M5’模型树在热电厂负荷优化中的应用  

The Application of M5' Model Tree on Optimal Load Distribution in Thermal Power Plants

作　　者：顾雅云[1] 胡林献[1] 

机构地区：[1]哈尔滨工业大学电气工程系,黑龙江哈尔滨150001

出　　处：《节能技术》2013年第5期426-429,共4页Energy Conservation Technology

摘　　要：机组汽耗特性是热电厂负荷优化分配的基础。为了改进传统单一多元线性回归模型无法适应非凸、非连续的汽耗特性,本文基于M5’模型树算法,滚动利用机组自动化系统的最新历史数据获取最新汽耗特性,并在此基础上建立实时厂级负荷优化模型,给出了应用差分算法求解全局最优解的方法。实例表明:所建立的汽耗特性模型的预测能力强且建模过程可控,方便应用于热电厂的负荷优化,具有实用价值。The steam consumption characteristics of turbine is the basis of optimal load distribution in thermal power plants. To solve the problem that traditional single multivariate linear regression model is not able to fit the non - convex and non - continuity steam consumption characteristics very well, a new method to achieve the latest steam consumption characteristics by a scroll way using latest measurement data from turbines＇ automation system was proposed in this article. Based on the above work, a real - time plant level optimal load distribution model was established, the global optimal solution of which was searched by differential evolution algorithm. The steam consumption characteristics achieved by the new method has many advantages which were proved by an example, such as good predictability, control lable modeling process, and practical value in optimal load distribution of thermal power plants.

关 键 词：热电厂 负荷优化 汽耗特性 M5’模型树算法 差分进化算法 

分 类 号：TK262[动力工程及工程热物理—动力机械及工程]