作者:胡樹云(杭州和利時自動化有限公司,浙江 杭州 310018)
王威(丹東金山熱電有限公司,遼寧 丹東 118000)
費盼峰(北京和利時工業軟件有限公司,北京 100176)
摘要:隨著環保要求的不斷提高,城市集中供暖小鍋爐被逐步關停,并被接入城市主干網,熱網不斷擴張。與此同時,熱量的生產也運用地熱、太陽能、工業余熱、電熱等多種熱源,使得集中供熱系統變得更加復雜。靠傳統手工運算方式、或者理想機理建模方式較難對熱網的結構設計及運行進行科學優化,需要通過計算機仿真建模的手段,并結合實際熱網運行的數據對熱網進行阻力特性辨識,才能真正起到有效的作用。本文研究了基于數據驅動與機理模型融合的集中供熱網水力平衡分析模型,并利用來自熱網SCADA運行數據通過多種機器學習算法對先驗知識模型的參數進行學習優化,最終建立與真實熱網相匹配的水力分析模型,此種方法可為熱力企業的熱網結構優化改造、經濟運行提供技術參考。
關鍵詞:智能熱網;水力分析;數據驅動;機理模型;機器學習
Abstract: With the continuous improvement of environmental protection requirements, small boilers for urban central heating are gradually shut down and their heating network are connected to the urban backbone network,therefore,the urban central heating network continues to expand. At the same time, heat production also uses multiple heat sources such as geothermal, solar, industrial waste heat, and electric heating which make the central heating system more complicated. It is difficult to scientifically optimize the structure design and operation of heat supply network by traditional manual calculation methods or ideal mechanism modeling methods. In order to really play an effective role, it is necessary to identify the resistance characteristics of the heat supply networks by means of computer simulation modeling and combined with the actual heating grid operation data. This paper studies the hydraulic balance analysis model of the central heating network based on the fusion of data-driven and mechanism model, uses the operation data from the SCADA of heating grid to learn and optimize the parameters of the prior knowledge model through a variety of machine learning algorithms, and finally establishes the hydraulic analysis model matching with the real heating grid. This method can be used for the structural optimization, transformation and economy of the heating network of thermal enterprises Provide technical reference for operation.
Key words: Smart heating network; Hydraulic analysis; Data-driven;Mechanism model; Machine learning
摘自《自動化博覽》2020年4月刊