液压机械无级变速箱换段液压故障诊断的BP方法

发布时间：2018-04-29 11:27

本文选题：湿式离合器 + 故障诊断　；参考：《中国农机化学报》2016年10期

【摘要】：提出一种针对液压机械无级变速箱离合器液压系统的故障诊断方法。首先,试验得到87组压力数据并进行数据预处理;而后,通过随机抽取的方法得到训练样本集与测试样本集,基于BP神经网络对训练样本集进行训练并得到故障判别模型,使用该模型对随机样本集进行分类测试;最后,基于神经网络特征选择方法对所得分类模型进行属性约简,并对BP神经网络的多值分类作了比较研究。结果显示:BP二值分类方法对正常及4种典型故障模式的平均识别率分别为98.89%、99.78%、99.78%、99.11%、99.78%;BP多值分类方法对5种典型油路状态模式的平均识别率分别为97.11%、99.78%、99.33%、99.56%、100%;正常模式、活塞卡住、密封圈损坏、油道阻塞与密封不严分别可约简3、4、3、3、3个样本属性。该结果表明:换段期间的油压波动与各故障模式之间存在特异性关联,可通过BP网络方法进行模式分类,但该方法对于组合故障判别无效,还有待进一步研究。
[Abstract]:A fault diagnosis method for hydraulic system of continuously variable gear box clutch of hydraulic machinery is presented. First, 87 groups of pressure data were obtained and preprocessed, then, the training sample set and test sample set were obtained by random sampling method, and the training sample set was trained based on BP neural network and the fault discrimination model was obtained. The model is used to classify the random sample set. Finally, the attribute reduction of the classification model based on the neural network feature selection method is carried out, and the multi-value classification of BP neural network is compared and studied. The results show that the average recognition rate for normal and four typical fault modes is 98.89 and 98.89 for normal and four typical fault modes respectively. The average recognition rate of BP multivalue classification method for five typical oil path state modes is 97.111,99.780.The average recognition rate of the BP multi-value classification method for five typical oil path state modes is 97.111,99.780.99.5699.33100, respectively; in normal mode, the piston clamps, the sealing rings are damaged, Oil channel obstruction and loose sealing can be reduced by 3 / 4 / 3 / 3 and 3 / 3 sample attributes, respectively. The results show that there is a specific correlation between oil pressure fluctuation and each fault mode during the period of change, which can be classified by BP neural network method, but this method has no effect on combined fault identification, which needs further study.
【作者单位】：山东农业大学机械与电子工程学院;南京农业大学工学院;农业部南京农业机械化研究所;
【分类号】：S219.07

【相似文献】