水力发电学报
          Home  |  About Journal  |  Editorial Board  |  Instruction  |  Download  |  Contact Us  |  Ethics policy  |  News  |  中文

Journal of Hydroelectric Engineering ›› 2021, Vol. 40 ›› Issue (1): 88-96.doi: 10.11660/slfdxb.20210109

Previous Articles     Next Articles

Neural network estimation methods for varying output coefficients of hydropower stations

  

  • Online:2021-01-25 Published:2021-01-25

Abstract: Focusing on how to improve the accuracy in estimating power output coefficients and power generation process of a hydropower station for its medium-long-term generation dispatching, this study develops three neural network models with an output of power output coefficient K and three alternative inputs–water head, generation flow, and inflow rate. These models adopt the neural network data mining technology and consider the influence of the station’s key status information comprehensively. And we formulate three neural network methods for estimating the varying K values in the station’s output calculations and compare the methods with traditional methods, in a case study of the many years’ operation data of the Three Gorges hydropower station. The results show our methods are more accurate in the estimations of K value, power output, and power generation, providing a new practical approach to the refined output calculation of medium-long-term power generation dispatching of hydropower stations.

Key words: hydropower station, medium-long term generation dispatching, refined power output calculation, estimation of varying power output coefficient, neural network

Copyright © Editorial Board of Journal of Hydroelectric Engineering
Supported by:Beijing Magtech