Abstract:
The thesis presents a hierarchal Energy Management System (EMS) approach for
designing a Virtual Power Plant (VPP) using novel machine learning and metaheuristic
algorithms in Smart Grid Distribution Systems (SGDS). The VPP is designed to
integrate renewable energy resources such as solar, wind and battery storage with
residential load dispatch. The decentralization of proposed VPP is maintained using
intelligent Multi-Agent Based Modelling (M-ABM). The preference of prosumer is kept
intact by efficient communication architecture. A detailed analysis of two Machine
Learning (ML) algorithms i.e. Linear Regression (LR) and Random Forest (RF) is
considered to forecast power generation according to the load demand. The proposed model implements four metaheuristic algorithms i.e. Artificial Bee Colony Optimization (ABCO), Gray Wolf Optimization (GWO), Squirrel Search Optimization (SSO) and Salp Swarm Algorithm (SSA) to solve power optimization problems. A case-wise detailed parametric and sensitivity analysis of algorithms is presented to aid the prosumers in terms of appliance scheduling. The test results of all six algorithms are devised in three optimization planning aspects, namely technical, economic andenvironmental to find out the most efficient and reliable EMS operation
