Il ruolo della digitalizzazione nell ottimizzazione del processo di manutenzione

Transcription

1 Il ruolo della digitalizzazione nell ottimizzazione del processo di manutenzione G. Guido, V.P.Operation&Maintenance N.Mazzino, V.P.Digital Railways and Innovative Technologies AICQ, Firenze 30 novembre 2017

2 2 Digitization, digitalization and.. digital transformation These terms are often used as synonims but they indeed have a different meaning and span Digitization is transformation in digital format of paper documents, signals, and data Master title Digitalization is the transformation of processes, functions and activities based on the availability of digital data Digital transformation is the overall effect of the digitalization on the business/customers/activities of a company or systems leading to a new definition of business models/operating procedures/ manufacturing processes

3 Digitalization = data and communication network Digitalization relies on the availability of manageable data Data are generated by different sources (equipment, people, operating machines, vehicles, etc..) in different locations. Their usability requires the possibility to aggregate and transport these data Digitalization can occur only if adequate and dependable communication means are available to transport the data Aggregation, analysis and elaboration of the collected data allows their transformation into 3

4 What can we expect from Digitalization Digitalization is expected to bring benefits in multiple areas/sectors: enhanced customer experience by offering better and added value for customers Smart Ticketing and intermodal mobility Human Flow Passenger information and On board entertainment systems Security Integrated Rail Operations by integrating information about resources and services Intelligent Traffic Management System embedding Rolling Stock and Crew management Dynamic Headway Higher safety for workers Cost reduction by collecting real time information about asset status Asset Management and Predictive Maintenance Spare parts and stock management Maintenance training 4

5 Which is the context behind? Successful stories in Digitalization of the Asset Management coming from other sectors (e.g. aerospace) made this concept a «must» also for railways and metro. Ansaldo STS started working in this field by setting up several research projects, gaining an important know-how for stepping into production of predictive maintenance solutions. To setup this new framework it is necessary to run in parallel for: Collecting and analysing data; Defining new optimized process; Creating new skill and competences. Signalling and Automation systems are a mine of data/information, but to make them part of a digitalized Intelligent Asset Management system actions are needed. 5

6 Intelligent Asset Management: the main Goals This presentation aims to show you the process we are following for upgrading our systems in order to achieve an Intelligent Asset Management solution to increase the monitoring, management and maintenance of the most important assets, also paving the road to «predictive» functionalities. Optimize processes Use data for events correlations (Big Data) Standardize Assets/Components Mapping and nomenclatures Minimize Risks Shift from corrective to on condition /predictive maintenance 6

7 Condition Based / Reliability Centered Maintenance based on big data analytics reduces maintenance cost and increase system availability 3.4 Asset Management The goals of these processes are: Gathering vast quantities of data Using predictive analytics to increase reliability Improving Train and components design Optimising maintenance operations and logistics Minimizing spares stock 7

8 1. Intelligent Asset Management the concept

9 Intelligent Assets Management for Optimized Decisions Decision-making, Strategies & Execution Information + Extracted Knowledge Maintenance Decisions Data Mining, Big Data & Predictive Analytics Diagnostic & Monitoring Systems Existing Asset Registers (and similar DBs) Signalling & Train Control Systems Other Dynamic Rail Information External Information (e.g. weather data, etc.) 9

10 Iterative process for building an Intelligent Asset Management System System Analysis Issues Identification Data Availability Heterogenous Data Iterative Lab-Testing Process Data Collection, Understanding & Preparation Refinement (if needed) Functionalities Implementation and testing Asset Status CBM Predictive Maintenance IAMS Functionalities Deployment 10

11 The first specific applications of the IAMS approach are focused on systems operated and maintained by AnsaldoSTS (metro systems first and new generation). On these two different systems the process described before was applied for performing: Preliminary analysis; Data collection, data analysis and dashboards visualization; Gap analysis and definition of possible upgrade. 11

12 Metro system (first generation)

13 ONBOARD ATC ATS Operational data ATP ATO WAYSIDE INTERLOCKING PSD SCADA

14 Preliminary Analysis steps Working sessions with the system experts, so to take advantage of their experience to identify recurrent issues and available datasets related to them Collection of available diagnostic data and related to maintenance/repair activities Data processing for the identification of issues (evidence in data of experts feedback) Gap analysis, definition of actions to bridge the gaps and identification of the updated architecture

15 Architecture Upgrade PSD Train Monitoring SCADA SAP Operational data Diagnostic and monitoring data

16 Data Analysis performed on the first Metro System generation SCADA 1 2 Events/Alarms coming from different systems DATA INGESTION (STORAGE AND PRE- PROCESSING) Non-structured data (logs files) have been acquired and stored on the Data Lake. After the cleaning and formatting process data are ready for the analysis step. 4 DASHBOARD TO VISUALIZE RESULTS PROCESS STEPS 3 DATA ANALYSIS to identify possible correlation between different events/alarms

17 Approach to the Analysis

18 Dashboard Visualization When the user clicks on a specific alarm the system calculates the possible causes and shows it in the pop-up screen. The graph represents the relations between the occured alarm and events/alarm occurred in the past to help maintenance team to identify the real failure causes and the proper intervention required.

19 Unstructured Data Structured Data Customized Business Insights and Big Data Use Cases The IAMS Platform Data Engineering Access Integrate Transfor m Profile Cleanse Enrich Advanced Analytics Predictive Analytics Data Mining Machine Learning Data Discovery Reports Dashboard s Charts Portals End-to-End Embeddability Data Lake: a modular, scalable, distributed storage system able to manage large amounts of data. This system could be easily enlarged to cover new assets and other future developments. Data analytics of the data contained in the Data Lake to perform the three main Asset Management functionalities: Asset Status Monitoring Condition Based Maintenance Predictive Maintenance

20 Metro System (new generation)

21 New Generation of a Metro Line In the next slides it will be shown what is currently feasable working on a new generation of metro lines already able to collect, store and make data available for analysis. More in details the IAMS Platform described above, has been applied to improve ASTS Track Circuits (TCS) monitoring and maintenance process. 21

22 Ongoing Activities Predictive Models Analysis of historical data Provide failures report to the maintenance team. Keep track of past failures number, causes distribution over the line. Data visualization techniques and dash board creation. Data-driven analysis to improve maintenance procedures Predict failures occurencies with different time-horizon. Finding «abnormal» beahaviour patterns in TC parameters to identify a degraded assets (anomaly detection) Failures nowcasting to investigate and asses failures causes in a real-time fashion 22

23 Data Sources LOGS FROM AF-GEN II MAINTENANCE REPORTS BINARY FILES FROM AF-GEN II CENTRAL ATC (AUTOMATIC TRAIN CONTROL) EVENTS AND ALARM LOGS 23

24 New Generation Metro Line Track Circuits (TCS) data for maintenance scheduling improvement TCS system available data (events/ alarms/ measures): historical real-time data are (acquired hourly) Interactive dashboard (uptaded each day) is used from the maintenance team to identify line sections or single TCS at risk in order to improve maintenance operations DIFFERENT PARAMETERS FROM TCS BOARD ACQUIRED INTERACTIVE DASHBOARD WITH RESULTS FOR THE MAINTENANCE TEAM PROCESS STEPS DATA INGESTION (STORAGE AND PRE- PROCESSING) 3 ANALYTICS ON NEW DATA COMBINED WITH HISTORICAL DATA Feedback for analysis process refinement Non-structured data (logs files) are acquired each day and stored on the Data Lake. After the cleaning and formatting process data are ready for the analysis step. When all daily data are collected, analysis is performed. Then, another anlysis is performed using also historical data to improve results reliability

25 Types of Analysis y Single TCS parameters trends over time: time series could be analyzed and visualized in different time windows (i.e. daily or monthly). Different analysis performed in order to provide support to maintenance Histograms for cumulative statistics distributions to describe an overall behavior (considering a single parameter) for all TCS involved. x Single TCS Analysis Overall TCS Analysis Failures Reports Bubble chart for failures number to visualize the distribution of failure occurrencies along the line during the selected interval of time. Histogram fo failure occurrencies for analysis and visualization of failed TCS depending on different failures type (different colors).

26 Interactive Dashboard Examples (1) (Failures report summary) Graph derived from the dashboard containing analysis results representing the elaboration of data collected in 3 month. This is used from the maintenance team to visualize failures distribution over the line in order to identify critical line sections. Moreover, the dashboard allows to visualize TCs associated stations.

27 Interactive Dashboard Examples (2) For each specific TCS, it is possible to track parameters trends over time to identify «abnormal» patterns and degrading status. Moreover, it is possible to monitor parameters values in relation with predefined thresholds.

28 Strategic System Level HMI showing high level information related to the status of the system and providing decision support tools based on specific KPIs (e.g. Cost saving, Failure rate, Risk reduction, ) focused on the needs of the each specific final user (e.g. infrastructure Owner, the Infrastructure Manager and the Global Maintenance Service Provider). 28

29 Tactical Subsystem Level HMI showing information related to the status each different subsystem and providing decision support tools based on specific KPIs (e.g. Spare parts availability, Failure rate, intervention procedures, ) and focused on the needs of the Maintenance Scheduler. 29

30 Operational Component Level HMI showing information related to each component and subcomponent of each different subsystem and providing decision support tools based on specific KPIs (e.g. Failure rate, intervention procedures, ) focused on the needs of the Maintenance Crews providing them detailed information about past, present and future status of each components and the related planned activities. 30

31 Conclusions There s no magic! To integrate a IAMS as decision support tools for the maintenance activities an iterative process is needed in order to reach the desired results. Physically upgrading the system, processing the new data acquired, creating an iterative process with all the experts of the systems to reach a concrete result able to pave the way towards a digitalization of the maintenance process.

32 THANK YOU FOR YOUR ATTENTION