基于遗传算法的信息融合在摩擦学组合故障诊断中的应用

根据8NVD48A－2u型柴油机现场运行油液监测试验结果和长期的实践经验，在600多个船舶柴油机故障诊断实例的基础上，总结出了各种零件磨损故障种类之间的关联度，结合D－S证据组合理论给出了零件磨损的摩擦学故障组合支持度；利用遗传算法搜索出了缸套活塞组的划伤、拉缸和腐蚀；轴承的擦伤、烧损、剥落和腐蚀、齿轮的点蚀、胶合和剥离等10种单一故障类型的5种组合故障，并确定了各种油液监测方法对这5种摩擦学组合故障的组合支持度。把遗传算法引入信息融合，为解决摩擦学组合故障问题提供了一种新的思路，对实际故障诊断具有一定的指导意义。

Application of Information Fusion Based on Genetic Algorithm to Diagnose of Combined Tribological Failures

The relevance of the wear type diagnose of various parts of model 8NVD48A 2u diesel engines is established by summarizing over 600 practices on wear failure diagnose of the diesel engines, based on the field oil monitoring results and various other testing experiences. The combined reliability of the wear failure diagnose to various parts of diesel engines is given based on Dampster Shafer theory (briefed as D S theory). Five combined wear failure modes of the parts have been found from the wear failure modes including the scoring, seizure, and corrosion of piston and piston ring/cylinder liner pair, the scratching, seizure, spalling, and corrosion of gears, and the pitting, seizure, and fatigue of gears, by using genetic algorithm. The applicability of various oil monitoring methods to the diagnose of the five combined wear failure modes has been established. Thus a new train of thought is suggested for the diagnose of the combined wear failure modes by introducing genetic algorithm into information fusion method.