Search

Electricity is crucial for daily life due to the number of activities that depend on it. To forecast future electric load, which changes over time and depends on various factors, grid operators (GOs) must create forecasting models for various time horizons with a high degree of accuracy because the results have a huge impact on their decision-making regarding, for example, the scheduling of power units to supply user consumption in the short or long term or the installation of new power plants. This has led to the exploration of multiple techniques like statistical models and Artificial Intelligence (AI), with Machine-Learning and Deep-Learning algorithms being the most popular in this latter field. This paper proposes a neural network-based model to forecast short-term load for a Colombian grid operator, considering a seven-day time horizon and using an LSTM recurrent neural network with historical load values from a region in Colombia and calendar features such as holidays and the current month corresponding to the target week. Unlike other LSTM implementations found in the literature, in this work, the LSTM cells read multiple load measurements at once, and the additional information (holidays and current month) is concatenated to the output of the LSTM. The result is used to feed a fully connected neural network to obtain the desired forecast. Due to social problems in the country, the load data presents a strange behavior, which, in principle, affects the prediction capacity of the model. Still, it is eventually able to adjust its forecasts accordingly. The regression metric MAPE measures the model performance, with the best predicted week having an error of 1.65% and the worst week having an error of 26.22%. Additionally, prediction intervals are estimated using bootstrapping. [ABSTRACT FROM AUTHOR]

Purpose: This work aims to evaluate demand forecasting models to determine if using exogenous factors and machine learning techniques helps improve performance compared to univariate statistical models, allowing manufacturing companies to manage demand better. Design/methodology/approach: We implemented a multivariate Auto-Regressive Moving Average with eXogenous input (ARMAX) statistical model and a Neural Network-ARMAX (NN-ARMAX) hybrid model for forecasting. Later, we compared both to a standard univariate statistical model to forecast the demand for electrical products in a Colombian manufacturing company. Findings: The outcomes demonstrated that the NN-ARMAX model outperformed the other two. Indeed, demand management improved with the reduction of overstock and out-of-stock products. Research limitations/implications: The findings and conclusions in this work are limited to Colombian manufacturing companies that sell electrical products to the construction industry. Moreover, the experts from the company that provided us with the data also selected the external factors based on their own experiences, i.e., we might have disregarded potential factors. Practical implications: This work suggests that a model using neural networks and including exogenous variables can improve demand forecasting accuracy, promoting this approach in manufacturing companies dealing with demand planning issues. Originality/value: The findings in this work demonstrate the convenience of using the proposed hybrid model to improve demand forecasting accuracy and thus provide a reliable basis for its implementation in supply chain planning for the electrical/construction sector in Colombian manufacturing companies. [ABSTRACT FROM AUTHOR]

Planning and operation of smart energetic projects have increased in complexity and requires a deeper analysis. The inclusion of renewable energy and distributed generation sources lead to an inherent difficulty to define appropriate models for decision-making. Nowadays, decisions in planning and operation are made level by level due to a disarticulation of information within the electrical sector and the stakeholders. Therefore, here is proposed an architectural decision model of multilevel-multiobjective to solve this problem. In this scheme, the problem of energetic assets sizing is on the upper level, while the problem of operation and dispatch of unit power is on the lower level. Both levels are related through quantitative tools (key performance indicators (KPI)) and qualitative tools (quality function deployment (QFD) and hierarchical analytical process (AHP)). This model was applied in the Yucatán peninsula. As a result, an integral solution with a lower cost and better environmental factor than the baseline for the region was obtained. [ABSTRACT FROM AUTHOR]

Planning and operation of Smart energetic have become more complex to analyse due to structural changes in the energy sector. The inclusion of distributed generation sources, generation with renewable sources, storage systems, and the dislocation of information between the different organization levels and actors lead to the inherent difficulty of defining appropriate models that help decision making. Nowadays, decisions in planning and operation are made level by level rather than integrated manner. To address this problem, this work proposes a multi-level methodological framework based on Key Performance Indicators and System of Systems concepts. Involving methods, both quantitative and qualitative, this work serves as guidance to managers, planners, or political decision-makers from any electrical enterprise to help them find suitable solutions for planning and operation of smart grid projects. [ABSTRACT FROM AUTHOR]