Date of Award

January 2017

Degree Type


Degree Name

Master of Science (MS)


Electrical Engineering and Computer Science


Sara Eftekharnejad


K-Nearest Neighbor, Linear Regression, Photovoltaic Forecasting, Probabilistic Power Flow, Symbolic Aggregate approXimation

Subject Categories



With the developments in renewable energy resources, more Photovoltaic (PV) generators are being built. Compared to traditional generators, a PV generator is less controllable which will adversely impact power system operation and planning. To ensue seamless operation of power systems, PV forecasting is essential and necessary. A challenge in PV forecasting is that PV generation behavior differs in different regions due to the fact that PV generation is highly dependent on weather conditions, in particular solar irradiance. This makes it important to study the power output data based on a specific region. In this thesis, I first analyze how PV forecasting will affect system planning by calculating probabilistic power flow (PPF). By using a variety of probabilistic models that can estimate solar irradiance, the PPF of each model is calculated and compared. The PPF will give us an idea of how accurate and inaccurate forecasts will affect power system operations and planning. I then seek to find out which method can forecast the power output more accurately. I used several methods such as Linear Regression, Artificial Neural Network (ANN), K-Nearest Neighbor (KNN) and Support Vector Machine (SVM), and compared these methods in 3-hours ahead forecasting. In addition, these methods are analyzed for future use, as the dataset used is constantly growing. Through my analysis of the data, I found out that, based on a small dataset, linear regression works better and as the dataset grows larger, the error for K-Nearest Neighbor reduces dramatically. In addition, a new approach named Symbolic Aggregate approximation (SAX) was used when an extremely large dataset was used to increase calculation speed and reduce dimensionality.


Open Access

Included in

Engineering Commons



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.