基于主成分分析和马氏距离的光伏系统健康状态研究

Research on photovoltaic system health state based on PCA-MD method

为准确评估光伏系统的性能状态,保障电网的安全运行,首先,以预测与健康管理技术为基础,利用多元统计分析理论,建立基于主成分分析和马氏距离的光伏系统健康状态评估方法,采用主成分分析对原始多维输入变量进行预处理,利用马氏距离表征光伏系统的健康程度;然后,通过旁路实验和模拟灰尘实验验证了该光伏系统健康状态评估方法能够避免PR方法评价结果的"伪正常"现象;最终,通过实验对该光伏系统健康状态评估方法与PR方法进行对比分析。研究结果表明,与PR方法相比,该光伏系统健康状态评估方法能够更加灵敏、准确地反映光伏系统的性能状态。

In order to evaluate PV system performance more accurately and ensure the safety of power grid, first, a PCA-MD evaluation method of photovoltaic system heath state is put forward based on PHM technology and multivariate statistical analysis. The PCA is used to preprocess original multi-dimensional input variables and the MD is used to show health degree of PV system. Then, the PCA-MD method is testified by bypass experiments and dust shelter experiments. The experiments results show that the PCA-MD method could avoid the ＂pseudo normal＂ results of PR method effectively. Finally, the contrast experiments is designed to show the difference of PCA-MD method and PR method and the outcome demonstrates that the PCA-MD method is more sensitive and accurate than PR method.