Development of a Fuzzy Logic Health Index for Generator Stator Insulation Condition Assessment

Muhamad Arief Afandy (Universitas Tidar)
Agung Trihasto (Universitas Tidar)
Risky Via Yuliantari (Universitas Tidar)

Aviation Electronics, Information Technology, Telecommunications, Electricals, Controls (AVITEC)
Vol 8, No 2 (2026): August

Stator insulation breakdown is the primary cause of generator failures, driven by cumulative thermal, electrical, and mechanical stresses. For generator Unit 2 at PLTU Palabuhan Ratu Coal-Fired Power Plant, the Mamdani fuzzy logic model provides an automated, rule-based diagnostic assessment utilizing tip-up tan delta, AC leakage current, and partial discharge measurements. This model predicts the Health Index of generator phases using historical test data 2019, 2023, and 2025. It successfully classifies phases R, S, and T into the "Poor" category. Model validation yields an error of 0.6% and passes monotonicity tests, sensitivity analysis, and boundary testing. Partial discharge is the most dominant parameter with a potential health index reduction of 44.3 points. The main novelty is integrating the three dielectric parameters into a Mamdani fuzzy model adapted to MELCO and IEC standards for generators in tropical environments.