Developing Prognostics Algorithms: Data-Based and Model-Based Approaches

Transcription

1 Developing Prognostics Algorithms: Data-Based and Model-Based Approaches Seth DeLand May 9, 2017 MathWorks Automotive Conference 2017 The MathWorks, Inc. 1

2 What is Prognostics?. 2

3 Why Prognostics? Improved operating efficiency New revenue streams Competitive differentiator Source: GE Oil & Gas Source: Oliver Wyman 2015 MRO Survey 5

4 How does it work? Prognostics Algorithm Workflow Access and Explore Data Preprocess Data Develop Predictive Models Integrate Analytics with Systems Business Data Data Reduction/ Transformation Model Creation Enterprise Systems Sensor Data Feature Extraction Model Validation Embedded Devices 6

5 Challenges for Prognostics Development How long will it take to collect fault data? How expensive is it to collect? How complex is the system? 7

6 Sources of Data for Prognostics Development Fleet Experiments Simulation 8

7 Spectrum of Approaches for Prognostics Algorithms First Principles Modeling Data-Driven Modeling Physical Models Kalman Filters Curve Fitting Statistical Methods Techniques for Modeling Dynamic Systems 9

8 Examples 1. Data-based prognostics using machine learning 2. Fault injection and failure analysis using simulation 10

9 Examples 1. Data-based prognostics using machine learning 2. Fault injection and failure analysis using simulation 11

10 Example 1: Data-based Prognostics with Machine Learning Data provided by NASA PCoE 12

11 Different Types of Learning Type of Learning Supervised Learning Predicting Failures Machine Learning Develop predictive model based on both input and output data Unsupervised Learning Discover an internal representation from input data only Anomaly Detection 13

12 Principal Components Analysis what is it doing? 14

13 Different Types of Learning Type of Learning Supervised Learning Predicting Failures Machine Learning Develop predictive model based on both input and output data Unsupervised Learning Discover an internal representation from input data only Anomaly Detection 15

14 Data-based Prognostics with Machine Learning - Takeaways Use machine learning to identify outliers and build predictive models Many choices for algorithms, apps make it easy to compare options Workflow-focused tools help you fine-tune the model to your particular data 16

15 Examples 1. Data-based prognostics using machine learning 2. Fault injection and failure analysis using simulation 17

16 Example 2: Fault injection and failure analysis using simulation 18

17 Run multiple parallel simulations from the parsim command Speed up simulations and simplify workflow Simplifies large simulation runs 19

18 Leverage Parallel Computing with Simulink Reduce the total amount of time it takes to Run multiple independent simulations (E.g. Parameter sweeps, Monte Carlo Analysis) 20

19 Fault injection and failure analysis Takeaways Use simulation when measured data is not available Run what-if analyses to explore scenarios that are difficult to recreate Comparing field data to simulation data can help diagnose cause-of-failure 21

20 Fleet Data for Prognostics Development Server Data Storage Vehicle Data 22

21 Deploying Prognostics Algorithms Server Data Storage Server-side prognostics Vehicle Data Onboard prognostics 23

22 Considerations for System Architecture Server Data Storage Connectivity is fragile Infinite resources on server Limited resources on ECU Some algorithms more viable than others 24

23 Considerations for System Architecture Server Data Storage Reasonable Compromise Hybrid approach split between onboard and server-side Onboard: buffer, preprocess Server-side: classify, take action Recommendation: Think about modularity when designing 25

24 Server-side Prognostics Server Data Storage RESTful Java.NET Python C/C++ Vehicle Data Package algorithms for MATLAB Production Server 26

25 Onboard Prognostics Server Data Storage Vehicle Data Generate standalone C/C++ code using Simulink Coder 27

26 Key Takeaways No one-size-fits-all approach to prognostics. Prognostics system architecture is evolving. MATLAB and Simulink provide a platform for developing prognostics algorithms. 28

27 Learn More mathworks.com/big-data mathworks.com/machine-learning Example: Model-based approach 29