大数据样本与半监督环境下基于生成对抗网络的故障诊断

在大故障样本条件下,提出一种基于生成对抗网络模型的故障诊断方法研究.构建生成对抗网络模型,保证模型判别器输出数据的总体分布与原始故障集趋同,并基于空间测量工具对梯度函数进行优化,降低损失;采用故障集图像转换方式实现对原始信号的降维处理,利用判别器的神经网络结构训练输入数据,并提取出机械故障数据集中的故障特征点.实验结果表明,提出方法具有良好的分类诊断性能,故障诊断精度能够达到９９．４５％.

Large Data Samples and Fault Diagnosis Based on Generation Countermeasure Network in Semi-supervised Environment

Under the condition of large fault samples,a fault diagnosis method based on the generation countermeasure network model is proposed．The model of generating countermeasure network is constructed to ensure that the overall distribution of output data of model discriminator is similar to the original fault set, and the gradient function is lost based on the spatial measurement tool; the method of fault set image conversion is used to realize the dimensionality reduction of the original signal,and the neural network structure of discriminator is used to train the input data and extract the fault features of the mechanical fault data set．The experimental results show that the proposed method has good performance of classification and diagnosis, and the accuracy of fault diagnosis can reach ９９．４５％．