| 引用本文: | 杨志淳,靖晓平,乐健,沈煜,张好,杨帆.基于MI-PSO-BP算法的配电设备状态实时评估方法[J].电力自动化设备,2019,39(12): |
| YANG Zhichun,JING Xiaoping,LE Jian,SHEN Yu,ZHANG Hao,YANG Fan.Real-time condition assessment method based on MI-PSO-BP algorithm for distribution equipment[J].Electric Power Automation Equipment,2019,39(12): |
|
| 本文已被：浏览 4097次 下载 1719次 |
|
|
|
| 基于MI-PSO-BP算法的配电设备状态实时评估方法 |
|
杨志淳1,2, 靖晓平3, 乐健4, 沈煜1,2, 张好4, 杨帆1,2
|
|
1.国网湖北省电力有限公司 电力科学研究院，湖北 武汉 430077;2.国家电网公司高压电气设备现场试验技术重点实验室，湖北 武汉 430077;3.湖北华中科技电力开发有限公司，湖北 武汉 430077;4.武汉大学 电气与自动化学院，湖北 武汉 430072
|
|
| 摘要: |
| 为了提高配电设备故障预测水平，提出了一种常规综合评估方法与实时评估方法相结合的配电设备运行状态实时评估方法。给出了两阶段综合状态评估方法的框架体系，通过互信息理论(MI)量化设备各属性与状态的相关关系，消除冗余属性。利用粒子群优化(PSO)算法对BP神经网络权值与阈值进行优化，以提高评估质量。利用该MI-PSO-BP模型对某地区配电变压器实时状态进行评估，评估结果及发展趋势与实际情况相吻合，验证了该评估方法的正确性和有效性。 |
| 关键词: 配电设备 实时评估 互信息理论 粒子群优化算法 BP神经网络 |
| DOI：10.16081/j.epae.201911008 |
| 分类号:TM732 |
| 基金项目:国家电网公司总部科技指南项目(521532180007)；国网湖北省电力有限公司电力科学研究院重点科技研发项目(52153217000T) |
|
| Real-time condition assessment method based on MI-PSO-BP algorithm for distribution equipment |
|
YANG Zhichun1,2, JING Xiaoping3, LE Jian4, SHEN Yu1,2, ZHANG Hao4, YANG Fan1,2
|
|
1.Electric Power Research Institute, State Grid Hubei Electric Power Co.,Ltd.,Wuhan 430077, China;2.Key Laboratory of High-voltage Field-test Technique of SGCC, Wuhan 430077, China;3.Hubei Huazhong Technology Power Development Co.,Ltd.,Wuhan 430077, China;4.School of Electrical Engineering and Automation, Wuhan University, Wuhan 430072, China
|
| Abstract: |
| In order to raise the fault prediction level of distribution equipment, a novel real-time condition assessment method is proposed for distribution equipment, which combines conventional comprehensive assessment method with real-time assessment method. The framework of the two-stage comprehensive assessment method is given. In order to eliminate redundant attributes, the relationship between the attributes and the condition of the equipment is quantified by MI(Mutual Information) theory. The weights and thresholds of BP neural network are optimized by PSO(Particle Swarm Optimization) algorithm to improve the quality of assessment. The real-time condition of the actual distribution transformer is evaluated with the propose MI-PSO-BP assessment model, and the evaluation result and trend are coincident with the fault report, which verifies the correctness and effectiveness of the proposed assessment method. |
| Key words: distribution equipment real-time assessment mutual information particle swarm optimization algorithm BP neural network |
