- Anomaly detection
Time series abnormal waveform detection technology*
Detect anomalies, learning only from normal time series data.
- Jointly learn anomaly detection models and unique waveform fragments constituting time series data.
- Currently being applied in infrastructure equipment, such as elevators and switchgears at transformer substations.
- One-Class Learning Time-series Shapelets (OCLTS)
Applications
- Detect anomalies in transformer substation switchgears.
- Detect anomalies using elevator brake signals.
- Can also be applied in detecting anomalies using operational data from various infrastructure equipment.
Benchmarks, strengths, and track record
- World’s first anomaly detection technology that learns waveform fragments using only normal time series data
- In performance verification using benchmark data, confirmed a 9% increase in accuracy compared to conventional technologies.
Inquiries
Please include the title “Toshiba AI Technology Catalog: Time series abnormal waveform detection technology” or the URL in the inquiry text.
Please note that because this technology is currently the subject of R&D activities, immediate responses to inquiries may not be possible.
References:
- A. Yamaguchi et al, One-Class Learning Time-Series Shapelets, IEEE Big Data 2018.
- A. Yamaguchi et al, OCLTS: One-Class Learning Time-Series Shapelets, IJDAT 2019.
- Develop highly explainable AI that detects anomalies based on learning only from normal waveform data (in Japanese)
