| 引用本文: | 王玉荣,朱奕飞,汤奕.考虑暂态波形特征的风电机组高电压脱网智能故障溯源方法[J].电力自动化设备,2023,43(3): |
| WANG Yurong,ZHU Yifei,TANG Yi.Intelligent fault source identification method for high-voltage trip-off of wind turbines considering transient waveform characteristics[J].Electric Power Automation Equipment,2023,43(3): |
|
| 摘要: |
| 大规模风电机组高电压脱网会对电力系统的电能质量和安全稳定运行产生影响。提出了一种风电机组高电压脱网故障溯源方法，后续可配合保护设备以快速切除故障，减少风电机组高电压脱网对系统产生的影响。首先基于风电机组高电压脱网时输出波形进行特征分析，构造了风电机组高电压脱网故障特征指标体系；然后采用Gini指数-最大相关最小冗余法对原始指标体系进行筛选，降低了原始指标体系的冗余度；最后采用遗传算法-蚁群优化算法-粒子群优化算法对BP神经网络的权重、偏差初值进行优化，从而保证BP神经网络的溯源准确率。在西北电网中进行了算例分析，验证了所提方法的有效性。与传统机器学习方法相比，所提方法具有更好的故障溯源性能。 |
| 关键词: 高电压脱网 故障溯源 指标体系 Gini指数 最大相关最小冗余法 BP神经网络 |
| DOI：10.16081/j.epae.202208035 |
| 分类号:TM614 |
| 基金项目:中国南方电网公司科技项目(090000KK52190162) |
|
| Intelligent fault source identification method for high-voltage trip-off of wind turbines considering transient waveform characteristics |
|
WANG Yurong1, ZHU Yifei2, TANG Yi1
|
|
1.School of Electrical Engineering, Southeast University, Nanjing 210096, China;2.State Grid Nantong Power Supply Company, Nantong 226000, China
|
| Abstract: |
| The large-scale high-voltage trip-off(HVTO) of wind turbines has an impact on the power quality and the safe stability operation of the power system. A fault source identification method for HVTO of wind turbines is proposed, which can be used as a support to fastly remove the fault by protection devices to reduce the impact on the power system brought by HVTO of wind turbines. Firstly, a fault characteristic index system is constructed to reflect the fault source characteristic based on the characteristic analysis of output waveform on HVTO of wind turbines. Secondly, the original index system is sorted out by Gini coefficient and maximum relevance and minimum redundancy(Gini-mRMR) method. Finally, the weights and initial deviation values of back propagation(BP) neural network are optimized by genetic algorithm, ant colony optimization algorithm, and particle swarm optimization algorithm(GA-ACO-PSO) to guarantee the accuracy of fault source identification. A case study is conducted in the Northwest Power Grid, which verifies the validity of the proposed method. Compared with the traditional machine learning method, the proposed method has better performance of fault source identification. |
| Key words: high-voltage trip-off fault source identification index system Gini coefficient mRMR method BP neural network |
