Multi-objective microgrid storage planning problem using plug-in electric vehicles

Abstract

Energy storage has been evolving towards a dynamic scenario with bidirectional communication between several autonomous agents. Efficient power dispatching systems have been mainly assisted by the use of Information and Communication Technologies, Distributed Systems and Artificial Intelligence. This thesis describes a Multi-objective Storage Planing Problem considering Plug-in Electric Vehicle (PEV) as storage units. The problem involves several PEVs and a Microgrid (MG) community, composed of small houses, residential areas, and different Renewable Energy Resources. The energy storage planning is formulated as a Mixed-Integer Linear Programming (MILP) problem, considering PEVs users requirements, minimizing three different objectives and analyzing three different criteria. Two novel cost-to-variability indicators, based on Sharpe Ratio, are introduced for measuring energy storage schedules volatility. By adding these additional criteria, energy storage planning is optimized seeking to minimize the following: total MG costs; PEVs batteries usage; maximum peak load; difference between extreme scenarios and two Sharpe Ratio indices. Since prediction involves inherent uncertainty, the use of probabilistic forecasting is proposed. Probabilistic forecasts of wind and solar power production, energy consumption and prices are used in order to perform smart energy storage, checking storage plan robustness. A novel Hybrid Forecasting Model (HFM) with automatic parameter optimization, done by metaheuristic procedures, is proposed for handling the different MG forecasting problems, generating probabilistic quantiles. Finding optimal values for the HFM fuzzy rules and weights is a highly combinatorial task. Thus, parameter optimization of the model is tackled by a bio-inspired optimizer, namely GES, which combines two heuristic approaches, namely the GRASP and the Evolution Strategies metaheuristics. The proposed forecasting model is applied to forecasts different mini/microgrid v time series. In particular, its results are highlighted for load forecasting in residential and commercial areas, which are typical microgrid scenarios. Due to the quick training phase done by the proposed metaheuristic calibration algorithm, the proposed model is suggested to be embedded into Smart Meters (SM) and other future SG devices. Storage planning is scheduled for different time horizons, according to information provided by lower and upper bounds extracted from those probabilistic forecasts. In order to find sets of non-dominated solutions, a matheuristic black box solves several weighted-sum MILP subproblems. Candidate non-dominated solutions are searched from feasible solutions obtained during the searching process over the branches of a Branch-and-Bound tree. Pareto fronts are discussed and analyzed for different energy storage scenarios. The sets of non-dominated solutions and their conflicts and harmonies indicate that only minimizing system costs goes against the system robustness, increasing the risk of having higher peak loads as well as more fluctuations over the expected costs. It was also noticed the trade-off between grid maximum peak load and the MG total costs.