基于双层聚类与GSA-LSSVM的汽轮机热耗率多模型预测  

Multi-model for turbine heat rate forecasting based on double layer clustering algorithm and GSA-LSSVM

作　　者：牛培峰[1,2] 刘超[1] 李国强[1] 张维平[3] 陈科[1] 

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004 [2]国家冷轧板带装备及工艺工程技术研究中心,河北秦皇岛066004 [3]秦皇岛职业技术学院机电工程系,河北秦皇岛066100

出　　处：《电机与控制学报》2016年第3期90-95,共6页Electric Machines and Control

基　　金：国家自然科学基金(61403331;61573306);河北省自然科学基金(F2010001318)

摘　　要：针对单模型难以精确描述具有复杂非线性特性的汽轮机热耗率的问题,提出一种新的热耗率多模型建模方法。首先应用GK算法分析出最优聚类个数以及初始聚类中心,避免了聚类数确定的盲目性;然后利用核模糊C均值算法对热耗率样本集做出聚类划分,在每个子空间中利用最小二乘支持向量机(LSSVM)辨识出相应子模型,同时,为了保证子模型精确度,采用引力搜索算法来解决LSSVM参数优化问题;最后,将子模型通过隶属度值加权融合得到精确的热耗率预测模型。以某600MW超临界汽轮机组为研究对象,基于现场数据建立汽轮机热耗率预报模型,仿真结果验证了提出的多模型建模方法具有较高的预报精确度和泛化能力。Aiming at the issue that the characteristic of complex nonlinearity of heat rate for steam turbine which was difficult to be descript accurately by the single-model,a new multi-model modeling method for heat rate was presented. Firstly,the initial cluster centers and optimized cluster numbers were obtained by the G-K algorithm. Then,the data set was clustered into several local regions with kernel-based fuzzy C-means clustering algorithm. In addition,the sub-model was built by least square support vector machine（ LSSVM）. Moreover,in order to guarantee accuracy of each sub-model,gravitational search algorithm was employed to solve parameter optimization problem of LSSVM. Finally,towards obtaining accurate predictions over the entire operating range,a degree of membership composition strategy to aggregate the local model predictions was proposed. A 600 MW steam turbine unit was used as the identification object,and multi-model for heat rate was established based on field data. Simulation results show that the multi-model modeling method has well regression precision and generalization ability.

关 键 词：多模型 热耗率 引力搜索算法 最小二乘支持向量机 聚类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]