| Title |
Pattern Classification Algorithm for Electricity Usage Patterns for Planning the Operation of Energy Storage System |
| Authors |
Young-Il Kim ; Sung-Man Choi ; Min-Kyu Baek ; Bok-Deok Shin |
| DOI |
http://dx.doi.org/10.5207/JIEIE.2019.33.12.021 |
| Keywords |
Clustering; Energy Storage System; Load Pattern; Pattern Classification Labeling; Peak Shaving; Scheduling |
| Abstract |
Prediction of electricity usage patterns plays an essential role in managing the usage efficiently. The prediction needs to be done sophisticatedly and accurately in order to operate energy storage system efficiently since the accuracy of prediction has a big effect on the storage operation plan. In recent years, machine learning-based solutions are being developed because regression analyses have limits when electricity usage has irregular patterns. As the accuracy of machine learning-based predictions depends on the quality and quantity of data to be learned, however, preprocessing process that classifies and labels the usage pattern is important. In this study, we suggest a PCL (Pattern Classification Labeling) algorithm to improve the machine learning-based prediction. It analyzes an actual load data to compare the PCL algorithm with K-means algorithm that has been used widely. In the result, PCL algorithm shows less error rate than K-means algorithm does by 12.2%. |