基于深度学习特征提取和粒子群支持向量机状态识别的齿轮智能故障诊断

针对齿轮故障诊断问题,利用数理统计特征提取方法、深度学习神经网络、粒子群算法和支持向量机等技术,提出了一种基于深度学习特征提取和粒子群支持向量机状态识别相结合的智能诊断模型。该模型利用深度学习自适应提取的频谱特征与数理统计方法提取的时域特征相结合组成联合特征向量,然后利用粒子群支持向量机对联合特征向量进行故障诊断。该模型在对多级齿轮传动系统试验台的故障诊断中实现了中速轴大齿轮不同故障类型的可靠识别,获得了满意的诊断结果。应用结果也验证了基于深度学习自适应提取频谱特征的有效性。

Intelligent Fault Diagnosis for Gears Based on Deep Learning Feature Extraction and Particle Swarm Optimization SVM State Identification

For the fault diagnosis of gears, using statistical methods for feature extraction, deep learning neural network (DNN), particle swarm optimization algorithm and SVM technology, a novel intelligent diagnosis model was proposed, which combined the deep learning feature extraction and particle swarm optimization SVM state identification. The model combined the frequency domain features which were extracted by deep learning adaptively and the time domain features which were extracted by mathematical statistics method to form a combined feature vectors, then using particle swarm optimization SVM to diagnose the feature vectors. This model provides reliable identification of intermediate shaft gears of different types of faults in the fault diagnoses of gear transmission system test bench, and the satisfactory diagnostic results are obtained. The application results also verify the effectiveness of the adaptive extraction of spectral features based on deep learning.