Non-intrusive load disaggregation based on deep Convolutional network

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Non-intrusive load disaggregation based on deep Convolutional network

ارائه دهنده: Provider: Ali Danaeifard

اساتید راهنما: Supervisors: Dr.Alireza Hatami

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: دکتر محمدحسن مرادی - دکتر محمدمهدی شهبازی

زمان و تاریخ ارائه: Time and date of presentation: 2021

مکان ارائه: Place of presentation:

چکیده: Abstract: In recent years, the energy consumption of residential subscribers has increased to a very high level. Due to the increasing use of energy consumption, the urgent need to find a way to manage energy consumption has been considered. One of the main reasons for not managing energy consumption is that subscribers do not have much information about the energy consumed by electrical appliances in their homes. As a result, if they are aware of their energy consumption and in other words, separate the power consumption of each device from the power consumption of the whole house, they can manage it. To achieve this goal, Mr. Hart proposed a method called intrusive load monitoring. In recent years, this idea has evolved using smart meters and the ability to record electrical signals at different frequencies, and different data mining methods to classify data. Is used, in order to separate the signals and classify them, we need to extract features from the data, which is possible due to complex mathematical methods, and neural networks and machine learning are used to classify the load, but today with the growth Science based on neural network and the deepening of this network and the ability to automatically extract features in deep matching and its higher accuracy, the researcher has paid more attention to the field of deep learning. In this research, deep Convolutional network has been used to disaggregation the household load. Deep neural networks with more layers can do more processing on the input data. Further processing will increase the accuracy of detection and classification. Meanwhile, convolutional neural network with convolution learning input layers will perform the process of feature vector extraction in detecting and classifying signals automatically and appropriately. Variety and design of feature vector extraction methods for classifying signals face many challenges. Therefore, in order to achieve an automated method that is more accurate than the existing methods, the method presented in this research will be based on deep convolution network. Key Words: Non-intrusive load Monitoring , Deep learning , Residential Household load disaggregation

فایل: ّFile: Download فایل