A smart monitor for pump system fault diagnostics

点击翻译按钮获取中文摘要

One of the fundamental problems facing a wide range of industries is how to effectively identify a machinery defect before it reaches critical levels so as to avoid machinery performance degradation, malfunction, and even catastrophic failures. The current strategy in fault diagnosis is to periodically shut down the machine service for manual inspection. Often, these routine examinations may lead to unnecessary downtime, which will add significant costs to the operation of machines. The objective of this research project, in collaboration with the Bare Point Water Treatment Plant in Thunder Bay, Ontario, is to develop new technologies and a smart monitor for remote diagnostics of pump stations. It aims to improve operation reliability and reduce maintenance costs. To achieve this goal, advanced research and development will be undertaken in multidisciplinary fields including machinery system modeling, signal processing, artificial intelligence, electronics and mechatronics. Specifically, a new universal transducer interface and smart sensors will be developed for remote data acquisition of vibration, current and temperature signals; new signal processing techniques will be proposed for fault detection in electric motors and pumps; a more accurate classifier will be developed for automatic fault diagnosis in water pumps; new training algorithms will be proposed to improve classifier's convergence and adaptive capability to accommodate different machinery conditions; a smart monitor platform will be developed for online condition monitoring of motors and pump workstations. The developed smart sensors, fault detection technologies and intelligent tools can also be applied in other Canadian industries for real-time machinery health condition monitoring, which will help the related companies improve production quality and operational safety, but reduce maintenance costs.