基于小波包遗传算法的冲压模具故障识别研究
骆志高，王祥，李举，范彬彬，郭啸栋
(江苏大学机械工程学院镇江212013)
摘 要:根据冲压模具失效机理分析及相应声发射信号的特点，讨论冲压模具的特征参数。基于小波包分析技术，提取各频带内的能量与总能量之比并确定为初始特征参数,并应用遗传算法对初始特征参数进行优化，生成新的特征参数。通过实验对冲压模具实时采集信号的参数与正常状态和失效状态两种情况下特征参数的隶属度比较,有效识别了模具的工作状态，验证了小波包遗传算法应用于冲压模具故障识别的可行性。
关键词:小波包；遗传算法；冲压模具；故障
STUDY ON PUNCHING DIE FAULT I DENTIFICATION BASED ON WAVELET ANALYSIS AND GENETIC
ALGORITHM
LUO Zhigao WANG Xiang LI Ju FAN Binin GUO Xiaodong
(Co1lege of Mechanical Engineering, Jiangsu University, Zhenjiang 212013)
Abstract: According to the failure analysis of punching die and the characteristics of acoustic emission signals,the parameters of punching die are discussed. extracting the comparison between the different frequency bands energy and total energy based on wavelet analysis technology, and take them as the initial characteristic parameters.Optimizing them by genetic algori thm to generate a new parameter. The punching die signal parameters were collected by experiment, then compare the collected signal parameters with normal state characteristic parameters and failure state characteristic parameters, which can identify the work station of the die effectively. Results show that application is feasible.
Keywords: wavelet package analysis; genetíc algorithm; punching die; fault
冲压模具的工作环境恶劣，不仅表现在承受高接触压力和剧烈的摩擦,还有循环加载引起的应力、应变和温度的周期性变化而使模具产生疲劳失效。它所造成的被迫停车维修，给企业带来了巨大的经济损失。对冲压模具的状态判别是解决其可靠性、安全性的关键途径之一。
由于模具声发射信号背景噪声复杂,很难提取到能反映实际物理状态的信号参数,状态与征兆之间表现出的也是一种非常复杂的非线性关系，这对模具状态进行判定带来了极大的困难。本文针对声发射信号的特点，把声发射信号进行小波包分解,频带内的能量与总能量之比为初始特征参数;再通过对各频带的特征参数进行遗传算法生成新的优化特征参数,以模糊诊断理论，辨别模具的工作状态。
