AbstractThere is a change in the traditional power system structure as a result of the increased incorporation of microgrids (MGs) into the grid. Multi‐area MGs will emerge as a result, and issues related to them will need to be addressed. Load frequency control (LFC) is a challenge in such structures, which are more complicated due to variations in demand and the stochastic characteristics of renewable energy sources. This paper presents a cascade fuzzy active disturbance rejection control technique to deal with the LFC problem. In order to tune different parameters of controllers, a newly developed heuristic algorithm called the Gazelle optimization algorithm (GOA) is also employed. Moreover, due to the fact that multi‐area MGs are regarded as cyber‐physical systems (CPSs), a relatively new concern for LFC problems is their resilience to cyberattacks such as false data injection (FDI) and denial of service (DoS) attacks. Therefore, this research also presents a novel machine learning approach called parallel attack resilience detection system (PARDS) to deal with the LFC problem in the presence of cyberattacks. The efficiency of the proposed strategy is investigated under different scenarios, such as non‐linearities in the power system or server cyberattacks.