梯级水电站群联合优化调度函数的逐步回归模型

发布时间：2019-05-24 09:11

【摘要】：针对梯级水电站群联合优化调度函数识别中存在的线型选择及变量筛选问题,采用逐步回归模型筛选关键作用变量并建立回归调度函数方程,以保证调度函数的简洁、有效性。将模型应用于金沙江与三峡四库梯级系统提取消落期系统联合优化调度函数,从四库梯级联合优化调度样本中通过逐步回归调度函数辨识溪洛渡、三峡水库协同消落机制并检验了模型精度。结果表明:在率定来水样本条件下,逐步回归调度函数拟合误差均在3%以内,拟合精度较高;在检验来水样本条件下,与多元线性回归调度函数相比,逐步回归调度函数对应总发电效益平均增加1.67亿元(0.69%)。由于逐步回归调度函数剔除了次要因素干扰,相较于多元线性回归调度函数泛化能力更优,具有更高实用价值。
[Abstract]:Aiming at the problem of line selection and variable selection in the identification of joint optimal operation function of cascade hydropower stations, the stepwise regression model is used to screen the key operational variables and the regression dispatching function equation is established to ensure the simplicity and effectiveness of the dispatching function. The model is applied to the Jinsha River and the three Gorges four-reservoir cascade system to extract the joint optimal scheduling function of the collapse time system, and the Xiluodu is identified by stepwise regression scheduling function from the four-reservoir cascade joint optimal dispatching sample. The cooperative collapse mechanism of the three Gorges Reservoir is tested and the accuracy of the model is tested. The results show that under the condition of calibrating incoming water samples, the fitting error of stepwise regression scheduling function is less than 3%, and the fitting accuracy is high. Under the condition of testing incoming water samples, compared with multivariate linear regression scheduling function, the total power generation benefit corresponding to stepwise regression scheduling function increases by 167 million yuan (0.69%) on average. Because the stepwise regression scheduling function eliminates the interference of secondary factors, compared with the multivariate linear regression scheduling function, the generalization ability of the stepwise regression scheduling function is better and has higher practical value.
【作者单位】：河海大学水文水资源学院;中国长江三峡集团公司;安徽省佛子岭水库管理处;
【基金】：国家自然科学基金项目(51609062、51579068) 国家重点研发计划项目(2016YFC0400909) 中央高校基本科研业务费专项资金资助项目(2015B28414)
【分类号】：TV737

【相似文献】