IoT oT and and Big ig Data BNS NSF F Railw ailway ay Sunny Bajaj 1
BNS NSF F at a a Glance Glance 32,500 mi of track 28 States/ 3 Canadian Provinces 1600 trains/ day 8000 Locomo>ves 13,000 Bridges 89 Tunnels 25,700 Grade Crossings 10m Carloads Shipped in 2014 2
I nternet o f Things Way ayside ide Det etect ector ors 3
Why hy Way ayside ide Det etect ection? ion? • Improve Safety, Availability, Reliability and Velocity of rolling stock • Augment manual inspec>ons • Reduce train delays associated with setouts • Proac>vely iden>fy “Bad Actors” 4
Det etect ector or Types pes 5 Types of Technologies 13 Types of Detector Systems 2000+ Individual Detectors Infrared T echnology Acoustic Sensors Vision Cameras Excessive friction in wheels Harnessing sounds of a target Recording images of and bearings generates component as it operates under components operating at elevated temperatures that load and speed can provide early track speed is proving an indicate a defect that, if not warnings about defects in a effective and modern way to addressed, can result in component that may not be spot defects that are hard to catastrophic failure. visible, like a crack or an internal identify while a car is sitting defect. in a yard. Force Detectors Laser T echnology It is normal for railcars to Measuring position of impart stable and balanced components forces to the rail, but excessive can provide useful information. impact forces or imbalanced Most recently, this information is forces in curves or straightaways indicate issues being used to plan maintenance that may result in component for locomotive wheels that damage or derailment. require attention and to monitor freight-car braking capabilities. 5
Det etect ector or Examples xamples • Acoustic Bearing Detector (ABD ) – acoustic systems used to evaluate sounds generated by specific bearing component defects • Hot Box Detector (HBD) – evaluates bearing temperature history for statistical outliers; brake issues, burned off journals • Cracked Wheel/Axle Detector (CWAD) – Rail mounted sensors Known Standard Cracked Wheel capable of detecting the difference between tones generated by normal vs. flawed wheels and axles 6 11/4/15
The he Res esult ults 7 11/4/15
Internet o f T hings The he Connect onnectiv ivit ity 8
The he Connect onnectiv ivit ity 9 11/4/15
Establis blishing hing a a Big ig Data a Pla latfor orm… m… and and Explor xploring ing the he Pos ossibilit ibilities ies… … 10
Big ig Data a Pla latfor orm m – – Any ny Data, a, Any ny Wher here, e, Any ny Time ime Analytics WEKA, R, etc. Standard Reporting Predictive & Prescriptive Data Analytics Op>mized for Analy>cs Op>mized for Opera>ons Discovery Sources Visualization Engr. Geo Cars Drones Weather Asset Mgmt Mechanical Detectors Industry Data Video & Audio Structured Data Internet & Log Data Real-Time Streams 11
Pot otent ential ial Bus usines iness Applica pplications ions… … Business Case Analysis PlaJorm Map weather paXers and Hadoop What is the impact of history to BNSF Network. IOC weather on Detector Use Detector data to run DB2 readings? predic>ve models. SAS/R/SPSS Hadoop What is the correla>on Use Geo/Detector Car & IOC between Track Quality and Mech. Sensor data to DB2 Truck/wheel condi>on? determine correla>ons SAP SAS/R/SPSS Combine Geo-Car EAM Hadoop Can we predict and drones data to predict GIS engineering track defects engineering track defects Teradata in advance? in advance SAS/R/SPSS 12
Recommend
More recommend
