Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/534338
Title: | Development of new imputation techniques using swarm intelligence algorithms in smart meter data |
Researcher: | Hemanth G R |
Guide(s): | Charles Raja, S |
Keywords: | Engineering Engineering and Technology Engineering Electrical and Electronic |
University: | Anna University |
Completed Date: | 2022 |
Abstract: | Smart meters are one of the essential components of Advanced newlineMetering Infrastructure (AMI) that pave the way for establishing a smart grid newlinethat can control and manage energy consumption more efficiently and reliably in newlinereal-time. These smart meters continuously record energy consumption data at a newlinepre-defined sampling frequency. However, due to practical reasons, there will be newlinebad data present in the smart meter data which has to be identified and treated. newlineIf bad data treatment is not being handled properly, then the prediction model newlinedeveloped using the smart meter data will be less accurate and cannot be relied on newlinefurther for analyzing the load behavior of customers. As a result, data newlinepreprocessing is being performed first on the smart meter data to ensure the quality newlineof data for further processing. Also, it is essential to identify the types of bad data newlinethat are present in the smart meter data. One such type of bad data is missing newlinevalues. These missing values can cause major problems when present abundantly. newlineIn general, missing values are treated using a process called imputation. newlineThis thesis primarily demonstrates the bad data treatment for newlinereal-time smart meter data of Thiagarajar College of Engineering (TCE), an Indian newlineeducational institution. Bad data treatment includes both identification and newlineclassification performed using a stage-wise data-driven approach comprising five newlinestages. Also, the missing values have been identified and classified into four newlinedifferent classes, namely, Class 1, Class 2, Class 3, and Class 4 using this newlineapproach. Next, multiple imputation has been performed for each class of missing newlinevalues. newlineThe next part of thesis demonstrates the multiple imputation of newlineClass 1 and Class 2 missing values using conventional imputation algorithms and newlineparticle swarm optimization based imputation algorithms. The imputation newlineperformance has been measured using the performance metric of root mean square newlineerror. The imputation algorithm that produces the minimum value of root mean newlinesquare error has been identif |
Pagination: | xvii,135p. |
URI: | http://hdl.handle.net/10603/534338 |
Appears in Departments: | Faculty of Electrical Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 240.83 kB | Adobe PDF | View/Open |
02_prelim pages.pdf | 2.23 MB | Adobe PDF | View/Open | |
03_content.pdf | 200.79 kB | Adobe PDF | View/Open | |
04_abstract.pdf | 173.49 kB | Adobe PDF | View/Open | |
05_chapter 1.pdf | 861.1 kB | Adobe PDF | View/Open | |
06_chapter 2.pdf | 1.72 MB | Adobe PDF | View/Open | |
07_chapter 3.pdf | 3.08 MB | Adobe PDF | View/Open | |
08_chapter 4.pdf | 3.07 MB | Adobe PDF | View/Open | |
09_annexures.pdf | 1.79 MB | Adobe PDF | View/Open | |
80_recommendation.pdf | 96.54 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: