Global Sensitivity Analysis of Islanded Microgrids with Correlated Random Variables
To accurately quantify the impact of correlated random variables on the operational state of an islanded microgrid with droop control,a conditional sampling method for random variables is proposed.The global sensitivity indices are computed using a Monte Carlo algorithm based on Latin hypercube sampling,considering the correlation of random variables to enhance the accuracy and efficiency of variance-based global sensitivity analysis.A case study is set up in a 38-node microgrid with droop control and uncertain sources,with traditional Monte Carlo simulation providing accuracy reference to validate the proposed method's performance.The case study results indicate that the correlation of random variables significantly affects the global sensitivity analysis results of the microgrid,and the proposed method accurately accounts for this correlation.This demonstrates the importance of precisely considering random variable correlation in global sensitivity analysis of microgrids and the necessity of implementing the proposed method for such analyses.