自编码器（Autoencoder, AE）因能从正常运行数据中学习非线性表示而被广泛用于工业过程故障检测，但常规AE过度依赖重构误差且易因过泛化（overgeneralization）漏检弱故障，同时对时序演化及工况变化建模不足会降低其对动态故障的敏感性。为此 自编码器 ...

A variational autoencoder (VAE) is one of several generative models that use deep learning to generate new content, detect anomalies and remove noise. Suited for generating synthetic time series data ...