Fast and Efficient Very Short-Term Load Forecasting Using Analogue and Moving Average Tools

The electricity markets continuous and secure operation depends on accurately predicting real-time demand. This study presents an innovative Analogue Moving Average (AnMA) method that uses classical statistical techniques like correlation, regression, and moving averages to improve the accuracy of load demand forecasting. AnMA is designed to correct for biases and unforeseen changes in load demand and offers several desirable attributes, such as high accuracy, speed, robustness, low maintenance, repeatability, and a low computational cost. The study evaluates the performance of AnMA against Naive, exponential smoothing, and Autoregressive Moving Average (ARMA) benchmarks for forecasting horizons ranging from five minutes to two hours multi-step ahead, using data from the preceding four months. The results show that AnMA is competitive with the benchmarks in terms of accuracy while offering dramatically lower computational costs, making it an efficient and highly attractive method.