System and method for predicting building thermal loads

摘要: A system for forecasting predicted thermal loads a building comprises condition forecaster weather conditions to be compensated by environmental control and load predictor modeling management components generate maintaining set of conditions. The the present invention is neural network and, preferably, recurrent that generates from short-term data. trained inputting mass data occupancy actual comparing generated measured at building. Training error attributed weights neurons processing Iteratively adjusting these minimize optimizes design non-weather inputs.