Equipment health monitoring (EHM) algorithm for electrical devices

• Main Author: Xie, Yihang
• Other Authors: Soong Boon Hee
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78039
• Institution: Nanyang Technological University

This dissertation report proposes a new scheme for fault detection and prognosis in electrical devices with intelligent data analysis method. In the background of industrial 4.0, AI and IoT technology, an industrial revolution was coming all over the world in every field. In this project, the target was using neural network and machine learning technology to applicate in the electrical machine fault detection and isolation. The method was defined as EHM (equipment health monitoring) algorithm which was the key to detect the status of the electrical machine. The scheme aims at developing the system with high efficiency and performance to detect and prognosticate the fault. The system algorithm will be developed based on the neural network and machine learning method combined with feature of electrical devices. The microcontroller unit and the sensor interface will be selected and designed to install a platform to implement the algorithm and function. The fault detection and prognosis system aim at establishing a network which can be implemented in advanced power system. The main significance in the fault detection and prognosis system is that effective electrical devices state management extend beyond the scope of equipment monitoring and go deep into the standardization and optimization of equipment. The development of the system reduces costs which exist in operation and maintenance of equipment. Besides, it increases the lifecycle of electrical devices while still complying the standard. This report concludes the FFT improvement method which reduce the frequency deviation and improve the FFT resolution, the method and test example was shown in the specific section. Besides, the instantaneous power based neural network method for fault detection and isolation II was developed. The raw power data and FFT power data was considered and processed in the experiments. Compared using the current and voltage data, the accuracy and speed of neural network was improved.