Predictive Maintenance for Automatic Weather Station (AWS) Based on Anomaly Detection Using Autoencoder: A Literature Review

Muhammad Hamdan Afif, Daffa Aly Meganendra

Journal of Computation Physics and Earth Science (JoCPES) · 01 October 2023

In one sentence. Autoencoder-based anomaly detection is proposed to flag faulty automatic weather station data for predictive maintenance, a low-novelty study with no quantitative results given.

Abstract

Automatic Weather Station or AWS is an instrument for measuring weather parameters automatically. The results of measuring weather parameters are very useful in the fields of meteorology and climatology, such as weather prediction, aviation and climate change. Especially in Indonesia, the Meteorology, Climatology and Geophysics Agency or BMKG has main tasks and functions in this field. Currently, data with accurate results is needed to produce accurate weather and climate predictions. However, sometimes there are anomalies in the data caused by AWS damage, resulting in inaccurate data. This will have an impact on modeling results in the fields of meteorology and climatology, where the modeling results are less precise. To overcome this problem, predictive maintenance is needed to avoid data errors in AWS operations. This research aims to build predictive maintenance at an Automatic Weather Station Based on Anomaly Detection using a Machine Learning Autoencoder. The anomaly data can be detected by machine learning autoencoders for monitoring AWS performance and conditions, that methodology applied in this study for build predictive maintenance in AWS. Finally, the expectation of this research is to make accurate predictive maintenance on AWS so perhaps that can reduce maintenance costs and increase the lifespan of the instrument before it breaks.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.