| 引用本文: | 焦鹏悦,杨德友,蔡国伟.基于Koopman算子与卡尔曼滤波的同步发电机动态状态估计[J].电力系统保护与控制,2024,52(9):27-35.[点击复制] |
| JIAO Pengyue,YANG Deyou,CAI Guowei.Dynamic state estimation for a synchronous generator based on the Koopman operator and Kalman filter[J].Power System Protection and Control,2024,52(9):27-35[点击复制] |
|
| 摘要: |
| 动态状态估计是监测同步发电机动态行为的重要手段，准确的动态状态估计结果对于指导电力系统安全运行与高效控制具有重要意义。从数据驱动的角度出发，提出了基于Koopman算子与卡尔曼滤波的同步发电机动态状态估计方法。该方法首先利用汉克尔动态模态分解算法从发电机动态响应数据中提取Koopman算子，进而以提取的Koopman算子为基础构建同步发电机状态空间模型，并利用卡尔曼滤波对同步发电机状态变量进行动态估计。该方法无须事先构建发电机模型及参数，实现了完全数据驱动的动态状态估计。仿真实验结果表明，在发电机模型及参数失配的情况下该方法估计精度明显高于传统以模型为基础的估计结果，具有较好的自适应性和鲁棒性。 |
| 关键词: 动态状态估计 模型 数据驱动 Koopman算子 卡尔曼滤波 汉克尔动态模态分解 |
| DOI：10.19783/j.cnki.pspc.231088 |
| 投稿时间：2023-08-23修订日期：2024-01-26 |
| 基金项目:国家电网有限公司总部科技项目资助(5108- 202299255A-1-0-ZB)；国家重点研发计划项目资助(2021YFB 2400800) |
|
| Dynamic state estimation for a synchronous generator based on the Koopman operator and Kalman filter |
| JIAO Pengyue1,YANG Deyou2,CAI Guowei1 |
| (1. School of Electrical Engineering, Northeast Electric Power University, Jilin 132012, China; 2. School of Electrical
and Electronic Engineering, Harbin University of Science and Technology, Harbin 150080, China) |
| Abstract: |
| Dynamic state estimation is an important means of monitoring the dynamic behavior of synchronous generators, and accurate results are important for guiding safe operation and efficient control of power systems. From a data-driven perspective, this paper proposes a method for estimating the dynamic state of synchronous generators based on the Koopman operator and Kalman filter. The method first extracts the Koopman operator from synchronous generator dynamic response data using the Hankel dynamic mode decomposition algorithm, and then constructs a state space model of the synchronous generators based on the extracted Koopman operator. The state variables of synchronous generators are dynamically estimated by Kalman filter. The algorithm does not require prior construction of generator models or parameters and achieves fully data-driven dynamic state estimation. Simulation results show that this algorithm has good adaptability and robustness and exhibits significantly higher accuracy than traditional model-based estimation results using mismatched generator models and parameters. |
| Key words: dynamic state estimation model data-driven Koopman operator Kalman filter Hankel dynamic mode decomposition |
