Fuel Injector Spray Inspections Using Vision Builder AI

[ExamplePrograms]

Overview

With the ever increasing focus of worldwide energy usage, the testing of fuel injection systems has become an important factor in the improvement of vehicles’ fuel economy.  Many times, testing is done with high speed cameras which produce images that are manually analyzed one-by-one. This article highlights two fuel injector inspections performed using actual test images and NI Vision Builder AI software to automate the measurements.  These inspections include a spray tip penetration measurement and a fuel consumption measurement using injector opening and closing detection.

See the Parts List for this application now

Spray Tip Penetration Inspection

The measurement of the spray tip penetration of a fuel injector provides the distance of the center plume from the fuel injector to the farthest extension of spray.  It additionally contains a Boolean indicator alerting the user when spray is detected. 

The method for quantifying the spray tip penetration is documented in SAE specification J2715. This inspection begins with an image of a multi-plume fuel injector as shown in Figure 1.  The spray is characterized in Figure 2 by imaging the single plume at a right angle to the camera lens axis, thresholding the image, and defining a coordinate system based off the location of the injector tip.  Edge detection is then performed below the tip to determine when spray is detected.  Once the edge detection is triggered, the center plume is measured from the spray-tip to the bottom of the injector.  The measurement is returned in pixels, but the inspection can also be calibrated to report the value in real-world units.

Figure 1: Raw Multi-plume Fuel Injector Image.	Figure 2: Multi-plume Fuel Injector Image after Processing.

After processing and measuring on the image is complete, an overlay is added.  It quickly provides the operator with a visual of a Boolean indicating the presence of spray and the center plume measurement directly on the image.

Opening and Closing Delay Inspection

The detection of fuel injector opening and closing delays can approximate fuel consumption during each spray event.  This process requires thresholding the images and detecting the time when spray is started and stopped through edge detection and pixel intensity analysis.

To detect the beginning of the spray as seen in Figure 3, a threshold is first performed on the images and the row of pixels below the fuel injector is monitored for line detection.  If at least one pixel with a value above the threshold is detected, fuel has been injected and a Boolean value flags the user to the occurrence of this event.  Finally, the angular spread of the spray at the fuel injector tip is calculated and displayed to the user.  An image showing the results of this analysis is shown in Figure 4.

Figure 3: Raw Fuel Injector Spray Image.	Figure 4: Fuel Injector  Spray Image after Processing.

The detection of when the spray has stopped is less objective because even though the injector has closed, fuel is still present in the images.  In these inspections, “stopped” is a relative term in which some spray is still present in the image but has become irrelevant to the entire amount of fuel consumed.

For this inspection, the detection of when spray has stopped is determined by measuring pixel intensity values for a region of interest below the fuel injector.  As the spray decreases, the intensity measurements increase until they reach a pre-defined threshold indicating that the fuel injector is closed and spray is no longer present.  For this set of sample images, the threshold was manually determined to be set at an intensity value of 105 based on the average pixel intensity values received during a full spray sequence as shown in Figure 5.  Once the average intensity inside the region of interest becomes greater than 105, the user is alerted that spray has stopped.

Figure 5: Average Pixel Intensity Values vs. Time.

Previously, the moment in which spray had “stopped” was determined by an operator viewing the images and arbitrarily choosing the time in which this had happened.  With the automation of fuel injector inspections through NI Vision Builder AI, measurements are calculated faster and more reliably than ever before.

Parts List

See the Parts List for this application now