Predictive maintenance for improving power system assets management

点击翻译按钮获取中文摘要

The new competitive deregulated electric power market is causing fundamental changes to the rules of the game of generating, transmitting and delivering electric energy.  These changes coupled with stringent regulatory demands, aging electrical equipment and reduction in manpower and expertise, make it crucial to operate the power system in a fundamentally different manner. The electrical power systems and market issues should be integrated, with both economical and engineering aspects coordinated. In this new environment asset management and predictive maintenance become key business instruments for efficiently operating the power system. The use of Predictive Maintenance can help organizations to develop a systematic maintenance program, meeting these requirements in a cost effective manner. Predictive maintenance basically combines different techniques and tools in a systematic approach to improve the managing of system assets as a basis for maintenance decisions and efficient operation of the system.   The objectives of this research are twofold: the first goal is to develop  dynamic maintenance models for the different components of the power system such as transformers, circuit breakers, cables, etc. and  the second goal is to develop a strategy to integrate all these maintenance schemes to cost-effectively manage the assets of the power systems.  Artificial intelligent-based condition monitoring techniques will be used in collecting, storing and conditioning of data. Signal processing -based technique will be used to extract features and classified information from the data. The heart of the predictive maintenance technique is the health assessment unit and the prognostics unit. The purpose of the first unit is to generate diagnostic records and propose fault possibilities, while the purpose of the second unit is to calculate the future health of an asset and the remaining useful life of it taking into account the future usage profiles.