A Deep Learning Approach for Peak Load Forecasting: A Case Study on Panama

Abstract

Predicting the future peak demand growth becomes increasingly important as more consumer loads and electric vehicles (EVs) start connecting to the grid. Accurate forecasts will enable energy suppliers to meet demand more reliably. However, this is a challenging problem since the peak demand is very nonlinear. This study addresses the research question of how deep learning methods, such as convolutional neural networks (CNNs) and long-short term memory (LSTM) can provide better support to these areas. The goal is to build a suitable forecasting model that can accurately predict the peak demand. Several data from 2004 to 2019 was collected from Panama’s power system to validate this study. Input features such as residential consumption and monthly economic index were considered for predicting peak demand. First, we introduced three different CNN architectures which were multivariate CNN, multivariate CNN-LSTM and multihead CNN. These were then benchmarked against LSTM. We found that the CNNs outperformed LSTM, with the multivariate CNN being the best performing model. To validate our initial findings, we then evaluated the robustness of the models against Gaussian noise. We demonstrated that CNNs were far more superior than LSTM and can support spatial-temporal time series data.

A Holistic Approach to Power Systems Using Innovative Machine Learning and System Dynamics

“…Long-term expansion plans, also known as integrated resource planning, must guide policymakers and stakeholders in managing the future transition of power systems. To develop and revise these plans accordingly, electric companies first predict future scenarios (pessimistic, moderate, and optimistic) regarding peak demand and the growth of energy consumption, based on the historical data and future indicators (i.e., economic growth, weather, and demographic projections) [8] to determine the requirements for them to be able to meet future demand. By forecasting demand, energy planners can calculate the generation capacity required to meet future demand while complying with environmental policies.…”

“…A list of the companies that are considered large clients in Panama includes Cemento Bayano, Gaming & Services de Panamá S.A., Gold Mills, Green Tower Properties, Hospital Punta Pacífica, Nestle Fabrica de Los Santos, and Nestle Fabrica de Nata. In our previous work [8], the SelectKBest method was used to determine which features contribute the most substantially to peak demand growth. The SelectKBest feature selection method is a tool from the Scikit-learn library that removes all the features except those with the highest scores.…”

“…The architecture for each of the models is demonstrated below. Initially, only four input features were considered for predicting month-ahead peak demand, based on a rigorous feature selection process that was previously conducted in [8]. These features comprise the categorical variable month, residential consumption, large client consumption, and the peak demand figures from the previous three months.…”

“…This paper is an extension of our previous work in [8,9] by providing a holistic framework that can support the actual decision-making needs of the energy industry. Most importantly, a decision-making tool is needed that can assist energy managers with the critical decisions related to day-to-day operations, while enabling them to closely manage and respond to peak demand growth more effectively in the long term.…”

Development of an Algorithm for Regulating the Load Schedule of Educational Institutions Based on the Forecast of Electric Consumption within the Framework of Application of the Demand Response

“…To control the load within an industrial enterprise, there are many algorithms for load redistribution, both by direct control by determining the optimal load [20], and by planning power consumption [21].…”