Study On Identifying Violating Power Users With Deep Learning
| Autor: | Yen-yu Chen, 陳彥宇 |
|---|---|
| Rok vydání: | 2019 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 107 Beside the normal profitability of power supply companies, illegal users will overload power consumption due to the reduction of electricity costs, increasing the power supply load in the region, which may hinder normal power supply and affect general users. Moreover, in the process of violating the rules, the wiring of wires and the modification of the meter may cause safety concerns, which may lead to accidents such as electric leakage and fire. This article uses the Python design program through deep learning of the neural network to find the characteristics patterns of the normal power user's electricity meter display, and complete the detection of abnormal power consumption data to identify the offending users, hoping to improve the accuracy of the investigation and reduce unnecessary time waste. In this article, deep learning using inverse neural network and long- and short-term memory neural networks are used for training. After proper adjustment of parameters and program settings, both models are showing good discrimination rate, indicating that both types of learning can identify the illegal use to a considerable extent. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|