What is the fingerprint deviations tool?
The 'Fingerprint Deviations' tool is one of two recommendation features that TrendMiner's 'Recommendation Engine' provides (the other being the 'Correlation Recommendation'), and is implemented under the light bulb icon, situated top right of the TrendMiner user interface.
For guidelines on the use of 'Correlation Recommendations', please see the 'Recommendation Engine' document.
The 'Fingerprint Deviation' tool provides you extended recommendation capabilities, exploiting the power TrendMiner's diagnostic capability. If a tag strays from a Hull (track) of a fingerprint, the 'Fingerprint Deviation' tool will help you identify other tags that also deviate from their hull; providing you recommendations for possible causes behind the deviation.
The 'Fingerprint Deviations' tool is analogous to TrendMiner's 'Layer Comparison' tool. The 'Fingerprint Deviations' tool is a diagnostics device, and is used to troubleshoot any batch, startup or other time dependent process prone to deviation.
Why use the fingerprint deviations tool?
The purpose of the fingerprint deviation tool is to help you identify possible causes behind process issues.
A fingerprint is composed of a number of hulls, some visualized on the focus chart, others undisclosed. The fingerprint is a depiction of a model or golden batch process, developed using tags from previous batches where the feedstock, reactants and assets performed optimally, resulting with the intended end product.
The fingerprint deviation tool works to help identify tags that deviate from the hulls that make up the golden batch fingerprint.
Note: A hull is a course or track created using the tags of good performing processes, normally displayed in the focus chart. Using TrendMiner's process monitoring tool, poor performing processes will seed the formation of tags that stray from preformed hulls. These are described as deviations from the hull, or, fingerprint deviations.
Undisclosed hulls are so, to prevent the visual display of hulls becoming over complicated. If you see a deviation on screen, this tool will enable you to find other tags that have also deviated. Any undisclosed tag deviating will be recommended for investigation, even if the displayed tags are behaving as intended.
How to use the fingerprint deviation tool
- Create fingerprint as normal (see fingerprint creation guide….)
Note: When you save the fingerprint there is currently no indication for which tags are undisclosed.
When you display a saved fingerprint (in whole) on screen, TrendMiner will check for deviations in the behavior of each tag corresponding to those in the fingerprint, as compared to the behavior defined by the hulls in that fingerprint. It is these deviations that will appear in the list of recommendations.
- With a displayed fingerprint, select the fingerprint deviation feature
A drop down panel appears showing results for correlations and fingerprint specific recommendations.
In this instance the fingerprint deviations tags option is selected to reveal recommended tags.
Once you have discovered a possible cause for your process error, it's time to address the issue.