基于深度学习的水电机组耗量特性封装与水电站厂内经济运行  

Packaging of Water Consumption Characteristics of Hydropower Units Based on Deep Learning and Economic Operation of Hydropower Plants

作　　者：包长玉 谢俊[1] 张秋艳 陈付山[2] BAO Chang-yu;XIE Jun;ZHANG Qiu-yan;CHEN Fu-shan(The College of Energyand Electrical Engineering,Hohai University,Nanjing 211100,China;Jiangsu Engineering Consulting Center Co.,Ltd.,Nanjing 210003,China)

机构地区：[1]河海大学能源与电气学院,江苏南京211100 [2]江苏省工程咨询中心有限公司,江苏南京210003

出　　处：《水电能源科学》2022年第12期173-177,共5页Water Resources and Power

基　　金：国家重点研发计划(2019YFE0105200);国家自然科学基金项目(U1965104)。

摘　　要：水电站厂内经济运行是一个高维、离散、非线性的优化问题,针对水电机组耗量特性非线性问题,提出了一种深度学习(DL)与微分进化(DE)算法融合的方法(DL-DE算法)。采用深度学习映射水电机组耗量与出力的非线性关系,采用微分进化算法优化水电站厂内经济运行模型。算例分析表明,利用深度学习技术提取和非线性表征能力再结合DE算法的全局优化能力,可较好地实现水电站厂内经济运行问题的高效求解。The economic operation of hydropower station is a high-dimensional,discrete and nonlinear optimization problem.Aiming at the nonlinear problem of consumption characteristics of hydropower units,the DL-DE method of combining deep learning(DL)and differential evolution(DE)was proposed.Deep learning was used to map the nonlinear relationship between hydropower unit consumption and output power,and differential evolution algorithm was used to solve the economic operation model of hydropower station.The analysis of a numerical example shows that the efficient solution of the economic operation problem of hydropower station can be realized by using the feature extraction and nonlinear representation ability of deep learning technology combined with the excellent global optimization ability of DE algorithm.

关 键 词：数据驱动 深度学习 微分进化算法 水力发电 经济运行 

分 类 号：TM73[电气工程—电力系统及自动化]