• • Environmental and electrical parameter identification errors are controlled below 1%, while controller parameter errors remain below 3%, establishing a quantitative benchmark for EMT model calibration in PV grid-connected systems where unified optimization typically yields controller parameter deviations exceeding 10%.
• • The Sobol global sensitivity analysis and dynamic response feature clustering with DBSCAN and inter-cluster mean difference index effectively screen dominant parameters, compressing the search space and improving computational efficiency by eliminating non-identifiable parameters that contribute negligible variance to system response.
• • The IQDBO algorithm, incorporating quantum angle encoding and a stagnation-based perturbation mechanism, outperforms PSO and GWO in both identification accuracy and convergence stability, reducing the risk of premature convergence in high-dimensional parameter spaces typical of multi-loop PV inverter controllers.
• • The two-stage framework prevents cross-interference between electrical and controller parameters during identification, directly addressing the weak identifiability of controller parameters that arises when all parameters are optimized simultaneously under a single fitness function.