Arti®cial neural network based electrical load prediction for food retail stores D. Datta *, S.A. Tassou Department of Mechanical Engineering, Brunel University, Uxbridge, Middlesex, UB8 3PH, U.K. Received 20 February 1998 Abstract It has been shown by a number of investigators that arti®cial neural networks (ANNs) can be more reliable and eective building energy predictors than traditional simulation models. This paper presents the results from comparisons of the predictive accuracy of two commonly used neural networks employed for the prediction of the electrical load of a retail food store. The networks used were the multi-layered perceptron (MLP) and radial basis function (RBF). The MLP network was found to perform better than the RBF network particularly in the prediction of ¯uctuations of the electrical energy around the base and maximum loads. Further work will be carried out to optimise the structure and prediction accuracy of the two networks. # 1998 Elsevier Science Ltd. All rights reserved. Keywords: Electrical load prediction; Retail food stores; Arti®cial neural networks 1. Introduction Retail food stores are amongst the greatest single end users of electricity with refrigeration systems accounting for more than 50% of the electricity used. Lighting accounts for about 25% with the Heating, Ventillation and Air Conditioning (HVAC) equipment and other utilities accounting for the remainder. The retail industry continues to increase the average store size while upgrading facilities to improve service, reliability, energy eciency and cost eectiveness. The energy consumption of retail store refrigeration systems is a function of a number of variables which include the building fabric, the ambient conditions (temperature, solar insolation and wind velocity), the occupancy of the store (i.e. sales activity), and the internal environment. In the U.K., it is a common practice for the refrigeration and HVAC system to be part of an integrated design to take advantage of the rejected heat from the Applied Thermal Engineering 18 (1998) 1121±1128 1359-4311/98/$19.00 # 1998 Elsevier Science Ltd. All rights reserved. PII: S1359-4311(98)00034-9 PERGAMON * Corresponding author.