Additive Ensemble Neural Network with Constrained Weighted Quantile Loss for Probabilistic Electric-Load Forecasting

Abstract

This work proposes a quantile regression neural network based on a novel constrained weighted quantile loss (CWQLoss) and its application to probabilistic short and medium-term electric-load forecasting of special interest for smart grids operations. The method allows any point forecast neural network based on a multivariate multi-output regression model to be expanded to become a quantile regression model. CWQLoss extends the pinball loss to more than one quantile by creating a weighted average for all predictions in the forecast window and across all quantiles. The pinball loss for each quantile is evaluated separately. The proposed method imposes additional constraints on the quantile values and their associated weights. It is shown that these restrictions are important to have a stable and efficient model. Quantile weights are learned end-to-end by gradient descent along with the network weights. The proposed model achieves two objectives: (a) produce probabilistic (quantile and interval) forecasts with an associated probability for the predicted target values. (b) generate point forecasts by adopting the forecast for the median (0.5 quantiles). We provide specific metrics for point and probabilistic forecasts to evaluate the results considering both objectives. A comprehensive comparison is performed between a selection of classic and advanced forecasting models with the proposed quantile forecasting model. We consider different scenarios for the duration of the forecast window (1 h, 1-day, 1-week, and 1-month), with the proposed model achieving the best results in almost all scenarios. Additionally, we show that the proposed method obtains the best results when an additive ensemble neural network is used as the base model. The experimental results are drawn from real loads of a medium-sized city in Spain.

Enhancing Short-Term Electric Load Forecasting for Households Using Quantile LSTM and Clustering-Based Probabilistic Approach

“…The authors proposed ANN probabilistic electricity forecasting with quantile optimization, considering the randomness of inputs and the output variation. In [27], the constrained weighted quantile loss (CWQLoss) for supervised regression has been proposed. The feature learning blocks are fully connected and convolutional layers and are trained end-to-end using stochastic gradient descent, which are the building blocks of the architecture.…”

Anomaly Detection with Machine Learning Algorithms and Big Data in Electricity Consumption

“…Using an Artificial Neural Network (ANN), the current consumption is repeatedly compared with the forecast or typical profile, and if the difference is over a threshold, a suspicion is raised. Furthermore, recent works using deep learning models and dynamic mode decomposition techniques for probabilistic and short-term load forecasting are proposed in [11,12]. A forecasting method based on gradient boosting neural networks is proposed for network traffic classification that can be adapted for electricity consumption forecast as input for fraud detection [13].…”

Improved Quantile Convolutional and Recurrent Neural Networks for Electric Vehicle Battery Temperature Prediction

“…Quantile regression can be used for time-series forecasting, such as in Quantile Random Forest (QRF) and Quantile Extra Trees Regressor (QETR) from the Python package scikit-garden [73] . It can be combined with neural networks [74,75] , CNN [24,76,77] , RNN [76][77][78] , and other DL models [77,79] . The models from scikit-garden can be used as reference models, as done by Refs.…”