Chaos quasi-opposition sea-horse based modified new tilt controller designed for multi-area deregulated AGC using deep learning against cyber-attacks

This article highlights the intermittent character of RESs and reveals how deep learning techniques reduce the frequency deviation in automatic generation control (AGC) of the deregulated Modern Power System (MPS). With increasing MPS complexity comes more difficulty controlling frequency deviation and frequency security. The stability of an MPS is significantly affected by the robustness of the controller. The proposed MPS consists of thermal, hydro, and nuclear power plants, which include generation rate constraint (GRC) and governor dead band (GRB). The incorporation of wind and solar photovoltaic systems, together with their intermittent nature, has been considered to provide a more reliable control mechanism for enhanced frequency regulation in multi-area deregulated environments. The PID, FOPID, and TID basic AGC controllers are insufficient to offer a plant with optimal performance because of MPS hybridization. So, a new modified Cascaded tilted-FO derivative with filter (CPDμF−TI) controller has been suggested, and its performance has been checked by comparing it with different existing controllers, which provide better performance in terms of overshoot, undershoot, and settling time. To optimize the different controller parameters, a newly modified chaos quasi-opposition-based sea horse optimization (CQOSHO) approach has been suggested to demonstrate its superiority it compared with several popular existing Meta-heuristic optimization. Moreover, an extensive investigation of the proposed AGC has been investigated and successfully implemented. A data-driven deep-learning-based forecasting technique is used to forecast real load with renewable energy (wind and solar) in MPS. The flexibility and reliability of the proposed controller in deregulated systems have been evaluated in various situations, including step, multi-step disturbance, and modified IEEE-118 bus. This research proposes a learning-based attack detection technique for the cyber-physical model's grid frequency deviation analysis. The suggested controller's efficacy has been compared and evaluated against previously published literature. The investigation of the compressive results shows strong evidence of the benefit and effectiveness of the suggested control technique.