应用混沌神经网络诊断发动机磨损故障

利用混沌神经网络,采用混沌动力学中的Logistic映射构造神经元,对发动机的相似早期磨损故障进行了有效诊断。通过故障模拟实验,对不同类型初期磨损故障的振动数据进行了采集和分析,发现其频谱特征具有较大的相似性,且各种初期磨损故障的李雅普诺夫指数大于0,具有混沌特征。用一般方法难以区分不同类型的初期磨损故障,而混沌神经网络能够有效地识别这些相似故障模式,对于发动机磨损故障的早期预防具有积极作用。

Engine's Friction Fault Diagnosis by Using Chaotic Neural Network

This paper describes a method to build chaotic neural network by using logistic mapping to identify the engine's early friction fault. By imitating different faults and collecting their vibration data, it is found that the spectrum of each fault is very similar, and has a chaotic feature as its Lyapunov index is plus, the normal neural network has difficulties in diagnosing them. Results show the chaotic neural network can identify the engine's similar fault pattern successfully and prevent the engine's early fault effectively.