Load frequency control in renewable based micro grid with Deep Neural Network based controller

Microgrids (MGs) offer numerous technical, economic, and environmental benefits, yet they face challenges due to high-frequency deviations caused by the unpredictable nature of the renewable energy source, and variable loads with the integration of Electric Vehicles (EVs). Numerous methods, algorithms, and controllers have been created to address these issues and preserve system stability and efficient load frequency control (LFC). This paper introduces a novel control strategy to optimise the load frequency model in a microgrid (MG) with vehicleto-grid interactions using Particle Swarm Optimisation - deep Artificial Neural Network (PSO-DNN). The performance of the suggested controller is evaluated against traditional techniques, including dynamic EV charging and discharging, renewable energy integration, and fluctuating generation, using the proportional integral derivative (PID) controller and the PSO-PID controller. The PSO-DNN controller achieves 99.308 % efficiency, with a minimal mean squared error and an integrated time absolute error reduced to. It achieves a transient time of 18.5626 s, demonstrating quick response, accurate control, and quick peak output capabilities with little undershoot and overshoot. This analysis for stability confirms that the PSO-DNN controller effectively ensures stability in the microgrid's LFC system amidst uncertainties and disturbances, as compared to PID and fuzzy controllers. This approach enhances resilience, reduces settling time, and ensures reliable frequency control, validating its efficacy in maintaining stable and efficient microgrid operations.