自适应参数高斯卷积核神经网络的滚动轴承故障诊断方法

• Format: Patent
• Language: Chinese
• Published: 27.06.2025
• Subjects: MEASURING; PHYSICS; TESTING; TESTING STATIC OR DYNAMIC BALANCE OF MACHINES ORSTRUCTURES; TESTING STRUCTURES OR APPARATUS NOT OTHERWISE PROVIDED FOR

本发明涉及自适应参数高斯卷积核神经网络的滚动轴承故障诊断方法，属于滚动轴承故障诊断类别领域，包括如下步骤：利用加速度计获取旋转设备中滚动轴承的振动信号，进行切片、求包络向量，并制作为数据集；将数据集的样本进行峭度值和平均偏度值计算，设计三种自适应参数的与轴承振动冲击信息特性相关的高斯卷积核；利用获得的三种自适应参数高斯卷积核设计自适应参数高斯卷积层进行特征提取，将得到的三个特征向量进行拼接得到新特征向量；将所获得的新特征向量输入到具有交叉熵损失函数的Softmax分类器中，并建立故障特征和故障类别之间的映射关系来实现故障诊断，提高故障诊断模型的训练效率和分类效果，增强模型的可靠性和可解释性。 The invention relates to a rolling bearing fault diagnosis method based on an adaptive parameter Gaussian convolution kernel neural network, and belongs to the field of rolling bearing fault diagnosis categories, and the method comprises the following steps: obtaining a vibration signal of a rolling bearing in rotating equipment through an accelerometer, carrying out the slicing, solving an envelope vector, and making a data set; carrying out kurtosis value and average skewness value calculation on samples of the data set, and designing Gaussian convolution kernels, related to bearing vibration impact information characteristics, of three adaptive parameters; designing an adaptive parameter Gaussian convolution lay