基于数据驱动的水泥熟料烧成系统能耗优化

Energy consumption optimization of cement clinker burningsystem based on data-driven

水泥制造业一直是我国的高能耗行业之一,对能源的依赖度高,能源消费在生产成本的占比达40%~60%。根据《建材工业“十三五”发展指导意见》,水泥企业的节能取得很大进展,但对比世界先进水平,吨水泥综合能耗仍存在差距。水泥烧成系统是水泥生产过程中的主要能耗部分,水泥烧成系统内部进行着复杂的理化反应,涉及众多环节与设备,采集的数据具有非线性、强耦合性、变量众多和大滞后等特点。随着人工智能的发展和工业数据采集的普及,分散控制系统(distributed control system,DCS)在各行各业都获得了广泛应用,人工智能分析优化方法成为工业数据分析优化的主流。为了提升水泥生产企业的生产效率,在分析水泥熟料烧成系统电力过程的历史运行变量和生产能效数据的基础上,采用一种基于数据驱动的水泥熟料烧成系统能耗优化与参数推荐混合算法。针对参数筛选问题,采用平均影响值算法进行能耗敏感度分析,对影响能耗的参数进行筛选。在建模阶段,采用改良BP神经网络对能耗进行建模,得到系统能耗模型后,通过遗传算法对其优化,使该能耗模型以吨熟料最低电耗的优化目标对可控运行参数进行寻优,能够获得运行参数的优化值。算法在白山水泥厂水泥熟料烧成系统中进行了实际部署,运行结果表明,该算法有效支撑了水泥熟料烧成系统的能耗管理,优化前水泥能耗始终在15000 kWh左右,通过仿真优化后,最优能耗为13661 kWh,约减少7%,同时可获得特征变量推荐值。

The cement manufacturing industry has always been one of the high energy consumption industries in China,which is highly dependent on energy.Energy consumption accounts for 40%-60%of production costs.In recent years,according to the"Thirteenth Five-Year Development Guidelines for the Building Materials Industry",cement companies have made great progress in energy conservation.However,compared with the world's advanced level,there is still a gap in the comprehensive energy consumption per ton of cement.The cement firing system is the main energy consumption part in the cement production process.The cement firing system carries out complex physical and chemical reactions,involving many links and equipment,and the collected data has the characteristics of nonlinearity,strong coupling,numerous variables and large lag.In recent years,with the development of artificial intelligence and the popularization of industrial data collection,distributed control system(DCS)has been widely used in various industries,and artificial intelligence analysis and optimization methods have become the mainstream of industrial data analysis and optimization.In order to improve the production efficiency of cement production enterprises,and based on the analysis of the historical operating variables and production energy efficiency data of the electric process of the cement clinker burning system,a data-driven energy hybrid algorithm for consumption optimization and parameters recommendation of the cement clinker burning system are used.First,for the parameter selection problem,the average influence value algorithm is used to analyze the energy consumption sensitivity,and the parameters that affect energy consumption are filtered.In the modeling phase,the improved BP neural network is used to model the energy consumption.After obtaining the system energy consumption model,the genetic algorithm is used to optimize it,so that the energy consumption model can be controlled to run with the optimization goal of the lowest power consumption per ton of clinker and can obtain the optimized values of the operating parameters.The algorithm has actually been deployed in the cement clinker firing system of Baishan Cement Plant.The operation results show that the algorithm effectively supports the energy consumption management of the cement clinker firing system.Before optimization,the cement energy consumption is always around 15000 kWh.Through simulation optimization,the optimal energy consumption is 13661 kWh,which reduces energy consumption by about 7%.At the same time,the recommended values of characteristic variables can be obtained.