SCADA-Based Real-Time Fault Diagnosis and Operational State Monitoring of a Biomass Power Plant Using Machine Learning

Guney E., DURSUN M.

13th International Conference on Electrical and Electronics Engineering, ICEEE 2026, Antalya, Türkiye, 27 - 29 Nisan 2026, ss.561-566, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/iceee69936.2026.11598051
  • Basıldığı Şehir: Antalya
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.561-566
  • Anahtar Kelimeler: anomaly detection, biomass power plant, condition monitoring, Isolation Forest, Random Forest, SCADA
  • Gazi Üniversitesi Adresli: Evet

Özet

This paper presents a machine learning-based framework for real-time fault diagnosis and operational state monitoring of a biomass power plant using SCADA data. Unlike conventional monitoring approaches that rely on static thresholds or single-stage prediction models, the proposed framework integrates unsupervised anomaly detection with a domaininformed diagnostic classification layer, enabling interpretable and actionable system-level insights. A dataset comprising 15 process parameters (spanning electrical output, steam cycle conditions, grate temperatures, and flue gas quality indicators) recorded over a 24-hour operating period at one-second resolution is resampled to one-minute intervals, yielding 1,413 observations. An Isolation Forest algorithm is employed for unsupervised anomaly detection, identifying 71 anomalous minutes (5.0%). A Random Forest classifier subsequently assigns each observation to one of four operational states: Optimal, Normal, Risky, or Critical. The proposed framework achieves a classification accuracy of 95.76% and a weighted F1-score of 0.9518. Feature importance analysis identifies power output, steam pressure, and front grate temperature as the most discriminative diagnostic parameters. The results demonstrate that the framework provides actionable real-time insight for condition monitoring and early fault intervention in biomass energy systems, supporting sustainable plant operation and alignment with predictive maintenance paradigms.