Predicting Remaining Useful Life using Time Series Embeddings based on Recurrent Neural Networks

We consider the problem of estimating the remaining useful life (RUL) of a system or a machine from sensor data. Many approaches for RUL estimation based on sensor data make assumptions about how machines degrade. Additionally, sensor data from machines is noisy and o‰en su‚ers from missing values in many practical seŠings. We propose Embed-RUL: a novel approach for RUL estimation from sensor data that does not rely on any degradation-trend assumptions, is robust to noise, and handles missing values. EmbedRUL utilizes a sequence-to-sequence model based on Recurrent Neural Networks (RNNs) to generate embeddings for multivariate time series subsequences. Œe embeddings for normal and degraded machines tend to be di‚erent, and are therefore found to be useful for RUL estimation. We show that the embeddings capture the overall paŠern in the time series while €ltering out the noise, so that the embeddings of two machines with similar operational behavior are close to each other, even when their sensor readings have signi€cant and varying levels of noise content. We perform experiments on publicly available turbofan engine dataset and a proprietary real-world dataset, and demonstrate that Embed-RUL outperforms the previously reported [24] state-of-the-art on several metrics.