Introduction to the Intelligent Asset Management System Mr

Introduction to the Intelligent Asset Management System Mr. Federico Papa (Project TMT leader) Hitachi Rail STS IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 1

Intelligent Asset Management Definition Although there is not an established definition of an Intelligent Asset Management, the terms Asset, Asset Management and Management System are all defined in ISO 55001 and these provide principles which influence the scope of application, including: 1 Asset Management is much broader than maintenance - additionally it includes the operation, renewal and upgrade of the railway assets (possibly including also those that are used to maintain and/or monitor the railway). 2 The Asset Management System is inclusive of, but comprises more than, an IT system – it is the complete set of interrelated elements that enable an organisation to take better decisions and implement them efficiently and effectively. IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 2

IN 2 SMART Architecture Intelligent Asset Management Strategies • Modelling methodology for the formulation of the mathematical framework • Design of a generic framework for decision support in maintenance and intervention planning • RAMS & LCC analysis and Risk Assessment • Particularization in a collection of real use cases for maintenance and intervention planning • Maintenance Execution, Work Methods and Tools Dynamic Railway Information Management System • An IT (Big Data) architecture and its interfaces Standard open interfaces suitable to manage privacy and IPRs • A set of algorithms compatible with (running on) the Big Data architecture Railway Information Measuring and Monitoring System • A set of heterogeneous monitoring systems IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 3

Railway Information Measuring and Monitoring System Data acquisition IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 4

Railway Information Measuring and Monitoring System (RIMMS) Integrated set of cutting-edge on-board and wayside asset-specific measuring and monitoring sub-systems IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 5

RIMMS Correlated Technologies Drones Robotics monitoring application Satellite Technologies Asset digitalization Io. T sensors 2 D and 3 D monitoring Vibration monitoring Fiber optics monitoring Electromagnetic field monitoring IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 6

Dynamic Railway Information Management System Standard Interfaces & Data Analysis IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 7

Dynamic Railway Information Management System (DRIMS) Aimed at defining an innovative approach to existing railway data management, processing and analysis to support intelligent asset management without the need of developing either a new database or yet another asset register. OBJECTIVES Interfaces Anomaly Detection Standard open interfaces to access existing heterogeneous multi-owner maintenance-related data Analytic tools for automatic detection of anomalies Data mining Predictive Models Analytic tools discovering and describing the maintenance workflow processes. Analytic tools predicting railway assets decay IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 8

Data Acquisition and Analysis Develop guidelines for Open standard Interface for Maintenance Data Heterogeneous Data Sources (Railway Assets and Systems) Design of Data mining and Big data Analytics for extracting knowledge from data Data Collection on IT Architecture IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 Analytics for assets status and behaviour modelling 9

Focus on Open Standard Interfaces The Canonical Data Model Definition of guidelines for Maintenance Data including models and data exchange for Railway Asset Management Focus on Definition of a conceptual and structured representation of data that need to be shared between systems Example developed within IN 2 SMART for collecting diagnostic data from legacy signalling system: • Static data of the railways and signaling equipment are modelled by using Rail. ML • Static details of the sensors deployed within the railway infrastructure is modelled by using Sensor. ML • Measurements and status are modelled by using CDM (Conceptual Data Model) to be further developed within IPx IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 10

Data Analytics ANOMALY DETECTION Detect unusual states or indications of future failures. PROCESS MINING Detect differences between planed processes and process executions. IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 ASSET OR PROCESS PREDICTIVE MODELS Predict future asset status 11

Focus on Process Mining Positioning of the three main types of process mining: (A) Discovery (B) Conformance checking (C) Enhancement IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 12

Focus on Anomalies Detection Anomaly Detection represents the data-driven modelling of the normal behaviour of monitored parameters of an asset/component and the consequent detection of anomalous behaviours. Anomaly detection could be seen as a diagnostic virtual sensor generating alerts and to be treated as such (i. e. it is an uncertain measure to be used as a supporting information and not as an alarm sensor). STEP 1 Model the normal behaviour of the monitored parameters IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 STEP 2 Define criteria to identify anomalous observations and test the model STEP 3 Apply the process to new observations for the online identification of anomalous assets 13

Focus on Prognostics and diagnostics for degrading railway infrastructures Nowcasting the process of exploiting past and present uncertain or incomplete data to make deductions about the present IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 Forecasting the process of exploiting past and present data to make deductions about the future 14

Intelligent Asset Management Strategies Decision Making IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 15

Data-driven Asset Management ASSET INFORMATION Diagnosis Forecasting Data Mining and Analytics BEHAVIOUR KNOWLEDGE Optimization PLANNING DECISIONS SENSOR DATA Sensor concepts Lean Execution Railway Infrastructure IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 16

Decision Making System Decision Support System Strategic Asset Mng Plan: Maint. Strategies & Policies Results from Data Analytics Asset Mng Plan Route Asset Plan: ANOMALY DETECTION Asset Mng Plan Route Delivery Plan: PREDICTIVE MODELS Work Orders Delivery plan Implementation of Asset Mng Plan: Dynamic adaptation & response IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 PROCESS MINING 17

Final Process & OC Contribution Decision Making Predictive, risk- and condition- based, reliability centered, integrative maintenance decision support Data Analysis Assets status and behaviour monitoring using anomaly detection, forecasting nowcasting and process mining tasks Data Acquisition Systems and Assets IN 2 SMART & MOMIT Final Event, Naples 10/10/2019 Periodic/Real-time acqusition of System Data with Standardized Format 18

THANK YOU FOR YOUR ATTENTION Contract No. H 2020 – 730539