国外文献 > 1.轴承 > 7. 智能方法与计算 > 神经网络
关键词搜索：
| 标 题： |
A fault diagnosis method combined with compound multiscale permutation entropy and particle swarm optimization-support vector machine for roller bearings diagnosis
结合复合多尺度置换熵与粒子群优化支持向量机的滚子轴承故障诊断方法 |
| 作 者： |
Xu, Fan
Tse, Peter Wai Tat
Fang, Yan-Jun
Liang, Jia-Qi |
| 关键词： |
Compound multiscale permutation entropy [复合多尺度排列熵]
support vector machine [ 支持向量机]
particle swarm optimization [ 粒子群优化]
roller bearings [ 滚子轴承]
fault recognition [ 故障识别] |
| 出版年： |
2019 |
| 期刊名： |
Proceedings of the Institution of Mechanical Engineers, Part J. Journal of engineering tribology
机械工程师学会会刊，J部分。工程摩擦学杂志 |
| 年卷期： |
2019, vol.233, no.4 |
| 摘 要： |
A method based on compound multiscale permutation entropy, support vector machine, and particle swarm optimization for roller bearings fault diagnosis was presented in this study. Firstly, the roller bearings vibration signals under different conditions were decomposed into permutation entropy values by the multiscale permutation entropy and compound multiscale permutation entropy methods. The compound multiscale permutation entropy model combined the different graining sequence information under each scale factor. The average value of each scale factor was regarded as the final entropy value in the compound multiscale permutation entropy model. The compound multiscale permutation entropy model suppressed the shortcomings of poor stability caused by the length of the original signals in the multiscale permutation entropy model. Validity and accuracy are considered in the numerical experiments, and then compared with the computational efficiency of the multiscale permutation entropy method. Secondly, the entropy values of the multiscale permutation entropy/compound multiscale permutation entropy under different scales are regarded as the input of the particle swarm optimization-support vector machine models for fulfilling the fault identification, the classification accuracy is used to verify the effectiveness of the multiscale permutation entropy/compound multiscale permutation entropy with particle swarm optimization-support vector machine. Finally, the experimental results show that the classification accuracy of the compound multiscale permutation entropy model is higher than that of the multiscale permutation entropy.
本研究提出了一种基于复合多尺度置换熵、支持向量机和粒子群优化算法的滚子轴承故障诊断方法。首先，采用多尺度置换熵和复合多尺度置换熵方法对不同工况下滚子轴承的振动信号进行分解，得到置换熵值。复合多尺度置换熵模型结合了各尺度因子下的不同粒度序列信息，将各尺度因子的平均值作为复合多尺度置换熵模型的最终熵值。该模型有效克服了多尺度置换熵方法因原始信号长度过长而引起的稳定性差问题。通过数值实验验证了该方法的有效性和准确性，并与多尺度置换熵方法的计算效率进行了比较。其次，将不同尺度下的多尺度排列熵/复合多尺度排列熵的熵值作为粒子群优化支持向量机模型的输入，用于实现故障识别； 通过分类准确率验证了基于粒子群优化和支持向量机的多尺度排列熵/复合多尺度排列熵方法的有效性。最后，实验结果表明，复合多尺度排列熵模型的分类准确率高于多尺度排列熵模型。 |
| 中文内容为网络AI翻译，专业术语未必准确，仅供参考。 |
京公网安备 11010202009788