Adaptive method for harmonic contribution assessment based on hierarchical K-means clustering and Bayesian partial least squares regression

作者：Zang, Tianlei; He, Zhengyou^*; Fu, Ling; Wang, Yan; Qian, Qingquan

来源：IET Generation Transmission & Distribution, 2016, 10(13): 3220-3227.

DOI：10.1049/iet-gtd.2015.1426

• POINT
• SEPARATION
• CUSTOMER
• SYSTEMS

The accuracy of harmonic contribution assessment is severely affected by the fluctuation of utility harmonic. This study proposes an adaptive method for harmonic contribution assessment of the distribution system with multiple harmonic loads based on hierarchical clustering and regression analysis. First, the utility harmonic impedance is determined by the dominant fluctuation filtering. Then, the utility harmonic voltage data are automatically clustered into different segments based on the hierarchical K-means clustering, which does not require a given clustering number of harmonic data. Subsequently, the harmonic contribution of each data segment is calculated with the Bayesian partial least squares regression, hence providing an adaptive estimation of the optimal principal components. Finally, the total harmonic contribution of the concerned period is obtained by weighted summation according to the length of each harmonic data segment. Simulations are carried out on the IEEE 13-bus and 69-bus systems. The results confirm that the approach can adapt to the fluctuation of utility harmonic and produce assessment outcomes more accurately compared with some other methods.

• 出版日期2016-10-6
• 单位西南交通大学