Participant-level anomaly detection for generation and load data using dual-side LSTMs
Qianya He, Qi Liu, Xueliang Gong, Jiaxun Liu, Liming Yao, Tianze Yu, Tong Zhao · 2026
Read Participant-level anomaly detection for generation and load data using dual-side LSTMs by Qianya He, Qi Liu, Xueliang Gong, Jiaxun Liu, Liming Yao, Tianze Yu, Tong Zhao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13500582.txt.
Description
Qianya He and six co-authors present an anomaly detector for power market data in a 2026 paper. Electricity markets settle accounts from meter data on generation and usage. Bad readings are rare and varied, and the user side holds tens of thousands of participants who change daily, so pinpointing who went wrong and when is hard.
Their method feeds each day's readings, totals, and rhythm patterns into a memory-based neural network built for sequences. Each participant gets a learned fingerprint, and a padding trick handles the shifting head count without polluting training. A two-step rule then screens by error size and confirms with model confidence. On one month of scrubbed provincial grid data, the system scored perfectly on their test measures, far above simple threshold tools. The authors offer it as a way to keep settlement data trustworthy.
Catalog record
| CREATOR | Qianya He, Qi Liu, Xueliang Gong, Jiaxun Liu, Liming Yao, Tianze Yu, Tong Zhao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |