1 / 12

Neural Networks in Sensor Fault Correction in Single Degree of Freedom Structural System

Neural Networks in Sensor Fault Correction in Single Degree of Freedom Structural System. AE 790 Intelligent Buildings. Term Project by Jason Borden. Presented May 28, 2002. Presentation Outline. FDNN & FANN Neural Networks Failure Detection Neural Network

zasha
Download Presentation

Neural Networks in Sensor Fault Correction in Single Degree of Freedom Structural System

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Neural Networks in Sensor Fault Correction in Single Degree of Freedom Structural System AE 790 Intelligent Buildings Term Project by Jason Borden Presented May 28, 2002

  2. Presentation Outline • FDNN & FANN Neural Networks • Failure Detection Neural Network • Failure Accommodation Neural Network • Single Mass Structural System • System Layout • System Parameters • Training and Testing

  3. Failure Detection Neural Network • Analyze Data Received in Real-time From Sensors • Automatically Detects Sensor Failures • Present Information to Human, Graphic Interface, or Second Neural Network

  4. Failure Accommodation Neural Network • Receives Data From FDNN • Sensor Failure Detected by FDNN Sent to FANN • Uses Data From Healthy Sensors to Estimate Reading for Faulty Sensor • Sends Estimated Reading to Control System to Maintain Performance & Effectiveness

  5. FDNN & FANN In Conjunction • Sensor Reading Delivered to FDNN • FDNN Determines if Sensor is Healthy or Failed • Sends Reading to Either FANN or to Control System • FANN Creates Estimate of Signal if Needed

  6. Single Mass Structural System

  7. Single Mass Structural System • System Parameters: • Actuator Mass-2924 kg • Stiffness (k)-13900 N/cm • Damping (c)-15.81 N*sec/cm • Input Force (V)-50 kg2*m2/sec3

  8. Training NN for Use in Structural Systems FDNN FANN • Input 100 Sample Velocity Time Histories • Trained to Produce Desired Output for Sample Histories (Magnitude of 1) • Input 100 Sample Displacement Time Histories • Desired Output is Corresponding Velocity Time History • After Training, Network can use Displacement to Determine Velocity

  9. Training NN for Use in Structural Systems

  10. Testing System Integration • Connect Both FDNN & FANN to Sensor & Active Control System • Simulate Controlled Structural Responses, with both healthy sensors and failed sensors

  11. Active Damping Force

  12. Conclusion Demonstrated Value of Applying Neural Networks to Structural Systems Equipped with Sensors and Active Control Systems Any Questions? All graphs and Illustrations from: Ankireddy, Seshasayee & Yang, Henry T.Y. “Neural Networks for Sensor Fault Correction in Structural Control,” Journal of Structural Engineering, pgs 1056-1064, vol 125, no. 9, Sept 1999

More Related