS with the ANT-TS algorithm.Finally,the correctness and stability of the ANT-TS algorithm are verified by an example and stochastic simulation experiments.The experimental results show that the running process of the ANT-TS algorithm is the same as the fault diagnosis strategy of MVAS.The algorithm has fewer parameters,expected test cost and number of cycles,and run faster when comparing with the traditional ACO algorithm.The ANT-TS algorithm can obtain the test sequences with less expected test cost compared with the traditional algorithm such as the multi-valued Rollout algorithm and the multi-valued IG algorithm.