Anomaly Detection for Solder Joints Using beta-VAE


Creative Commons License

ÜLGER F., YÜKSEL ERDEM S. E. , YILMAZ A.

IEEE TRANSACTIONS ON COMPONENTS PACKAGING AND MANUFACTURING TECHNOLOGY, vol.11, no.12, pp.2214-2221, 2021 (SCI-Expanded) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 11 Issue: 12
  • Publication Date: 2021
  • Doi Number: 10.1109/tcpmt.2021.3121265
  • Journal Name: IEEE TRANSACTIONS ON COMPONENTS PACKAGING AND MANUFACTURING TECHNOLOGY
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC
  • Page Numbers: pp.2214-2221
  • Keywords: beta-variational autoencoder (VAE), automated optical inspection (AOI), solder joint inspection (SJI), unsupervised anomaly detection, VAE, VISUAL INSPECTION SYSTEM, CLASSIFICATION, ALGORITHM
  • Hacettepe University Affiliated: Yes

Abstract

In the assembly process of printed circuit boards (PCBs), most of the errors are caused by solder joints in surface mount devices (SMDs). In the literature, traditional feature extraction-based methods require designing hand-crafted features and rely on the tiered red green blue (RGB) illumination to detect solder joint errors, whereas the supervised convolutional neural network (CNN)-based approaches require a lot of labeled abnormal samples (defective solder joints) to achieve high accuracy. To solve the optical inspection problem in unrestricted environments with no special lighting and without the existence of error-free reference boards, we propose a new beta-variational autoencoder (beta-VAE) architecture for anomaly detection that can work on both integrated circuit (IC) and non-IC components. We show that the proposed model learns disentangled representation of data, leading to more independent features and improved latent space representations. We compare the activation and gradient-based representations that are used to characterize anomalies and observe the effect of different beta parameters on accuracy and untwining the feature representations in beta-VAE. Finally, we show that anomalies on solder joints can be detected with high accuracy via a model trained directly on normal samples without designated hardware or feature engineering.