Modern technology has made it possible to harvest Renewable Energy Sources (RES) on a large scale. A sustainable method of generating electricity is projected to be Smart Grids (SGs), which mix traditional and RES sources. Additionally, all RES are affected by environmental factors, which alters the amount of power generated by these sources. Additionally, accessibility is based on yearly or daily periods. Although real-time demand forecasting is made possible by smart metres, accurate models that forecast the electricity generated by RES are also necessary. The reliability of grid stability, effective scheduling, and energy management are all guaranteed by prediction models (PMs). For instance, if the model predicts a period of Renewable Energy (RE), the SG must smoothly transition into the conventional energy source for that time and ensure that the electricity generated satisfies the anticipated demand loss. The research also recommends several learning-based PMs for sources of RE using open data sources and scheduling techniques for demand-supply matching. This study created a model that faithfully reproduces a microgrid, forecasts supply and demand, flexibly plans power delivery to meet demand, and provides practical insights about how the SG system functions. The Demand Response Programme (DRP) is also developed in this work employing cost-saving incentive-based payment packages. The test results are evaluated in several scenarios for the multi-objective ant colony optimisation algorithm (MOACO) with and without the input of the DRP to optimise operating costs.