SCADA-Based Health-Constrained Active Power Regulation of Aging Wind Turbines

Wind turbines of the same design develop different thermal and mechanical margins during long-term operation, requiring active power allocation to consider individual health conditions. This study proposes a supervisory control and data acquisition (SCADA)-based method for determining a health-constrained power boundary for aging wind turbines. A normal behavior model establishes references for key thermomechanical variables matched to current operating conditions. Standardized residuals are combined using the Mahalanobis distance and a trimmed temporal window to construct a turbine-specific health index and reduce the influence of isolated extremes. Short-horizon state prediction evaluates candidate power levels. The maximum feasible candidate power is determined separately under statistical, temperature, and mechanical constraints, and their minimum defines the health-constrained power boundary. Combined with available power and local operating limits, this boundary enters the wind farm power allocator as a turbine-level inequality constraint. Field data from 30 turbines rated at 1.5 MW are used to evaluate representative events and farm-wide performance. Turbines show different operating margins under similar wind conditions. In the evaluation sample, the health-constrained operation group had a shorter non-generating duration and a higher energy yield normalized by wind speed frequency; group differences remained statistically uncertain.