基于IWT_SE与GA_SVM的齿轮磨损检测
|
摘要
为了实现齿轮运行过程中的磨损程度准确识别,提出了基于改进小波阈值样本熵(IWT_SE)与遗传算法优化支持向量机(GA_SVM)的齿轮磨损程度检测方法。首先,对齿轮振动信号进行改进小波阈值降噪;其次,计算降噪后信号的样本熵,组成特征向量;最后,将特征向量输入基于GA_SVM建立的分类器进行故障识别分类。通过齿轮实验数据分析了算法中的参数选取问题;将该方法用于齿轮实验数据,并与传统小波阈值函数样本熵分别与BPNN,PNN,SVM,PSO_SVM相结合的方法进行对比分析,结果表明,IWT_SE与GA-SVM相结合时识别准确率最高,达95%,证明文中所提方法对齿轮磨损程度识别具有一定实际应用价值。
In order to monitor and diagnose degrees of wear during gear operation more accurately, a method was proposed based on improved wavelet threshold_sample entropy(IWT_SE) and genetic algorithm_support vector machines(GA_SVM). Firstly, the original vibration signal was denoised by the method of wavelet improved threshold function. Then, the sample entropies of signals after de-nosing were extracted as fault feature vectors. Finally, the feature vector was input to the GA_ SVM for classification of degrees of gear wear. The parameters of algorithm was selected using gear experimental data. The comparison is made with traditional threshold function method combined with neural network of BPNN, PNN, SVM and PSO_SVM. The experimental results show that the proposed method can recognize and classify the degrees of gear wear with high accuracy which is 95%, which proves that the proposed method has practical value in some degree.
In order to monitor and diagnose degrees of wear during gear operation more accurately, a method was proposed based on improved wavelet threshold_sample entropy(IWT_SE) and genetic algorithm_support vector machines(GA_SVM). Firstly, the original vibration signal was denoised by the method of wavelet improved threshold function. Then, the sample entropies of signals after de-nosing were extracted as fault feature vectors. Finally, the feature vector was input to the GA_ SVM for classification of degrees of gear wear. The parameters of algorithm was selected using gear experimental data. The comparison is made with traditional threshold function method combined with neural network of BPNN, PNN, SVM and PSO_SVM. The experimental results show that the proposed method can recognize and classify the degrees of gear wear with high accuracy which is 95%, which proves that the proposed method has practical value in some degree.
引文
[1] 桂勇,韩勤锴,李铮,等.风机行星齿轮系统齿轮裂纹故障诊断[J].振动、测试与诊断,2016,36(1):169-205.
[2] 宋萌萌,肖顺根,陈肇祥.基于提升小波变换与EEMD 的神经网络齿轮故障诊断方法[J].组合机床与自动化加工技术,2017(9):93-98.
[3] 张强,刘志恒,王海舰,等.基于BP神经网络的截齿磨损程度在线监测[J].中国机械工程,2017,28(9):1062-1068.
[4] 李力,倪松松.基于改进小波去噪预处理和EEMD的采煤机齿轮箱故障诊断[J].中南大学学报(自然科学版),2016,47(10):3394-3400.
[5] 何巍,袁亮,章翔峰.改进小波去噪-Teager算子的齿轮微弱故障提取方法[J].振动、测试与诊断,2018,38(1):155-213.
[6] Xiaoan Yan,Minping Jia.A novel optimized SVM classification algorithm with multi-domain feature and its application to fault diagnosis of rolling bearing[J].Neurocomputing,2018(6):1-18.
[7] 柏林,刘小峰,秦树人.小波-形态-EMD综合分析法及其应用[J].振动与冲击,2008,27(5):1-4,35.
[8] Richman J S,Moorman J R.Physiological time-series analysis using approximate entropy and sample entropy[J].American Journal of Physiological Heat and Circulatory Physiology,2000,278(6):2039-2049.
[9] 张学清,梁军,张熙,等.基于样本熵和极端学习机的超短期风电功率组合预测模型[J].中国电机工程学报,2013,33(25):33-40.
[10] 方一鸣,胡春洋,刘乐,等.基于主动学习GA_SVM分类器的连铸漏钢预报[J].中国机械工程,2016,27(12):1609-1614.
[11] Ke Li,Lingyu Wang,JingJing Wu,et al.Using GA_SVM for defect inspection of flip chips based on vibration signals[J].Microelectronics Reliability,2018,81:159-166.
[12] 郑小霞,周国旺,任浩瀚,等.基于变分模态分解和排列熵的滚动轴承故障诊断[J].振动与冲击,2017,36(22):22-27.
[13] 陶剑锋.历程图中的弱目标轨迹增强方法[J].西安交通大学学报,2018,52(2):76-81,117.
[2] 宋萌萌,肖顺根,陈肇祥.基于提升小波变换与EEMD 的神经网络齿轮故障诊断方法[J].组合机床与自动化加工技术,2017(9):93-98.
[3] 张强,刘志恒,王海舰,等.基于BP神经网络的截齿磨损程度在线监测[J].中国机械工程,2017,28(9):1062-1068.
[4] 李力,倪松松.基于改进小波去噪预处理和EEMD的采煤机齿轮箱故障诊断[J].中南大学学报(自然科学版),2016,47(10):3394-3400.
[5] 何巍,袁亮,章翔峰.改进小波去噪-Teager算子的齿轮微弱故障提取方法[J].振动、测试与诊断,2018,38(1):155-213.
[6] Xiaoan Yan,Minping Jia.A novel optimized SVM classification algorithm with multi-domain feature and its application to fault diagnosis of rolling bearing[J].Neurocomputing,2018(6):1-18.
[7] 柏林,刘小峰,秦树人.小波-形态-EMD综合分析法及其应用[J].振动与冲击,2008,27(5):1-4,35.
[8] Richman J S,Moorman J R.Physiological time-series analysis using approximate entropy and sample entropy[J].American Journal of Physiological Heat and Circulatory Physiology,2000,278(6):2039-2049.
[9] 张学清,梁军,张熙,等.基于样本熵和极端学习机的超短期风电功率组合预测模型[J].中国电机工程学报,2013,33(25):33-40.
[10] 方一鸣,胡春洋,刘乐,等.基于主动学习GA_SVM分类器的连铸漏钢预报[J].中国机械工程,2016,27(12):1609-1614.
[11] Ke Li,Lingyu Wang,JingJing Wu,et al.Using GA_SVM for defect inspection of flip chips based on vibration signals[J].Microelectronics Reliability,2018,81:159-166.
[12] 郑小霞,周国旺,任浩瀚,等.基于变分模态分解和排列熵的滚动轴承故障诊断[J].振动与冲击,2017,36(22):22-27.
[13] 陶剑锋.历程图中的弱目标轨迹增强方法[J].西安交通大学学报,2018,52(2):76-81,117.