论文部分内容阅读
本文根据模糊识别原理，引入集合分类概念，提出了一种用于在线监测发动机缸内部件故障的无监督竞争学习模糊神经网络。该网络仅需正常状态下的振动信号样本集及少量的故障状态样本进行学习，并且它可根据噪声及运行工况的变化，自适应地调整代表正常状态下的网络权值；采用对比增强及加权方法，抑制各样本中的噪声影响。用本文方法对ＥＱ６１００汽油机和１９０Ａ柴油机人工设置的缸套活塞磨损故障进行诊断，取得了理想的效果。本文为以理论指导为主、少量实验为辅的在线诊断内燃机缸内部件故障，探索了一条有效、便捷的途径。
In this paper, based on the principle of fuzzy recognition, the concept of ensemble classification is introduced and an unsupervised competitive learning fuzzy neural network is proposed for on-line monitoring of engine cylinder internals. The network only needs to learn the vibration signal sample set and a few fault state samples under normal conditions and it can adaptively adjust the network weights representing the normal state according to the changes of noise and operating conditions. Weighted method, to suppress the impact of noise in each sample. The method of this paper is used to diagnose the worn wear of cylinder liner piston of EQ6100 gasoline engine and 190A diesel engine, and the ideal result is obtained. This article is based on theoretical guidance, a small number of experiments supplemented on-line diagnosis of internal combustion engine cylinder parts failure, to explore an effective and convenient way.