Machine Learning Support for Board-Level Functional Fault Diagnosis
The ever-increasing integration density and design complexity of printed-circuit boards are making functional fault diagnosis extremely challenging. The cost associated with the testing, diagnosis and repair is one of the highest contributors to board manufacturing cost. To improve board-level functional fault diagnosis, machine-learning techniques are advocated, which can identify functional faults with high accuracy. In this chapter, we discuss machine learning support for board-level functional fault diagnosis. Section 1 presents an overview of board-level manufacturing tests and conventional fault-diagnosis models. Section 2 discusses the motivation of utilizing machine-learning techniques and the existing machine-learning-based diagnosis models. To address the practical issues that arise in real testing data, Sect. 3 presents a diagnosis system based on online learning algorithms and incremental updates. Section 4 presents a diagnosis system that utilizes domain-adaption algorithms to transfer the knowledge learned from mature boards to a new board. Section 5 concludes the chapter.
