Open Collections

Photovoltaic systems classification and sizing based on the historical power flow data

University of British Columbia

As the development of sensor and data storage technology, more data have become available for analysis. A commercial database stores the power flow data for more than 4000 Photovoltaic systems. Various optimization methods have been researched for reducing the PV system's cost by sizing each system component appropriately. For most of the existing optimization methods, they focus on the computational efficiency, system modelling or data availability. The disadvantage is they always assume the system's type and operation strategy are known. However, in the given database, the system's type is unlabeled. This thesis proposes a method for sizing PV systems based on their historical power flow data stored in the multivariate time-series format. The method is presented in three consecutive steps. In the first step, a validation rule is applied to filter out the problematic PV systems. The systems whose battery monitor is incorrectly installed can also be detected by the Gaussian Mixture Model (GMM) method, and the related data can be fixed afterward. In the second step, seven features are determined to differentiate the PV systems. The GMM method is applied to cluster the PV system based on the proposed features, so we can identify the system's type through visualizing the classification results. Once we know the system type, in the last step, the PV system is modelled mathematically. The Artificial Bee Colony (ABC) method is implemented based on the system model, historical data, and operation strategy to determine the optimal size of each system component. As an example, a stand-alone system is chosen to demonstrate the process that determines the sizes of the PV panel, diesel generator, and battery bank.

Text

2019-07-31

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/71165

Electrical and Computer Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/