Forecasting Short-Term Electricity Load Using Validated Ensemble Learning

Abstract

As short-term load forecasting is essential for the day-to-day operation planning of power systems, we built an ensemble learning model to perform such forecasting for Thai data. The proposed model uses voting regression (VR), producing forecasts with weighted averages of forecasts from five individual models: three parametric multiple linear regressors and two non-parametric machine-learning models. The regressors are linear regression models with gradient-descent (LR), ordinary least-squares (OLS) estimators, and generalized least-squares auto-regression (GLSAR) models. In contrast, the machine-learning models are decision trees (DT) and random forests (RF). To select the best model variables and hyper-parameters, we used cross-validation (CV) performance instead of the test data performance, which yielded overly good test performance. We compared various validation schemes and found that the Blocked-CV scheme gives the validation error closest to the test error. Using Blocked-CV, the test results show that the VR model outperforms all its individual predictors.

Short-term Electricity Demand Forecasting using Deep Neural Networks: An analysis for Thai data

“…Result Reference MLR with AR(2) Bayesian estimation provides consistent and better accuracy compared to OLS estimation [32] PSO with ANN Implementing PSO on ANN model outperformed shallow ANN model [46] OLS Interation of variable improves the prediction accuracy [31] OLS and Bayesian estimation Including temperature variable in a model can improved the prediction accuracy upto 20% [45] PSO & GA with ANN PSO+GA outperformed PSO with ANN [35] OLS, GLSAR, FF-ANN OLS and GLSAR models showed better forecasting accuracy than FF-ANN [36] Ensemble for regression and ML Lowers the test MAPE implementing blocked Cross Validation scheme. [37] FNN, RNN based LSTM & GRU For weekdays and for aggregate data GRU shows better accuracy In this study Weather conditions have a significant impact on short-term electricity demand forecasting and are commonly incorporated into forecasting models [43]. For short lead times of up to six hours, univariate methods that do not include weather variables are often deemed adequate [12].…”

“…The proposed methods can be categorized into three frameworks: data pre-processing, model design and estimation, and comparative study. Since this work is the extended version of [36,37], we have excluded details of data characteristics, variable identification procedure and their pre-processing stage to avoid an ambiguous presentation. However, carefully grouped datasets, hereafter named the scenarios are considered for further discussion.…”

“…Now, we have used the same dataset by reducing the existing four scenarios into two scenarios as recomended in [36]. As the continuation of our previous work, Chatum et al [37] also tested the same dataset by developing machine learning models with an ensemble learning approach. Therefore, one reason of the extension was to full-fill the gap of DNN implementation and this work mainly focus on finding the best DNNs model on the basis of forecasting accuracy.…”

“…However, their results, as measured by Mean Absolute Percentage Error (MAPE), were not very impressive. [37].…”

“…• increasing the number of hidden layers does not ensure the improvement of forecasting accuracy. • as this dataset has been tested in various methods such as Bayesian [32], Regression [32,36,45], machine learning [33][34][35]48], ensemble learning [37] etc, the experimental gap on deep learning network is now fulfilled by this paper.…”