摘要
针对水轮机结构复杂、建模方法多样,研究分析了理想水轮机模型、分段线性化模型、神经网络模型和插值法水轮机模型的不同建模方法,对比分析各模型仿真时间、水轮机流量、力矩、水头等参数仿真输出,结果表明:采用理想水轮机模型和分段线性化模型时仿真时间短、建模过程简单,小波动工况下能较为精确地进行动态过程仿真,但进行大波动工况计算需要调整模型参数;插值法水轮机模型能对各个工况的静态过程进行精确仿真,但在动态波动瞬间仿真存在误差比另外3种方法稍大;神经网络模型仿真动态过程误差相对较小,但建模过程复杂、仿真时间较长。
Considering the water turbine structure is complex,modeling method diversity,this paper analyzes the ideal turbine model,the piecewise linear model,neural network model and the interpolation method of different modeling methods of hydro-turbine model,through the study of the modeling of a certain power plant,the analysis of the actual simulation time,turbine’s flow changes,momentum changes,water head variation,research shows that:when using piecewise linear model and ideal turbine model the simulation time is shorter,the modeling process is simple,small fluctuations can more accurately under the condition of dynamic process simulation.When the interpolation turbine model is used,the static process of each working condition can be accurately simulated,but the simulation error increases at the moment of dynamic fluctuation.The error of simulation dynamic process is the smallest when neural network model is used,but the actual simulation time is longer and the modeling process is more complicated.
作者
陈上
陈金保
王安林
丘涛基
陈齐灯
肖志怀
CHEN Shang;CHEN Jin-bao;WANG An-ling;QIU Tao-ji;CHEN Qi-deng;XIAO Zhi-huai(School of Power and Mechanical Engineering,Wuhan University,Wuhan 430072,China;Fujian Huadian Jinhu Electric Power Co.Ltd,Sanming 353399,Fujian Province,China)
出处
《中国农村水利水电》
北大核心
2021年第11期203-208,共6页
China Rural Water and Hydropower
基金
国家自然科学基金项目“复杂多变网构下水电机组稳定性机理与机网协调控制研究”(51979204)。
关键词
水轮机
神经网络
分段线性化
二元三次插值
仿真模型
hydraulic turbine
neural network
piecewise linearization
Lagrange’s interpolation
simulation models
