摘要
进相运行是解决大电网无功过剩问题的重要措施,入网发电机组必须进行深度进相试验。由于无法做到进相安全边界的准确量化及风险预知,现场经常发生进相试验的反复失败及相关运行事故。该文基于数字孪生的高保真性与互操作性特点,采用机理及数据混合驱动的方式,提出一种面向发电机进相深度极限与安全边界量化的数字孪生方法,解决建模准确性与现场应用场景实用性需求之间的矛盾。该方法通过发电机进相运行原理及限制因素构建出孪生机理数学模型,再根据进相过程中的实时量测数据采用粒子群优化算法实现模型参数自主校正;最终实现高风险深度进相前的各类安全边界距离定量预测计算。开发发电机进相工况的风险实时分析平台,并应用于2台核电机组的入网进相试验。结果表明:平台计算耗时小于1 min,工况预测与边界分析误差均不大于0.8%。系统满足现场实时性与准确性要求,可为电力生产运行提供安全保障。
Generator phase-leading operation serves as a critical solution for addressing reactive power surplus in large-scale grid systems.Prior to grid connection,generators must undergo rigorous phase-leading capability testing.However,the inability to precisely quantify safety boundaries and predict operational risks has frequently led to experimental failures and field accidents during phase-leading operations.Leveraging the high-fidelity and interoperability features of digital twin technology,this study proposes a hybrid mechanistic-data approach to establish a digital twin framework for phase-leading depth limitation and safety boundary quantification,effectively resolving the inherent conflict between modeling precision and practical application requirements.This method constructs a twin mathematical model through the operation mechanism and limiting factors of the phase leading,and then uses the particle swarm optimization algorithm to achieve self-calibration of model parameters based on real-time measurement data.Finally,it achieves quantitative prediction of various safety boundary distances before high-risk deep phase leading.The real-time analysis platform is developed and applied to the grid connection test of two nuclear power units.The results show that the calculation time of the platform is less than 1 minute,and the error of working condition prediction and boundary analysis is not more than 0.8%.The system meets the real-time and accuracy requirements on site,and it can provide safety assurance for power production and operation.
作者
王泽民
郝亮亮
鲁秦怡
赵岩
何俊
涂水员
WANG Zemin;HAO Liangliang;LU Qinyi;ZHAO Yan;HE Jun;TU Shuiyuan(School of Electrical Engineering,Beijing Jiaotong University,Haidian District,Beijing 100044,China;China Nuclear Power Engineering Co.,Ltd.,Shenzhen 518117,Guangdong Province,China;Guangxi Fangchenggang Nuclear Power Co.,Ltd.,Fangchenggang 538001,Guangxi Province,China)
出处
《中国电机工程学报》
北大核心
2025年第11期4336-4350,I0018,共16页
PROCEEDINGS OF THE CHINESE SOCIETY FOR ELECTRICAL ENGINEERING
基金
中央高校基本科研业务费专项资助项目(2024YJS051)
中广核集团公司科技项目(E23GY00050)。
关键词
数字孪生
进相深度极限
安全边界
混合驱动
工况预测
digital twin
phase leading limit depth
safety boundary
hybrid drive
working condition prediction
