Daily Electric Load Forecasting Based on RBF Neural Network Models

Abstract

This paper presents a method of improving the performance of a day-ahead 24-h load curve and peak load forecasting. The next-day load curve is forecasted using radial basis function (RBF) neural network models built using the best design parameters. To improve the forecasting accuracy, the load curve forecasted using the RBF network models is corrected by the weighted sum of both the error of the current prediction and the change in the errors between the current and the previous prediction. The optimal weights (called "gains" in the error correction) are identified by differential evolution. The peak load forecasted by the RBF network models is also corrected by combining the load curve outputs of the RBF models by linear addition with 24 coefficients. The optimal coefficients for reducing both the forecasting mean absolute percent error (MAPE) and the sum of errors are also identified using differential evolution. The proposed models are trained and tested using four years of hourly load data obtained from the Korea Power Exchange. Simulation results reveal satisfactory forecasts: 1.230% MAPE for daily peak load and 1.128% MAPE for daily load curve.

Improving economic values of day‐ahead load forecasts to real‐time power system operations

“…Among various forecasting models adopted for day-ahead load forecasting [4][5][6][7][8][9], ANNs are wieldy applied in research and industry practice because of their excellent non-linear approximation properties. In this paper, two widely used ANNs, back-propagation (BP)-ANN [14,37,38] and radial basis function (RBF)-ANN [39], are adopted to build the combining forecast framework.…”

“…forecast is smaller than actual value) might not bring enough units on-line in the day-ahead operation schedule, which consequently requires turning on extra expensive fast-response units in real-time and in turn increase the total operation cost. In order to reduce statistical load forecast errors, various techniques such as linear regression [2,3], autoregressive moving average [4,5], fuzzy regression [6], artificial neural network (ANN) [7,8], support vector machine (SVM) [9][10][11][12], and extreme learning machine [13] have been studied. Indeed, a general conclusion on which forecasting model always stands out may not be readily available, because performance of individual forecasting models largely depends on specific parameter settings and size/ character of data set [2,14].…”

“…Indeed, a general conclusion on which forecasting model always stands out may not be readily available, because performance of individual forecasting models largely depends on specific parameter settings and size/ character of data set [2,14]. Furthermore, although these models have been shown effective in improving day-ahead load forecast accuracy, the studies [2][3][4][5][6][7][8][9][10][11][12][13] do not demonstrate if and how much such improvements in load forecast accuracy might bring economic values to power system operations, i.e. savings in operation costs brought by such reduced statistical load forecast errors.…”

Day-Ahead Prediction of Microgrid Electricity Demand Using a Hybrid Artificial Intelligence Model

“…Artificial neural networks (ANNs), in different forms, among the famous AI methods have been widely utilized for electricity demand prediction. For instance, Multi-layer Perceptron (MLP) ANN [24], RBF ANN [25], SOM ANN [26], and feedforward multi-layer (FFML) ANN [27].…”

Short-term Forecasting of Electricity Consumption in Buildings for Efficient and Optimal Distributed Energy Management

“…Artificial neural networks (ANNs), in different forms, among the famous AI methods have been widely utilized for electricity demand prediction. For instance, Multi-layer Perceptron (MLP) ANN in [24], RBF ANN in [25], SOM ANN in [26], and feedforward multi-layer (FFML) ANN in [27].…”