Re-engineering the data from multiple UOA sources to create a population-wide dataset in a marine context
Abstract
Most, but not all large ship operating companies, use more than one provider of off line oil analysis for their fleets because the provider is normally the contracted lubricant supplier and it is considered a risk to have only a single source of supply under contract.
This means that without significant co-operation between the stakeholders, the oil analysis and by association all externally provided condition monitoring data cannot be used to review the whole company’s machinery population. The value of a population wide assessment cannot be overstated as it provides the basis for future policy and strategy development.
With the emergence of "big data" approaches and the Industrial Internet of Things, it is becoming more attractive to produce a data centric approach to managing machinery health. Prior to the worlds fleet being fully "sensorized" and data being collected at high frequencies there is a need to build data history and to start to collect data from all assets.
This paper will illustrate one example by using commercially available cloud based Optical Character Recognition software solutions in association with both Marine and Oil Analysis domain expertise, it is possible to build a data set for historical oil analysis data from multiple providers in multiple pdf formats and will also demonstrate how it is possible to automatically parse data from offline reports into any database where the intended insight will come from UOA in isolation or as part of a fusion of useful information.
This means that without significant co-operation between the stakeholders, the oil analysis and by association all externally provided condition monitoring data cannot be used to review the whole company’s machinery population. The value of a population wide assessment cannot be overstated as it provides the basis for future policy and strategy development.
With the emergence of "big data" approaches and the Industrial Internet of Things, it is becoming more attractive to produce a data centric approach to managing machinery health. Prior to the worlds fleet being fully "sensorized" and data being collected at high frequencies there is a need to build data history and to start to collect data from all assets.
This paper will illustrate one example by using commercially available cloud based Optical Character Recognition software solutions in association with both Marine and Oil Analysis domain expertise, it is possible to build a data set for historical oil analysis data from multiple providers in multiple pdf formats and will also demonstrate how it is possible to automatically parse data from offline reports into any database where the intended insight will come from UOA in isolation or as part of a fusion of useful information.
