Sensitivity-based analyses of WECC load model parameter identification accuracy under various voltage dip characteristics

Abstract

This thesis presents a sensitivity-based parameter identification framework for the WECC Composite Load Model under varying voltage sag conditions. The developed MATLAB implementation of the WECC model, including simplified DER_A, is first validated through frame-by-frame comparison with the standard WECC reference model, showing excellent agreement in both active and reactive power responses. The original model consists of 56 parameters for combined active and reactive power estimation. Through the application of a windowed normalized RMS-based sensitivity analysis, this parameter set is reduced to 38 parameters, representing an approximate reduction of 32.1% and significantly reducing the search space. Particle Swarm Optimization is then applied for parameter estimation under multiple disturbance scenarios. The optimized results closely match the reference responses, demonstrating that reduced-parameter estimation maintains high accuracy while improving computational efficiency. The proposed approach enables more robust and practical parameter identification for large-scale power system studies.