Deep learning assisted optimal dispatch for renewable-based energy system considering consumer incentive scheme

Techno-economic development and implementation of renewable-based technologies require a strategic shift in the power system decision-making process. Hence, accurate power forecasting and optimal dispatch strategy are essential to achieve desirable power management. Owing to this, the research incorporates a hybrid one-dimensional convolutional neural network (1-DCNN) and long short-time memory (LSTM) deep learning model optimized with the walrus optimization algorithm (WAOA) for accurate forecasting of wind and solar power. Further, an optimal power dispatch strategy is formulated for a hybrid solar-wind-diesel-based energy system employing the Corona Virus Optimization Algorithm (CVOA) and results obtained through a hybrid deep learning forecasting model. This novel strategy seeks to minimize the overall system cost while incentivizing consumers for load shedding. The forecasting model has shown an accuracy of 96.31% and RMSE of 52.706 W/m2 as compared to the conventional model and also the CVOA optimization algorithm has shown the power curtailment by consumers for the three seasons are 102.307, 104.823, and 104.497 kWh and the incentive received in dollars are 44.760, 35.533, 34.350. This work highlights the potential for major energy savings and financial advantages in hybrid energy systems and advances our understanding of micro grid optimization, demand response programs, and renewable energy forecasts.