Online Convolutional Neural Network-based anomaly detection and quality control for Fused Filament Fabrication process

Abstract

ABSTRACT Additive Manufacturing (AM) technologies are experiencing rapid growth in the past decades. Critical objectives for the AM processes are how to ensure product quality and process consistency. The detection and correction of part and process anomalies show great promises and challenges. This paper presents an online laser-based process monitoring and control system to improve the geometric accuracy and in-plane surface quality for the AM process. The point cloud dataset obtained from the 3D laser scanner provides the current part height in the Z direction and in-plane surface depth information for each layer. A Convolutional Neural Network (CNN) model is designed with the pre-trained VGG16 model and validated using the monitoring data to effectively classify the in-plane anomalies. Two developed PID-based online closed-loop control systems are implemented which can significantly reduce the height deviation errors between the fabricated part measurements and design values, and correct the in-plane surface anomalies.