Optimized Energy Consumption Model for Multi-Source Small Structures

Abstract

The main purpose of this thesis is to find methods for optimizing energy sources so that the cost of electricity is minimum. There are many methods and algorithms for solving optimization problem, but the method used in this paper consists of two parts. The first part is applying Forecasting algorithm such as Long Short Term Memory (LSTM) algorithm to predict and forecast the energy consumption and the second part is to schedule the energy sources according to the forecast-ed results from the algorithm using Mixed Integer Linear Programming (MILP) and Genetic Algorithm (GA), but before all this the dataset should be accurate and it should be time series because this problem is time related and it is a major constraint in this problem, which means that at some time the electricity is used a lot more. For this problem three energy sources are used. First is Traditional Grids (TG’s), second is Solar Panels (SP’s) and third is Batteries. Optimizing these energy sources is a hustle as each of these have their own constraints like grid energy can either be available or not, solar panels are only available in daylight, batteries need to recharge. By following all of these constraints the algorithms should still provide solutions.