基于Java的发动机紧致型小波神经网络故障诊断

The compact type of wavelet neural network fault diagnosis for engine based on java

鉴于发动机是一种复杂的机电液一体化设备,其故障现象和原因之间存在复杂的非线性关联。本文结合小波变换的良好时频域特性和神经网络良好的非线性映射的优势,将MexicanHat小波基作为神经网络的传递函数,组建紧致型小波神经网络,用于发动机的故障诊断;本文以小波神经网络为算法基础,应用具有跨平台、可移植优点的Java语言和SQL Server2005数据库,开发出发动机智能故障诊断软件。

As the engine is a complex electromechanical device,the complex non-linear correlation exists between the fault phenomenon and the reason of the engine.With the advantages of wavelet transform time-frequency domain characteristic and the neural network nonlinear mapping,this paper makes the MexicanHat wavelet as the neural network＇s transfer function and constructs a compact type of wavelet neural networks for engine fault diagnosis;Based on the wavelet neural network algorithm,this paper develops the intelligent fault diagnosis software with the Java language＇cross-platform,portable benefits and the SQL Server2005 database.