A Framework for Designing Prognostic Systems

"Prognostics is an emerging field within equipment condition monitoring, which looks to predict the occurrence of a failure ahead of time. When a fault develops there is generally some period of time before failure, where the equipment can continue to function even although its condition is deteriorating. For many years, diagnostic systems have tried to identify the specific fault that is occurring during this period of deterioration.

Prognostic systems aim to move beyond diagnosis, and predict the remaining life of the equipment. With a better understanding of the time until failure, the asset owner can schedule maintenance or replacement more effectively, to extract the full life from the equipment while reducing the chance of a failure in service. This in turn reduces costs associated with periodic maintenance and early replacement of assets.

Within the power industry, the need for accurate prognostics is pressing. National Grid statistics show that the majority of power transformers were installed before 1970, and thus exceed their original design life. Continued service relies on confident predictions about future health, traditionally provided by engineering judgement. At the opposite extreme, new technologies such as HVDC and offshore wind introduce new assets and new uses of established assets, where little operational experience can provide such engineering judgement. Both situations would be enhanced by the widespread adoption of prognostic systems.

Currently, there exist no standards or common approaches to developing prognostic systems. Such systems have been developed for a number of applications, but the lack of commonly-agreed terminology makes it difficult to compare approaches and methods. When faced with a new application, the designer must establish their own needs and requirements without any support or guidance. It is difficult to ensure all possible options have been adequately considered.

This research aims to create a design framework for prognostic systems in the power industry. Such a framework can be used as a methodology for developing new prognostic systems, guiding the designer through different options and design decisions. The framework will also include the terminology needed to describe and compare different prognostic approaches and system components, allowing assessment of advantages and disadvantages of different choices. The existence of a design framework will make it faster and easier to build prognostic systems, leading to wider deployment of this technology, and ultimately better maintenance scheduling."

Within this project we have developed a methodology for designing prognostic systems. Machinery and equipment ages and deteriorates over time, and will eventually reach a point of failure if no action is taken. Prognostic systems can predict the time remaining before failure. An accurate prediction allows the equipment operator to confidently plan maintenance for a point in time before failure occurs. Before this project, there was no standard process or guidelines for how to design a prognostic system. We have identified the key stages required to design and implement a prognostic system, and the important decision points in the process. This will allow future prognostic systems to be designed in a more formal and systematic way, increasing confidence that the resulting prognostic system is fit-for-purpose, and allowing clear comparison between different modelling approaches.