Investigation of Transient Engine Performance Analysis Methods, 09-R8022

Printer Friendly Version

Principal Investigator
Matthew B. Ballew

Inclusive Dates:  01/01/09 – 12/31/09

Background - This project is investigating the application of advanced statistical analyses and decompositions for the interpretation of complex signals found in transient jet engine performance data. The need for these types of analyses arose during a previous internal research project that explored the analysis of transient jet engine performance data for improved Engine Trending and Diagnostics (ET&D) purposes. Although the immediate application of these methods will be for transient engine data, the methods could be applied to data from many different systems operating under various conditions. The methods should also provide for more precise detection and decomposition of steady-state data, which would be an enhancement of capabilities provided by SwRI-developed tools that are currently in use by customers.

Approach - The objective of this project is to apply new statistical tools to analyze and interpret the various complex signals that exist in transient data. Overcoming these challenges in a transient data set will require exploring new methods of statistical analysis and fusing those results with prior knowledge and expertise of the subject system.

Major Tasks include:

  1. Implement Pattern Recognition Algorithms. Project team members have implemented pattern recognition algorithms that detect statistically significant patterns.
     
  2. Implement Variance Shift Detection Algorithms. Project team members have implemented algorithms that detect increases in variance of test data to augment existing mean shift detection algorithms.
     
  3. Other Analysis Enhancements. Project team members have implemented other analysis enhancements including automated assumption checks on data and graphical representations of data in the various stages of analysis. Both allow for more rapid analyses to be conducted.
     
  4. Analyze Data. This task is currently in progress and will include the analysis of mission abort data and corresponding maintenance data.
     
  5. Document Results. The third quarterly report has been delivered and the final report will be completed at the conclusion of the project.

Accomplishments - Preliminary results indicate these advanced detection and decomposition algorithms will greatly aid the analysis of complex signals originating from transient jet engine performance data. These improvements directly influence the detection, diagnostic and prognostic capabilities applied to gas turbine engines and other monitored systems.

Figure 1. This chart shows analysis results from an engine with faulty sensors that cause reduced safety margins. This issue eventually caused a mission abort. SwRI-developed algorithms detected this problem three years prior to the mission abort.

2009 Program Home