XAI4HEAT: Towards Demand-Driven, AI Facilitated Management of District Heating Systems
摘要
In this paper, we introduce the methodology and system architecture for achieving the demand-driven, AI facilitated, end-to-end management of District Heating Systems (DHS). The approach is based on the capability to forecast the future heat demand, where those forecasts consider historical data of transmitted heat energy, human operator procedures, weather forecasts and thermal comfort of the end consumers. The proposed approach aims to bring significant improvements into costs and environmental impact of running a DHS by introducing a new intelligent control scheme. This scheme will replace a traditional regulation curve used to manage the flow and heat exchange in the DHS secondary line based on the outside ambient temperature, with the hyperplane defined by introduced predictor features.