Data Mining of Smart Metering Data for Abnormality Detection in Electric Energy Consumption

Soleymani, Mohammad | 2021

218 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54282 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Safdarian, Amir
  7. Abstract:
  8. The development of smart meters enables gathering and analysis of a large amount of data about electrical energy consumption in electric power distribution systems. This data and the obtained behavioral patterns of customers have a wide variety of applications. To name a few, classification of customers based on their consumption patterns, damaged smart meter identification, non-technical loss identification and measuring participation rate of customers in demand response programs are among the applications. So far, many studies have been done for consumption pattern identification. However, abnormality detection in electric energy consumption has captured growing attention due to the difficulty in distinguishing the behavior pattern during an abnormal time. Abnormality is broadly defined as any unusual electricity consumption instance or trend that falls outside of the normal power consumption patterns for a load or load sector, whether in terms of magnitude, time, duration, etc. Therefore, abnormality detection approaches in electric energy consumption using data mining of smart metering data have been proposed to identify time of occurrence and location of the abnormality in power distribution systems. Advanced classification methods such as regression and k-NN, along with an innovative method have been used to detect time of abnormality as the first aim. Furthermore, an advanced statistical method based on conditional correlation has been proposed to detect the distribution network configuration. Identification of anomalous lines as the second aim has been evaluated based on deep learning method. For this purpose, a deep neural network has been proposed by considering emerging algorithms and influencing distribution network configuration on connections of the neurons. Comparing abnormality detection models, three main types of learning in machine learning namely supervised, unsupervised, and reinforcement learning are studied. Receiver operating characteristic (ROC) curve and the indices obtained from the confusion matrix represent the very desirable efficiency of the proposed methods for detecting the time of occurrence of the abnormality. Also, the appropriate speed and accuracy of deep learning method in order to locate anomalies in electrical energy consumption is another observation of the study. It is worth mentioning that the results of the study can be used in energy management systems at the level of the electric power distribution systems to provide strategies and achieve energy efficiency goals
  9. Keywords:
  10. Data Mining ; Electrical Distribution System ; Supervised Learning ; Smart Power Grid ; Anomaly Detection ; Energy Consumption ; Smart Meter System

 Digital Object List