@ARTICLE{ iet:/content/conferences/10.1049/cp.2016.1188, author = {K. Pipe}, affiliation = {Humaware, Southampton}, author = {B. Culkin}, affiliation = {Humaware, Southampton}, keywords = {state detection methods;prognostics capability;adaptive anomaly detection techniques;predictive analysis;CI alerts;health assessment;prognostic processing;maintenance intervention optimisation;R&D projects;condition indicators;automated data-driven toolset;automated diagnostics;humaware data-driven toolset;decision support components;raw sensor data;risk based maintenance planning;predictive maintenance capability;}, language = {English}, abstract = {Humaware's data-driven toolset has been developed through a number of R&D projects to provide asset managers with a predictive maintenance capability with actionable information to optimise maintenance intervention to predict and prevent asset failure. The paper will explain how each tool relates to the State Detection, Health Assessment, Prognostics and Decision Support components of ISO 13374 to provide a “Detect, Diagnose, Prognose, Action” capability. The paper will demonstrate how Humaware's use of Condition Indicators (CI) i.e. features of the data that relate to defects improve on current state detection methods, which are based on simple thresholds applied to raw sensor data. The technology to provide accurate and robust alerts that identify events in the data such as changes in level and the initiation of trends using adaptive anomaly detection techniques is described. It will be shown that these event based alerts are dependable enough to support automated diagnostics and prognostic processing A data driven Bayesian approach is used to provide accurate diagnosis of the root cause defect based on the CI alerts. It also supports a risk based approach to deriving a Remaining Useful Life to provide a prognostics capability.}, title = {An automated data-driven toolset for predictive analytics}, journal = {IET Conference Proceedings}, year = {2016}, month = {January}, pages = {1 (7 .)-1 (7 .)(1)}, publisher ={Institution of Engineering and Technology}, url = {https://digital-library.theiet.org/;jsessionid=1qfrnv0ftern8.x-iet-live-01content/conferences/10.1049/cp.2016.1188} }