Hourly Heat Load Prediction for District Heating Systems based on Long Short-Term Memory Network

Abstract

In conjunction with the rapid progress in intelligent heating technology, heat load-based prediction and on-demand heating have been considered as the most important factor for the realization of optimized regulation in a centralized system. The prediction of heat load in heating systems will be useful for the optimization of automatic control and green heating. But traditionally applied methods are less effective and vulnerable to noise, hence they cannot meet the demands of industrial practice. Recently, deep learning techniques have been intensively researched and used in order to extract meaningful features from the data and construct models. LSTM model has proven superior compared to other approaches in terms of performance. In this regard, the purpose of this study is to use a deep learning method of Long Short-Term Memory Network for the accurate heat load prediction in DHSs in order to reduce the waste of resources.