Semi-supervised Learning with Deep Generative Models for Asset Failure Prediction
It addresses asset failure prediction for industries with sparse labeled data, representing an incremental advance in semi-supervised learning for prognostics.
This paper tackles the problem of predicting asset failures with limited labeled health status data by proposing a semi-supervised learning approach using deep generative models and non-linear embedding, achieving significant improvement in remaining useful life estimation even with only 1% labeled data.
This work presents a novel semi-supervised learning approach for data-driven modeling of asset failures when health status is only partially known in historical data. We combine a generative model parameterized by deep neural networks with non-linear embedding technique. It allows us to build prognostic models with the limited amount of health status information for the precise prediction of future asset reliability. The proposed method is evaluated on a publicly available dataset for remaining useful life (RUL) estimation, which shows significant improvement even when a fraction of the data with known health status is as sparse as 1% of the total. Our study suggests that the non-linear embedding based on a deep generative model can efficiently regularize a complex model with deep architectures while achieving high prediction accuracy that is far less sensitive to the availability of health status information.