基于多种算法的多目标配煤优化方法  

Uniform design based multi-objective optimization method for coal blending

作　　者：刘玉娇[1] 张海英[1] 关海盈 

机构地区：[1]西安理工大学复杂系统控制与智能信息处理重点实验室,陕西西安710048

出　　处：《热力发电》2014年第9期108-112,共5页Thermal Power Generation

摘　　要：针对试验法动力配煤存在操作繁琐、信息滞后较大以及配煤方案粗略等缺陷,将BP网络、均匀设计、遗传算法等方法相结合,建立了多目标遗传算法优化模型,快速得到优化配煤方案。其中,BP算法用于实现混煤与单煤的发热量等煤质信息之间的非线性映射关系;采用均匀设计建立适应度函数;利用遗传算法进行寻优。研究结果表明:预测模型具有较高的可靠性和置信度。There are several disadvantages when using the determination method for coal blending,such as complicated operation,lag information and rough blending scheme.Therefore,by combiring the BP network,uniform design and genetic algorithm,a multi-objective genetic algorithm optimization model was established,to get coal blending optimization solution faster.The model employs the BP neural network to determine the nonlinear relationship between the blending coal and single coal during the learning process,the uniform design to establish fitness function,and the genetic algorithm to optimize.The results proved that the prediction model has high reliability and confidence.

关 键 词：配煤 多目标 均匀设计 BP网络 遗传算法 

分 类 号：TK018[动力工程及工程热物理]