Quasi-Newton BP神经网络算法在电梯急停故障诊断中的应用  

Application of Quasi-Newton BP Network Algorithm in Fault Stop Diagnosis of Elevator

作　　者：夏文杰[1] 杨建武[1] 李屹[1] 李昆[1] 

机构地区：[1]北京工业大学机械工程与应用电子技术学院,北京100124

出　　处：《计算机测量与控制》2011年第1期23-25,共3页Computer Measurement &Control

基　　金：北京市科技计划项目(Z07050601480701)

摘　　要：为了判断电梯运行是否故障急停,结合电梯动态智能检测系统的结构和特点,采用基于多传感器数据融合技术的3层BP神经网络方法,并将其应用到电梯动态智能检测系统中;因对不同传感器采集的信号采用不同的处理方法,训练样本包括基于小波包分析的能量特征向量,峭度系数、峰峰值时域特征值;Quasi-Newton BP算法经104步完成对样本训练,精度是2.6×10-4,实现检测系统的智能化急停诊断;结果表明该算法优于弹性BP算法。In order to diagnose the elevator fault stop precisely, and according to the characteristics and structure of the intelligent elevator detecting system, this paper applies a method of three--layer BP neural network based on multi--sensor data fusion. For different sensor signals using different approach, training samples includes wavelet packet energy eigenvector, also involves such as kurtosis, peak to peak value and noise signal. Take the above eigenvectors as fault samples, the Quasi-Newton 13P trained by the 104--step, and the numerical ac- curacy is 2.6X 10-4. The trained network can exactly diagnose the elevator fault stop. The results show that the Quasi--Newton BP algo rithm is batter than the resilient back BP algorithm.

关 键 词：多传感器数据融合 电梯检测 神经网络算法 小波包特征向量 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]