基于引力搜索算法与LSSVM的机组初压优化

Unit Initial Pressure Optimization Based on Gravity Search Algorithm and LSSVM

主蒸汽压力是影响汽轮机热经济性的重要参数。以精确确定汽轮机滑压运行时最优初压为目标,提出了一种基于引力搜索算法(GSA)与最小二乘支持向量机(LSSVM)相结合的汽轮机初压优化方法。首先,采用LSSVM建立汽轮机热耗率预测模型,同时,GSA算法被用来优化LSSVM模型超参数以改善GSA-LSSVM模型的泛化能力;然后,在GSA-LSSVM热耗率预测模型基础上利用GSA算法搜索各个负荷下热耗率最小时所对应的主蒸汽压力,即为最优初压。最后,对某电厂600MW机组进行初压优化实验,仿真结果验证了该方法能够优化搜索到较好的主蒸汽运行初压。

The main steam pressure is an important parameter that affects the thermal economy of steam turbine.In order to accurately determine the optimal initial pressure of steam turbine during sliding pressure operation,a hybrid method of steam turbine initial pressure optimization based on gravity search algorithm(GSA)and least square support vector machines(LSSVM)is proposed in this paper.Firstly,with the aid of LSSVM the regression forecasting model is established for the heat rate of steam turbine,in which GSA algorithm is developed to find the optimal parameters of LSSVM to improve the regression accuracy and generalization capability for heat rate forecast.Then,based on the established model,the GSA algorithm is re-applied to seek the optimal main steam pressure with respect to the minimum heat consumption rate under each load level.Finally,taking a 600 MW supercritical steam turbine as the research object,the simulation results show that through optimization search the proposed method can be employed to provide a satisfactory initial pressure value for the main steam operation.