基于人工免疫的异步电机故障诊断系统

ASYNCHRONOUS MOTOR FAULT DIAGNOSIS BASED ON ARTIFICIAL IMMUNITY

基于免疫系统的原理,提出了一种用于异步电机故障诊断的新方法。针对阴性选择存在的不能辨识故障种类的缺点,提出一种改进的阴性选择方法,该法为实数编码、用欧氏距离来判断匹配,增加了记忆集,具有学习能力,能够辨识故障种类。并借鉴生物免疫亲和度较高的抗体之间抑制作用,增加对进入记忆检测器集的有效检测器与已有的记忆检测器亲和度计算,抑制亲和度高的检测器;根据电机定子匝间短路的实验结果,提出短路负载电流的基波经过时频变换后的结果,在一定的区域与负载成正比关系;使用所提出的改进的阴性选择方法对匝间短路进行检测,检测结果证明了改进的阴性选择方法克服了传统阴性选择方法的学习能力差,不能辨认故障的种类缺点,对故障检测的准确率很高。

Based on the principles of the immune system, a new method of fault diagnosis for asynchronous motor is presented. The improved negative selection algorithm （INSA） is proposed to overcome NSA＇s disadvantage of indistinguishable fault kind, which is coded by real number and concluded matching by Euclid Distance （ED）. Using restrained action between high affinity antibodies in biology immune system for reference, affinity between candidate detectors and memory detectors is computed to restrain the higher affinity antibodies based on INSA. Based on the experiment results of motor stator turn-to-turn short-circuit, that outcomes and load have linear relations in certain area is proposed after the fundamental component of shorted load current is transformed by time-to-frequency. The INSA is applied in checking turn-to-turn short-circuit. The INSA is proved to overcome the conventional negative selection algorithm＇s shortcomings of poor learning ability and undiscriminating between fault kind, be exactness and validity by experimental results.