摘要
提出基于最小二乘支持向量机(LS-SVM)的电力设备载流故障趋势的预测算法,并采用粒子群优化(PSO)算法对其参数进行优化。采用主元分析法(PCA)对各触点温度序列进行特征分析,在温度分布异常的情况下提取故障的早期特征;以此时刻为起点,采用PSO与最小二乘支持向量机相结合的方法,并结合实时更新的现场温度信息,对载流故障发展的短期趋势和长期趋势分别进行预测。基于实际运行数据的实验结果表明,将长期预测时间裕量大与短期预测精度高的优势相结合,可以对载流故障的发展趋势做出较为准确的预测。
The Least Squares Support Vector Machine （LS -SVM） is applied to predict the trend of current-carrying faults of electric equipments, and the particle swarm optimization （PSO） algorithm is employed for the optimization of algorithm parameters. Firstly, the Principal Component Analysis （PCA） is applied for the characteristic analyzing of temperature series of all contacts in the equipment, and the early fault features are abstracted in case there exists abnormal temperature distribution. Secondly, taking that moment as the starting point, and with the help of the real-time updated temperatures, the hybrid algorithm of PSO and LS-SVM is applied for the short-term and long-term trend prediction of current-carrying faults. Based on actual operation data, the experimental results show that, with the combination of large time margin from long-term prediction and high accuracy from short-term prediction, it is helpful to accurately predict the trends of current-carrying faults.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2012年第10期19-23,29,共6页
Power System Protection and Control
关键词
支持向量机
最小二乘法
粒子群优化
载流故障
温度预测
support vector machine
least squares
particle swarm optimization
current-carrying fault
temperature prediction
