论文部分内容阅读
针对水轮机复杂的非线性特性以及水轮机调节系统仿真与分析所需的水轮机流量与力矩特性难以准确描述的情况,介绍了基于遗传算法参数寻优的支持向量回归水轮机综合特性建模。由已知的综合特性曲线、飞逸特性曲线得到离散数据,向导叶小开度区和低转速区数据进行合理延拓;利用具有很强全局优化搜索能力的遗传算法寻找支持向量回归(SVR)模型的最优参数,训练获得流量和力矩特性的SVR模型。仿真结果表明,与传统的BP神经网络模型比较,所提出基于遗传算法参数寻优的支持向量回归模型具有更高的精确度。
In view of the complicated nonlinear characteristics of turbines and the difficulty in accurately describing the flow and moment characteristics of hydraulic turbines required for the simulation and analysis of hydraulic turbine governing systems, the support vector regression (IHM) regression model based on genetic algorithm parameters optimization is introduced. From the known integrated characteristic curve and the fly-off characteristic curve, the discrete data are obtained, and the data of the guide vane small opening area and the low rotation speed area are reasonably extended. The support vector regression (SVR) is searched by the genetic algorithm with strong global optimization search ability. The optimal parameters of the model, the SVR model trained to obtain the flow and torque characteristics. Simulation results show that compared with the traditional BP neural network model, the support vector regression model based on genetic algorithm parameter optimization has higher accuracy.