Print Email Facebook Twitter Predictive Estimation of the Real-time Electricity Market Price Title Predictive Estimation of the Real-time Electricity Market Price Author De Hoogt, I. Contributor Beijer, D. (mentor) Verhaegen, M. (mentor) Faculty Mechanical, Maritime and Materials Engineering Department Delft Center for Systems and Control Date 2017-03-20 Abstract Electricity can not be stored efficiently in the grid, resulting in a need for demand and supply to be in balance. The Transmission System Operator operates a real-time electricity market to acquire extra supply or load. Traditionally, electricity usage is forecast as accurately as possible a day ahead and deviations from this electricity program are kept to a minimum as financial risk is involved with these deviations. Deviations from the submitted electricity program are settled at the real-time electricity market clearing price, which can be profitable to the parties involved. The aim of this project is to achieve accurate, real-time and interpretable prediction of the electricity market price. To this end the Generalised Fuzzy Neural Network formulation of a Non-linear Auto-Regressive with eXogenous inputs model structure (GFNARX) model structure is proposed as contribution to existing fuzzy modelling literature. Accuracy of the proposed model is shown to be comparable to that of state-of-the-art fuzzy models on a popular literature benchmark, prediction of the Mackey-Glass chaotic time series, while using computationally cheaper means. As there is no substantial amount of literature on Real-time Electricity market Price (REP) forecasting which enables comparison of GFNARX predictions to any other, two other models from time series literature are used to generate comparison material: Seasonal Auto-Regressive Integrated Moving Average with eXogenous inputs and Generalized Auto-Regressive Conditional Heteroskedasticity (SARIMAX-GARCH) and Nonlinear Auto-Regressive with eXogenous inputs (NARX). To assess whether information about the real-time electricity price can be used to reduce electricity consumption costs for assets in a demand response portfolio, a benchmark method which simulates control of the cooling motor of a developed cold storage warehouse model according to the predicted state of the real-time electricity market, is proposed. Using the GFNARX model to predict the real-time price throughout the year 2015 based on historical data, a 25.5% reduction in cumulative electricity costs is obtained compared to the reference case where all electricity consumption is bought on the day ahead market. When comparing the GFNARX prediction result to using the most recent electricity price observation as naïve forecast which achieves 16.9% cost reduction, a relative 10.3% cost reduction is achieved. Subject electricity marketfuzzyneural networkblack-box identification To reference this document use: http://resolver.tudelft.nl/uuid:9f8c5f40-c914-48f8-96c0-f056b2e449d3 Embargo date 2018-03-20 Part of collection Student theses Document type master thesis Rights (c) 2017 I. De Hoogt Files PDF mscThesis_1357956.pdf 3.18 MB Close viewer /islandora/object/uuid:9f8c5f40-c914-48f8-96c0-f056b2e449d3/datastream/OBJ/view