Anomaly detection in industrial sensing system using recurrent neural networks

Anomaly detection in industrial sensing system using recurrent neural networks


Anomaly detection in industrial sensing system using recurrent neural networks

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Anomaly detection in industrial sensing system using recurrent neural networks

ارائه دهنده: Provider: Mehran biglari khoshmaram

اساتید راهنما: Supervisors: Dr muharram mansurizadeh

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: dr hatam abdoli - dr reza mohammadi

زمان و تاریخ ارائه: Time and date of presentation: 26/2/2024

مکان ارائه: Place of presentation: Amphitheater

چکیده: Abstract: One of the attractive fields of artificial intelligence in recent years in industries is anomaly detection in the sensor network. One of the applications of this field is the diagnosis of electrical motor abnormalities. Due to the high application and importance of electric motors in industries, their service and maintenance is very important. Currently, the service and maintenance of the electric motor is done when it is seriously damaged. This damage manifests itself in the form of unusual sound and vibration that can be recognized by humans. In order to detect the damage of the electric motor early and the need for service, it is possible to learn the electrical and mechanical conditions of the electric motor in healthy conditions by artificial intelligence using various sensors, and in case of deviation from the normal behavior, the situation can be quickly recognized. So far, many efforts have been made in the field of electric motor abnormality diagnosis, but it has rarely been done completely, taking into account all mechanical and electrical defects and realistically looking at the needs of the industry. In this thesis, an attempt has been made to perform abnormality detection by taking into account all the functional conditions of the electric motor, including electrical, mechanical and environmental, and realistically looking at the needs of the industries (reasonable price, reliability and real-time detection of the abnormality without the need to accurately state the source of the abnormality. ) suitable hardware and model are suggested. Due to the low price and industrial nature of microcontrollers or small industrial computers, there is a need for a model compatible with the processing ability and limited memory of these equipments; For this reason, an attempt has been made to propose a model with a small number of parameters, along with high accuracy of anomaly detection, by creating a simplified descriptor of feature vectors. And by building a 3-phase electric motor test operating table, the sensor network and the model are implemented and the test results are mentioned. Finally, the created model classifies the detected anomalies with 99% accuracy

فایل: ّFile: