This paper presents the application of improved SFLA to solve the unit commitment problem including operational and power flow constraints to ensure system security. The SFLA combines both memetic and social behavior of genetic PSO techniques. It includes both deterministic and random techniques for effective use of the search space. The introduction of cognitive component speeds up the local search process by widening the search space. In this proposed model the optimal power flow is carried out for every frog to obtain the minimal operating cost generator scheduling while satisfying the operational and network constraints. Gaussian random noises are introduced for hourly load variations. The proposed algorithm has been applied for IEEE 6 bus and IEEE 30 bus test systems for 24 h. The proposed algorithm is simulated using MATLAB’11 software and the results are quite encouraging because of its excellent convergence. © Springer India 2015.