الفهرس 
INTRODUCTION 1.1 PREFACEOnly 14 pages are availabe for public view
 |  | from | 95 |  |  |
| | | from | 95 | | |
Abstract
Today, almost all the devices used around the world are powered by electrical energy, either as a sole source or a contributor. Moreover, some applications require a continuous electrical power source, and a blackout might cause catastrophic consequences. These consequences can impact personnel safety, such as the safety and firefighting systems. Additionally, from a business perspective, a blackout can result in failure in HVAC systems in the pharmaceutical industry (which requires a relatively large UPS from 500 to 700 kVA). This failure can lead to excursions in temperature, relative humidity, and differential pressures between areas, potentially causing serious losses. Consequently, a reliable uninterruptible power supply is a vital necessity. Online monitoring and diagnosis of the uninterruptible power supply (UPS) is a key player to guarantee the UPS reliability and to notify the system owners prior any issue to arise within the UPS to react and eliminate the impact or reduce it to be minimal. The UPSs and batteries have several types such as dynamic, static … etc.; the scope of the thesis covers the static UPS. The thesis proposes a solution based on the Decision Trees Machine Learning model, that uses data from the IEC and IEEE standards (Voltage, Current and Temperature) to create a basic dataset that will evolve after installing it on the Battery Management System (BMS) and after gathering the data from the system by live online monitoring, also using the temperature of the battery cells as well as the ambient temperature is strongly advised as it gives certainty to the model and it can predict/detect serious breakdowns and unsafe situations such as thermal runaway or even explosions. The reliability, efficiency and simplicity of the model are key attributes that have been kept in mind during the thesis phases.