CLIENT/INDUSTRY BACKGROUND
Our client is an American developer, manufacturer, and supplier of products for automotive, commercial, aerospace, marine, rail, and off-road vehicles and power-generation applications. Piston rings are one of the very important elements of the piston used in engines. The chamfer is used to control and scrape oil back down the cylinder wall. Leaving it upwards will allow oil to coke (carbonize) and create a layer of burned oil between the second and top ring set.
PROBLEMS
- Inaccuracy in identifying chamfer on the rings as the dimensions are very small causes false acceptance
- Identifying chamfer on the piston rings is detailed oriented and hence very time consuming
PROBLEM IMPLICATIONS
- Piston rings without chamfers will allow oil to coke (carbonize) and create a layer of burned oil between the second and top ring set
- Black smoke along with exhaust gases from the engine
CLIENT REQUIREMENTS
- To automate the process of inspection to identify the presence of chamfer on the piston rings with the help of machine vision
- To reduce the time of inspection to 2-3 seconds
- To achieve high accuracy in order to reduce/eliminate false acceptance rate
CURRENT PROCESS
The inspection is being done manually by operators.
BUSINESS IMPACT
1. Increase in labor training cost
2. Decrease in profitability due to product inefficiency
3. Increase in COQ(Cost Of Quality)
SOLUTION USING MACHINE VISION AND AI
A camera or set of cameras with appropriate illumination (white square panel) is set up to identify the presence of chamfer on the piston ring. Images are captured and sent to the software (Qualitas EagleEye® Platform) cloud where the training is done using the DL algorithm. Once the program is trained, real-time defect detection takes place, based on which the results are sent to PLC to take action.
The presence of the chamfer would give a broader edge compared to the side where there is no presence of chamfer. This enables the clear identification of the presence/absence of chamfer.
CONCLUSION
A POC(Proof Of Concept) was conducted and the following conclusion was observed –
1. The presence/absence of the chamfer was identified with 99% of accuracy
2. Inspection cycle time is reduced to 1 second
3. The machine vision system has the potential to reduce/eliminate human intervention in the inspection process