关键词: 集中供热, 管网分级, 智能控制, 分时分区分温, 智慧供热, 人工智能, 源网协同

Abstract: To eliminate the energy waste caused by hydraulic imbalance and excessive heating in district heating systems，a source-network systematic energy saving technology integrating pipeline classification and intelligent control was proposed and applied in its demonstration project. The technology is based on the integration of artificial intelligence into thermal systems. Technically，the technology coordinates various heat sources（water，electricity and natural gas）as well as the software and hardware.Practically，application of this technology should be made by hierarchy，grades and stages and take various operable decoupling factors into consideration. Applying LSTM deep neural network in the heat load prediction， combining the twin model equivalent to the actual system with the simulation model，the optimal heating scheme for the system is obtained by taking the optimal indoor temperature as the objective function. In the heating season of 2019—2020， the technology being applied in a power station of a university in Tianjin，has realized intelligent and integrated regulation on its 28 thermal inlets ，and achieved remarkable energy-saving effect.Compared to the last heating season，the gas and electricity consumptions in the heating season of 2019—2020 reduced by 25.66% and 22.87%，respectively.Considering the influence of heating time，weather and uncontrolled thermal inlets，the gas could be saved by 31.07%.Proven by the performance of the demonstration project，the comprehensive energy-saving rate is 20%—30% at the initial stage theocratically.The project laid a foundation for the targeted 60%—70% energy saving based on the simulation model and theoretical calculation，by taking precise control upon different time，zone and temperature with a minimal time and space scale.

Key words: district heating；network grading；intelligent control；division by different time, zone and temperature； intelligent heating；artificial intelligence；collaboration of source and network