BP神经网络在机械故障诊断中的应用  

Application of BP Neural Network in Mechanical Fault Diagnosis

作　　者：张新海[1] 雷勇[1] 

机构地区：[1]西北工业大学动力与能源学院,西安710072

出　　处：《噪声与振动控制》2008年第5期95-97,共3页Noise and Vibration Control

摘　　要：随着现代化生产的进行,机械设备的故障诊断技术日益受到重视。介绍BP算法神经网络模型,给出BP神经网络应用于机械故障诊断的基本步骤。阐述三层BP神经网络采用数据驱动正向推理的故障诊断策略。即从初始状态出发,向前推理到达目标状态为止。介绍BP神经网络在机械故障诊断中的应用及实例,对某型号拖拉机变速箱中的主要零件——齿轮和轴承进行故障诊断振动测试,对部分齿轮和轴承发生故障时的振动信号进行测量。虽然BP网络在机械故障诊断中已得到广泛的应用,但由于神经网络本身还处于发展之中,还有不少问题需要进一步深入研究。With the progress of modernization of production, people pay more attention to the fault diagnosis techniques for mechanical devices. The model of back-propagation （BP） algorithm network was introduced in this paper. The essential steps of BP network in mechanical fault diagnosis were provided. The fault diagnosis strategy using data-driven positive-direction reasoning method in a 3-layered BP neural network was elaborated. That meant starting with the initial state, reasoning forward until the goal state was reached. Several examples of application of BP neural network in mechanical fault diagnosis were introduced. With this method, the vibration signals of the faults in some gears and bearings in a tractor were measured. Although BP neural network has been widely applied in mechanical fault diagnosis, quite a few problems need to be deeply studied since neural network itself is in developing nowadays.

关 键 词：振动与波 BP算法 神经网络 机械故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TH165.3[自动化与计算机技术—控制科学与工程]