This paper addresses a novel hunger games search (HGS) based on specular reflection-based learning (SRL) and dynamic quasi-opposition-based learning (DQOL), named HGS_RQ, for improving the optimization performance of the classical HGS while dealing with load frequency control task. By these learnings, a fitter solution can be generated whether by SRL or DQOL and therefore, the quality of the best solution can be refined. The effectiveness of the proposed HGS_RQ is demonstrated and validated on two-area interconnected power system with considering nonlinearity effect of governor dead band. Additional supplementary controller is proposed to reinforce frequency regulation through solid oxide fuel cell. The objective function is adapted to minimize the integral time absolute error in frequency deviations and tie line power. The efficacy and superiority are affirmed by the comparisons with some of prominent recent methods. It can be noted that the adequate response is proved, since the maximum frequency deviation is 0.088 Hz, the settling time is about 2 s, and the steady-state frequency change is zero in the two areas. On the other hand, there is a significant reduction in tie line power transient response with maximum deviation of 1.318% for the studied cases. Furthermore, the statistical measures and analysis of variance test are analyzed to exhibit the superior performance of the HGS_RQ in terms of accuracy and reliability.