Project Details
[Return to Previous Page]AI and Machine Learning System for Gear Mesh Pattern Analysis
Company: Dana Incorporated
Major(s):
Primary: CMPEN
Secondary: CMPSC
Optional: EE, IE
Non-Disclosure Agreement: YES
Intellectual Property: YES
Problem Statement: Gear pattern operations lack the ability to automatically detect gear pattern alignment causing Noise, Vibration or Harshness out of tolerance conditions. Benefits: Reduce rework due to Noise Vibration Harshness (NVH) tester fallout Eliminate the need for the operator to review images at station Reduce premature failure and warranty claims Project Requirements: Dana to provide 10 (minimum) master sample images for basis of comparison Dana to provide 1000 (minimum) images for use with analysis to build detection system Automated method to check gear mesh pattern against master images Artificial Intelligence and machine learning related Applicable to the following manufacturing facilities: Toledo, Dry Ridge, Fort Wayne (requires vision system), Columbia Project Deliverables: Base scope is limited to operator decision (Pass/Fail) support at Op200 Gear Pattern Verification operation. Create a simple Graphical User Interface for use on a tablet/PC with the following functions: User ability to select multiple files to ingest as “master” good images and train the system Ability to analyze images against the master data set. Images will be generated by the operation Provide a summary report listing all checked files, with the status of Pass or Fail Provide functionality to export the results to CSV file, including pertinent fields (Plant, Operation #, Time stamp, serial #, Pass/Fail, Failure reason/mode) Stretch targets: Provide functionality to define and categorize different failure modes when they are detected Provide capability to search/filter for specific failure modes that are detected by the system Provide data analytics functionality to provide data trending, pareto charts by failure mode type Provide functionality to allow for additional machines/operations to be added to the system.
