FDD Evaluator

Goal: The FDD Evaluator is intended to quantify the performance of FDD protocols that are intended to be applied to unitary air-cooled air-conditioning equipment. FDD Evaluator implements an evaluation methodology developed by David Yuill and Jim Braun at Purdue University.

Approach: The methodology is described in Yuill and Braun 2013 with updates in Yuill et al. 2014 and Yuill 2014. In summary, a large set of inputs are fed to a candidate FDD protocol, and the results are collected and analyzed. The inputs consist of almost any kind of input that an FDD protocol may require. They represent the kinds of quantities that can be measured on a running air-conditioner. These inputs come from a data library of laboratory measurements of systems operating with, and without fault imposed under a range of different operating conditions. The combination of fault condition and operating condition is referred to as a scenario.

The results from each scenario are classified as one of the following: No Response, Correct, False Alarm, Misdiagnosis, Missed Detection, or No Diagnosis. These classifications are described in detail in the references. The results from all scenarios are aggregated to generate statistics about the general performance. For most of the outcome classifications, the statistics are presented on the basis of impact on performance (capacity or efficiency). Plots showing the aggregated results as a function of performance impacts are provided by the FDD Evaluator. The numerical result summaries are also output in a way that can be copied/pasted for further analysis and storage.

Built-in protocols: There are several built-in protocols in FDD Evaluator. These allow users to familiarize themselves with the FDD Evaluator before using the Custom feature to evaluator their own FDD protocols. The built-in protocols are discussed in the references.

Custom: To conduct an evaluation of your FDD protocol there are six steps.

  1. Download a file describing the available input types, and their associated codes
  2. Upload the file to FDD Evaluator with the appropriate codes entered
  3. Download a unique input file in CSV format
  4. Feed the input values from the CSV file to the custom FDD protocol of interest
  5. Upload a CSV file of the FDD protocol’s outputs to FDD Evaluator
  6. Click “Evaluate” to see the evaluation results

References:

Yuill, D.P., H. Cheung and J.E. Braun. 2014. Evaluation of Fault Detection and Diagnostics Tools by Simulation Results of Multiple Vapor Compression Systems. Proceedings of 15th International Refrigeration and Air Conditioning Conference, Paper 2605, West Lafayette, July 2014.

Yuill, D.P. 2014. Development of methodologies for evaluating performance of fault detection and diagnostics protocols applied to unitary air-conditioning equipment. Doctoral Dissertation. Purdue University. West Lafayette, IN.

Yuill, D.P. and J.E. Braun. 2013. Evaluating the performance of FDD protocols applied to air-cooled unitary air-conditioning equipment. HVAC&R Research, 19(7).


Contact Information:

David Yuill
Select Units
Select Fault Conditions
Enter Operation Condition Limits





Upload Custom Protocol Outputs

Click to upload file.

Please wait...

False Alarm Rates
Misdiagnosis Rates
Missed Detection Rates
No Diagnosis Rates

Note: The confusion plot is only generated if all 'Faults' are selected.

Confusion Percentage Plot