多值属性系统的故障诊断策略最优化方法

Optimal method for fault diagnosis strategy with multi-value attribute system

针对过程工业系统故障诊断的检测次序问题,其检测属性存在多个取值。为了降低此类系统的故障诊断费用,提出使用人工智能的搜索剪枝技术来生成最优故障诊断策略。根据多值属性系统及其诊断策略问题的描述,给出了穷举诊断策路的生成树剪枝方法。根据此方法,形成2种故障诊断策略最优化的算法:广度优先生成树深度优先剪枝算法(BFS-DFC)和启发广度优先生成树动态规划剪枝算法(HBFS-DPC)。算法分析和实例结果表明此方法适用于小系统的最优故障诊断策略设计,并通过例子展示了两种算法的差异。

The problem of constructing optimal test sequences to diagnose permanent faults in process industrial systems whose test attribute may have multi-values is considered. Searching and cutting technologies of artificial intelligence were provided to reduce the economic losses due to cost variances of detection for generating optimal fault diagnosis strategy. Our method of spanning-cutting is based on searching all the strategies for the multi-value attribute system and the problem of diagnosis strategy. According this method, we provide two algorithms for searching the optimal diagnosis strategy, one of them is the breadth first spanning tree and depth first cutting tree algorithm （BFS-DFC）, and the other is the heuristic breadth first spanning and dynamic programming algorithm （HBFS-DPC）. The algorithm analysis and example result demonstrate that the method is efficient for the optimal fault diagnosis strategy design of small systems, and further examples were used to exhibit the difference between BFS-DFC and HBFS-DPC algorithms.