摘要
针对复杂工况下的滚动轴承振动信号,提出一种基于广义回归神经网络-柔性最大值分类模型的故障诊断分类方法,实现故障模式的识别。对滚动轴承振动信号进行变分模态分解,特征提取等预处理得到特征数据集,并将其划分为训练集,验证集和测试集;使用训练集和验证集训练广义回归神经网络-柔性最大值分类模型,同时引入灰狼优化算法优选该模型的关键参数平滑因子得到理想的分类模型;将训练好的模型应用测试集,输出故障识别结果;通过模拟试验采集不同工况下的轴承故障数据,进行方法有效性验证。结果表明该方法能在小样本训练集下实现对不同工况下的轴承故障的有效诊断,是一种适用于实际工况的故障诊断方法。
Aiming at rolling bearing vibration signals under complex working conditions, a bearing fault diagnosis classification method based on the generalized regression neural network-SOFTMAX(GRNN-SOFTMAX) classification model were proposed to realize bearing fault mode identification. Firstly, the variational mode decomposition(VMD) was performed for rolling bearing vibration signals to do feature extraction and other pre-processing, and obtain a feature data set. The feature data set was divided into a training one, a verification one and a test one. Then, the training set and test set were used to train the GRNN-SOFTMAX classification model. The grey wolf optimizer(GWO) was introduced to optimize the key parameter’s smoothing factor of the above model, and obtain an ideal classification model. Finally, the trained model was applied in the test set to output the fault identification results. Through simulation tests, bearing fault data under different working conditions was collected to verify the effectiveness of the proposed method. Results showed that the proposed method can use a small sample training set to realize effective diagnosis of bearing faults under different working conditions;it is a fault diagnosis method suitable for actual working conditions.
作者
陈剑
吕伍佯
庄学凯
陶善勇
CHEN Jian;L Wuyang;ZHUANG Xuekai;TAO Shanyong(Institute of Sound and Vibration Research,Hefei University of Technology,Hefei 230009,China;Anhui NVH Engineering&Technology Research Center Anhui Province,Hefei 230009,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2020年第21期1-8,16,共9页
Journal of Vibration and Shock
基金
安徽省科技重大专项(17030901049)。