Accurate fault section diagnosis of power systems with a binary adaptive quadratic interpolation learning differential evolution

Fault section diagnosis (FSD) is essential for ensuring the effective operation of power systems. To determine the faulty sections accurately, we proposed an improved binary adaptive quadratic interpolation learning differential evolution called BAQILDE for solving the FSD problem. By comparing the received warning data with the anticipated states of circuit breakers and protective relays, an analytical 0-1 integer programming function is established. To tackle the resultant function accurately, the population in BAQILDE is directly encoded in binary instead of floating-point to facilitate the solving convenience. Besides, three enhanced strategies including adaptive mutation operator, time-varying crossover rate, and dual transformation operator are developed to equilibrate the population diversity and convergence well to strengthen BAQILDE. To evaluate BAQILDE's performance, four test systems were used for verification, including 4-substation power system, IEEE 118 bus system, and two actual failures that occurred in Guangzhou and Jilin power grids, China. The results show that BAQILDE can diagnose various failures within 0.12 s with 100 % success rate and 0 diagnosis error, consuming an average of 32.21 function evaluation times. It outperformed other well-known peer algorithms in success rate, diagnosis error, robustness, convergence, and statistical analysis, which demonstrates its strong competitiveness in solving the FSD problem.