Adaptive model predictive control for electricity management in the household sector. (May 2022)

• Record Type: Journal Article
• Authors: Colmenar-Santos, Antonio; Muñoz-Gómez, Antonio-Miguel; Rosales-Asensio, Enrique; Fernandez Aznar, Gregorio; Galan-Hernandez, Noemi
• Is Part Of: International journal of electrical power & energy systems. Volume 137(2022)
• Publication Date: 2022-05
• Subjects: Adaptive Model Predictive Control -- Home Energy Management System -- Smart Grid -- Demand Response -- Distributed Energy Resources -- Artificial Neural Network -- Genetic Algorithm; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2021.107831
• Languages: English
• ISSNs: 0142-0615

Abstract

Graphical abstract: Highlights: Adaptive model predictive control (AMPC) is proposed for household building control. Energy flow models use artificial network and weighted moving average methodology. A genetic algorithm performs the heat pump comfort and cost optimization. MILP optimization is applied to achieve a cost-optimal energy strategy. The building management performance is validated through a set of scenarios. The electricity bill is reduced and the elasticity of demand is increased. Abstract: This paper focuses on the optimisation of electricity consumption in residential buildings. To deal with the increase in electricity consumption, the intermittency of renewable energy generation and grid contingencies, a greater effort is required towards residential management optimisation. A novel adaptive model predictive control algorithm is proposed to achieve this objective. The challenges for this research included recognising and modelling the economic and technical constraints of the sources and appliances and addressing the uncertainties concerning the weather and user behaviour. Data-driven models are developed and trained to predict the user behaviour and buildings. Artificial neural networks and statistical models based on the weighted moving average are proposed to capture the patterns of deferrable and non-deferrable appliances, battery storage, electric vehicles, photovoltaic modules, buildings and grid connections. A dual optimisation method is devised to minimise the electricity bill and achieve thermal comfort. The proposed optimisation solver is a two-step optimisation method based on genetic algorithm and mixed integer linear programming. A comprehensive simulation study was carried out to reveal the effectiveness of the proposed method through a set of simulation scenarios. The results of the quantitative analysis undertaken as part of this study show the effectiveness of the proposed algorithm towards reducing electricity charges and improving grid elasticity.