Hyperparameter Optimization of a Faster R-CNN Model For Fault Detection In Quality Control

Quality control in the industry is becoming increasingly important as a result of increasing market competition and the need for quality products. The increasing complexity of industrial processes and the increase in the amount of data available have promoted the development of intelligent systems for automated error prediction/detection, mainly based on Industry 4.0 technologies and especially deep learning methodologies. Therefore, deep learning has an important position for quality control processes. In this study, we propose an intelligent error/fault detection system that will use the Faster Region-based Convolutional Neural Network (Faster R-CNN) and integrates deep neural networks with the particle swarm optimization (PSO) to self-tune the system hyperparameters and improve its performance. The model will be evaluated on different performance metrics such as accuracy, recall, precision, false alarm rate, false negative rate, F1 score and error rate. Finally, the mean average precision of the proposed PSO based Faster R-CNN model is 97.96%. The experimental results illustrated that the PSO based Faster R-CNN has a good accuracy and fast detection ability by a large margin. In conclusion, it can be inferred that the hyperparameter optimization in fault detection systems based on DL model has the high importance and effects.