A Study on Failure Prediction Using Time Series Data of Hydraulic Excavator

Since unexpected machine failures are huge losses for users, maintenance activities essential. If the can be predicted in advance using a supervised learning, machines maintained before they break down and some prevented. However, although large number of failure data required to predict rarely occur actual field. In this study, we propose detect hydraulic excavator an autoencoder, which is unsupervised learning. By autoencoder model normal state data, advance. This paper shows results evaluating predictions LSTM (Long Short-Term Memory) excavators.