Typically for any manufacturing company, production quality and yield are two important performance metrics. Manufacturing processes typically include one or more steps where the product is visually inspected for defects. Typically, visual inspection is a highly manual process that can be time consuming and prone to errors. Poor production quality control results in significant operational and financial costs in the form of reworked parts, scrap generated, reduced yield, increased work in process inventory, post-sale recalls, warranty claims, and repairs. After a detailed analysis and research, Ai labs developed and implemented a Proof of Concept (POC) using its proprietary engine Minsky™ to highlight the defects of casting product operations. The objective of this case study is to automate inspection process and identify the defects by categorizing them as (Defective, Pass) in the casting process using deep learning classification model.
After comprehensively evaluating the client‘s challenges/data, we used Minsky™ to accurately model the historical data which comprised of image data and were labelled, classified into two categories.
Minsky™ automated visual inspection tasks using Neural Network models to detect product defects in the casting process thus enabling manufacturers to automate the quality control processes. Historical labelled image data was used for detailed for the selected deep learning Algorithms, real time image data from the production line was fed to the trained Minsky™ models to predict and classify the images as Defective or Pass for casting process.
Based on the data from Fig 1 above shows the results for Defective Disc Brake ( Cast_Defect: 99%)along with best matching image. Our analysis showed that CNN (88.0%) gave the highest model accuracy.
Based on the data from Fig 2 above shows the results for Non-Defective Disc Brake (Cast_Ok 88%)along with best matching image. Our analysis showed that CNN (88.0%) gave the highest model accuracy