基于粗糙集与最小二乘支持向量回归的汽轮机主蒸汽流量预测

针对传统主蒸汽流量计算方法的不足,提出了一种新的主蒸汽流量预测方法,该方法综合了粗糙集理论与最小二乘支持向量回归算法的优点,利用ROSETTA V1.4.41研究实验平台中的遗传约简算法对输入变量的属性进行约简,再利用最小二乘支持向量回归算法建立主蒸汽流量的预测模型。实验表明,与未经粗糙集理论处理过的BP神经网络、支持向量回归算法和最小二乘支持向量回归算法所建模型相比,该方法具有更好的预测精度和泛化能力,且建模速度显著提高。

Forecasting of Turbine Main Steam Flow Based on Rough Sets and Least Squars Support Vector Machine Regression

A new prediction method is put forward in view of the shortages of traditional main steam flow calculation method,which combines the advantages both rough set theory and least squares support vector regression algorithm.  Therefore,this new method is called RS-LSSVR.  In RS-LSSVR,the attributes reduction of input variable by genetic algorithm is carried out on the ROSETTA V1.4.41 research experimental platform,then the main steam flow prediction model is established by LSSVR algorithm.The simulation results show that the method based on RS-LSSVR has better prediction precision and generalization ability compared with BP algorithm, support vector regression algorithm and LSSVR algorithm without treated by the RS theory.Moreover,the modeling speed increases significantly.