An application of support vector machine to detect anomalies in time series data

Maintaining the equipment in a good performance is mandatory for business. It needs to reduce the production cost, minimize downtime, maintain product quality, safety and reduce risks. The maintenance department shall predict whether the equipment may fail in near future to minimize downtime. Parame...

Full description

Saved in:
Bibliographic Details
Main Authors: Satriawan, Yoga S., Saputro, Joko Slamet
Format: Conference Proceeding
Language:English
Subjects:
Anomalies
Computer simulation
Downtime
Machine learning
Product safety
Production costs
Support vector machines
Time series
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Maintaining the equipment in a good performance is mandatory for business. It needs to reduce the production cost, minimize downtime, maintain product quality, safety and reduce risks. The maintenance department shall predict whether the equipment may fail in near future to minimize downtime. Parameters data which we used in this paper are temperature and vibration. We used anomaly detection to predict failure of the equipment. Anomaly detection is one of the methods which can be used to predict the failure. It detects anomalies in time series data with evenly time-spaced numerical values. Paper proposes to use Support Vector Machine method to detect anomalies. This testing was done by using Azure Machine Learning Development Studio simulation result shows that support vector machine method can detect anomalies with accuracy 0.906.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0000659