摘要
提出了一种基于稀疏贝叶斯回归的用于检测谐波电流异常值的方法。该方法在稀疏贝叶斯框架下,利用谐波电流数据建立回归模型,以预测值与实测值间的残差作为异常检测的判据。假设噪声模型服从高斯分布,有效抑制了异常数据对回归曲线拟合的影响,从整体上保证了回归的平滑性。对实际工程数据进行分析计算,并对比其他方法,证明所提方法的有效性和准确性。
This paper proposes a new method to detect the harmonic current outliers based on sparse Bayesian regression. In the framework of sparse Bayesian, a regression model is built with harmonic current data, and the residuals between prediction values and measured values are taken as criteria for the anomaly detection. Under the assumption that the noise model obeys Gaussian distribution, the proposed method can effectively inhibit the impact of outliers on the re- gression curve fitting and ensure the smoothness of the results overall. Compared with other methods, the effectiveness and accuracy of the proposed method are proved through the calculation and analysis of practical engineering data.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2017年第5期104-107,共4页
Proceedings of the CSU-EPSA
关键词
电能质量
稀疏贝叶斯回归
谐波电流
异常检测
残差
power quality
sparse Bayesian regression
harmonic current
anomaly detection
residual
