论文部分内容阅读
通过对齿轮系统在不同的运转状态下不同的故障类型进行试验测试分析,获取了有关的测试信号,对振动特征信号进行了小波阈值去噪,采用离散小波变换(DWT)对去噪后的信号进行8层分解处理,对各层的小波系数进行了小波重构,得到8层细节信号和1层近似信号,并计算了各层信号的能量,得到了信号的能量分布特征.在此基础上把各层信号特征作为神经网络的输入,进行了网络的研究、分析处理和故障分类,并对小波神经网络方法与单独采用神经网络方法的故障诊断结果进行了比较评价.研究表明,去噪处理后的效果比没有去噪的信号特征更加明显,而采用小波神经网络诊断方法,对于齿轮无故障、齿根裂纹故障、分度圆裂纹故障和齿面磨损故障能够进行很好地区分与诊断,其诊断成功率均在95%以上,可对实际工程工作的齿轮系统进行故障诊断.
Through the test and analysis of the different fault types of gear system under different operating conditions, the relevant test signals are obtained, the wavelet threshold is used to denoise the vibration characteristic signals, and the signal of the denoised signals is decomposed by discrete wavelet transform (DWT) Then the wavelet coefficients of each layer are reconstructed by wavelet to get 8 layers of detail signal and 1 layer of approximate signal, and calculate the signal energy of each layer, and obtain the signal energy distribution characteristics. On this basis The signal characteristics of each layer as the input of neural network, the network research, analysis and fault classification, and the wavelet neural network method and neural network alone fault diagnosis results were compared and evaluated.Research shows that the noise reduction The effect is more obvious than the no-denoising signal characteristic. The method of wavelet neural network diagnosis can distinguish and diagnose the gear without fault, the root crack, the index circle crack and the tooth wear fault well, The diagnostic success rates are above 95%, the actual engineering work of the gear system for fault diagnosis.