Additive Model Decision Tree-Based Adaptive Wide-Area Damping Controller Design

An adaptive wide-area damping controller (WADC) scheme based on gain scheduling (GS) technique and decision tree (DT) approach is presented. For the predefined operating scenarios, several robust controllers are designed, so that each controller performs well within a certain range of load/generation variation and contingencies. Considering the best acting controller in each operating condition as the target of classification, several DTs are trained using stage-wise additive modeling by multiclass exponential loss function (SAMME). The weighted combination of DTs is referred to as additive model DT (AMDT) that is used to interpolate the controller coefficients from phasor measurement unit (PMU) signals. Afterward, the combined controller is applied to damp the interarea oscillations. Case studies on the IEEE 39-bus 10-machine and 68-bus 16-machine test systems demonstrate the effectiveness of the proposed controller in a wide range of operating conditions as well as its robustness in the presence of uncertainty in the predicted scenarios.