Building asset monitoring and prognostics systems using cost effective technologies for power generation applications

Abstract

Cost effective smart industrial data recorders promise to automate the collection of condition indicating sensor data. Automatic and pervasive data recording creates a wealth of condition assessment data that couples with operational history to yield a data store rich in opportunity for data driven prognostics as well as model development. Storing, managing, scoring, and otherwise utilizing this new found wealth of machinery condition indicators challenges the prognostics designer. Implementation of new and existing prognostic algorithms and techniques in an automated and useful way are the challenge of the day. While the application is not yet complete, this paper describes the motivation, the tools, the vision, and the current state of the power generation prognostics application with over 300 “balance of plant” machines under automatic surveillance.