Adaptive Type‐3 Fuzzy PI Control for Robust DC‐Bus Regulation in Grid‐Connected PEM Fuel‐Cell‐Assisted Hybrid Microgrids

ABSTRACT

This study proposes a Type‐3 fuzzy proportional–integral (T3‐FPI) controller for robust DC‐bus voltage regulation in a grid‐connected hybrid microgrid integrating photovoltaic and wind generation, a lithium‐ion battery, a proton exchange membrane fuel cell (PEMFC), and the utility grid. The controller performs online fuzzy gain scheduling of and according to the DC‐bus voltage error and its rate of change, while a rule‐based energy‐management strategy coordinates battery, grid, and PEMFC operation. For a fair assessment, T3‐FPI is compared with a reproducibly retuned PI, a Type‐1 fuzzy PI (T1‐FPI), and an interval Type‐2 fuzzy PI (T2‐FPI) under identical operating and tuning conditions. In the controller benchmark, the DC‐bus voltage returns to and remains in the vicinity of the 500‐V reference following the imposed load disturbance, confirming stable post‐disturbance regulation. Relative to the retuned PI, T3‐FPI reduces the post‐step peak DC‐bus voltage deviation by approximately 10.40% and 10.93% in Scenarios 1 and 2, respectively, while the corresponding post‐step RMSE reductions are approximately 7.74% and 7.54%. The extended severe‐uncertainty assessment further shows that T3‐FPI does not universally outperform T1‐FPI and T2‐FPI in every scalar voltage‐error metric. Instead, it maintains closely comparable voltage‐regulation performance while exhibiting smoother bounded adaptive‐gain trajectories. In a representative severe PV‐loss stress case, the total variation of decreases from 9.867 and 9.868 for T1‐FPI and T2‐FPI to 9.491 for T3‐FPI, while the corresponding total variation decreases from 10.812 and 10.806 to 10.498. A scan of the complete bounded T3‐FPI gain domain confirms local stability for all 961 evaluated – combinations, with a worst‐case maximum real eigenvalue of . The full‐microgrid results additionally demonstrate coordinated power sharing, conditional PEMFC support, balanced three‐phase operation, and adequate converter‐capacity margin under dynamic loading. These results position the proposed T3‐FPI primarily as a bounded and smoother adaptive gain‐scheduling solution rather than as a controller that is universally superior to lower‐order fuzzy alternatives in every individual voltage‐error metric.