基于旋转机械振动参数图形融合灰度共生矩阵的故障诊断方法

A Fault Diagnosis Method Based on Combination Gray Co-occurrence Matrix of Vibration Parameter Image for Rotating Machinery

旋转机械振动多维图形信息一直没有得到充分利用,在一定程度上影响了诊断技术的推广和应用,针对这一问题,该文研究了直接提取和挖掘旋转机械振动状态参数图形中的纹理特征信息的方法,提出利用描述图形灰度空间分布特性和空间相关性的灰度共生矩阵分析图形纹理特征,解决了灰度共生矩阵受所选取的方向影响的问题,最后利用人工免疫算法实现旋转机械故障诊断。在600MW模化汽轮机转子试验台上进行了转子正常、转子不平衡故障、转子不对中故障及轴承松动故障的试验并应用上述方法进行了故障诊断,诊断结果表明可以获得较高的诊断精度,为旋转机械故障诊断探索了一条新途径。

For scarce consideration of multi-dimensions image information, ,which affects the diagnosis technology promotion and utilization in a certain extent, this paper develops a method which directly extracts and mines texture feature in vibration parameter image for rotating machinery. It presents using the gray co-occurrence matrix for analyzing the texture feature, which describes gray space distribution characteristic and space correlation. At the same time, this method solves the problem that gray co-occurrence matrix is affected by the selected direction. Finally, this paper adopts the artificial immune method to diagnose rotating machinery fault. On the modeling of 600 MW turbine experimental bench, rotor＇s normal state, fault of unbalance, misalignment and bearing pedestal looseness are being examined. The diagnosis results show that gray co-occurrence matrix combination can obtain more accurate result. This is a new way for investigating fault diagnosis of rotating machinery.