SPIL Fault Detection Model

Objective

Developing the fault detection system to earlier identify anomaly and reduce false alarm.

Procedure

1. Researched on recent studies discussing the related problems about anomaly detection using RNN-based structure.
2. Modified the structure of related works in order to fit our dataset and generate the proper results.
3. Implemented RNN-based time series fault detection model and calculated anomaly score by the difference between predict and actual value.

Results

Due to closing to graduating, I only participated in the proof of concept stage of this project. However, We already implemented the fault detection system locally in Pytorch using the relatively small data (with four sensors) to obtain acceptable results.