عنوان

Supervised Machine Learning Techniques for Short-Term Load Forecasting

Number

TLpq2315539734

.Language of Text, Soundtrack etc

انگلیسی

Title Proper

Supervised Machine Learning Techniques for Short-Term Load Forecasting

General Material Designation

[Thesis]

First Statement of Responsibility

Amarasundar, Harish

Subsequent Statement of Responsibility

Matin, Mohammad A.

Name of Publisher, Distributor, etc.

University of Denver

Date of Publication, Distribution, etc.

2019

Specific Material Designation and Extent of Item

93

Dissertation or thesis details and type of degree

M.S.

Body granting the degree

University of Denver

Text preceding or following the note

2019

Text of Note

Electric Load Forecasting is essential in today's world for the utility companies to allocate their resources economically and plan accordingly for future consumption based on the demand. Machine Learning Algorithms has been in the forefront for prediction algorithms. This Thesis is mainly aimed to provide utility companies with a better insight about the wide range of Techniques available to forecast the load demands based on dierent scenarios. This Thesis is focused on modelling Supervised Machine Learning Algorithms to come up with the best possible solution for Short-Term Hour ahead Electric Load forecasting. The Data set for this Thesis comprises of Hour ahead Real time Load data from Electrical Reliability Council of Texas from the year 2018. The input Data set has the hourly load values, Weather data set and other details of a Day. The models were evaluated using Mean Absolute Percentage Error (MAPE) and R-Squared (R2) as the scoring criterion. Support Vector Machines yield the best possible results with the lowest Mean Absolute Percentage Error of 1.46%, a R2 score of 92% and the least computation time for the data set used in this Thesis. Recurrent Neural Networks univariate model serves its purpose as the go to model when it comes to Time-Series Predictions with a MAPE of 2.44%. The observations from these Machine learning models gives the conclusion that the models depend on the actual Data set availability and the application and scenario in play.

Artificial intelligence

Computer engineering

Electrical engineering

Amarasundar, Harish

Matin, Mohammad A.

Electronic name

p

[Thesis]

276903

a

Y