一种改进的BP神经网络在故障诊断中的应用研究

Application of an Improved BP Neural Network in Fault Diagnosis

人工神经网络作为一个具有高度非线性映射能力的计算模型,在工程中具有广泛的应用前景。它不需要预选确定样本的数学模型,仅通过学习样本数据即可以进行故障诊断。本文结合计算机综合业务的一些实际问题,探讨了利用改进的BP神经网络进行故障诊断的方法和应该注意的问题,在分析神经网络的基础上提出了基于改进的BP模型神经网络的故障诊断推理方法。结果表明,基于BP神经网络的故障诊断方法是行之有效的。

Artificial neural network （ANN）, as a computing model possessing high-nonlinear mapping ability, has a wide application future in engineering. It needn＇t predefine the sample＇s mathematical model, only by means of studying the sample data it can be used for fault diagnosis. Combined with some practical problems in fault diagnosis of computer integrated service, the methods of how to use the improved BP-NN in fault diagnosis and the problems which should be paid attention to were discussed. The result indicates that this method is feasible.