Using the Fuzzy Integrals for the Ensemble-Based Classification Problem PCB Defects

The production of printed circuit boards is one of the important areas of the electronics industry, as they are the main components of electronic devices and are used in many other industries. The demand for printed circuit boards is constantly increasing, but at the same time the requirements for reliability are also increasing because the quality of printed circuit boards has a significant impact on the fault tolerance and performance of end devices. However, printed circuit boards are prone to many defects, which can cost companies large losses, as a faulty board can lead to undesirable circuit behavior and defects in the final device, therefore, research is constantly being conducted to improve the quality of the production process. In our work, based on a benchmark method for PCB defect detection, we addressed the defect classification problem by investigating the feasibility of applying ensemble methods based on fuzzy fusion using Sugeno and Choquet integrals to combine the estimates of three pretrained deep learning models: ResNet50, DenseNet169, and InceptionV3.