数据挖掘方法在实时厂级负荷优化分配中的应用  

Application of Data Mining Method in Real-time Optimal Load Dispatching of Power Plant

作　　者：曾德良[1] 杨婷婷[1] 程晓[1] 刘吉臻[1] 

机构地区：[1]华北电力大学控制与计算机工程学院,北京市昌平区102206

出　　处：《中国电机工程学报》2010年第11期109-114,共6页Proceedings of the CSEE

基　　金：国家863高技术基金项目(2007AA041105;2007AA04Z163)~~

摘　　要：介绍一种解决实时厂级负荷优化分配问题的方法。通过大量的分析现场数据,认为机组煤耗特性曲线不能进行回归和拟合,而是一些离散点连接而成的曲线。在默认热力学计算准确的条件下,分析了影响机组供电煤耗率的因素,研究了基于信息熵的多因素权重分配方法,结合海量历史数据相对准确地预测出各个工作点下的供电煤耗率,从而建立负荷与供电煤耗率之间的实时离散模型,采用动态规划法求得快速性和经济性约束条件下的全局最优解。该方法在大唐国际盘山电厂得到实际验证,结果证明该方法准确、实用。A solution to real-time optimal load dispatching problem of power plant was introduced in this paper. Based on analysis of operating data, it was proved that the characteristics curve of unit coal consumption rate （CCR） against load are constructed by discrete points and it can＇t be fit by linear regression. In the default condition of accurate thermodynamic calculation, factors impacting unit CCR for power supply are studied. Multi-factors weight distribution method based on information entropy was employed to predict CCR for power supply at different operating points based on history and real-time data. Thus, discrete mathematical model between load and CCR was build up for each unit. Based on the proposed speediness and economy constraints, global optimal solution for optimal load dispatching was approached by dynamic programming method. The method proposed in this paper was applied in Panshan power plant of China Datang Corporation and it was proved to be practical and effective.

关 键 词：厂级负荷优化分配 数据挖掘 供电煤耗率 动态 规划法 全局最优解 

分 类 号：TK247[动力工程及工程热物理—动力机械及工程] TM261[一般工业技术—材料科学与工程]