Fault localization of a switched mode power supply based on extended integer-coded dictionary method. (September 2018)

• Record Type: Journal Article
• Authors: Ye, X.; Chen, C.; Zhai, G.
• Is Part Of: Microelectronics and reliability. Volume 88/90(2018)
• Page Start: 339
• Page End: 344
• Publication Date: 2018-09
• Subjects: Fault localization -- Integer-coded dictionary -- Gaussian mixture model -- Switched mode power supply; Electronic apparatus and appliances -- Reliability -- Periodicals; Miniature electronic equipment -- Periodicals; Appareils électroniques -- Fiabilité -- Périodiques; Équipement électronique miniaturisé -- Périodiques; Electronic apparatus and appliances -- Reliability; Miniature electronic equipment; Periodicals; 621.3815
• DOI: 10.1016/j.microrel.2018.06.107
• Languages: English
• ISSNs: 0026-2714

Abstract

Abstract: Switched mode power supply (SMPS) is widely used in various industrial fields. Due to the complex operational condition, various faults may occur in the SMPS. Localizing the faults efficiently is necessary for the SMPS. Based on the classical integer-coded dictionary (ICD) method, which has been widely used in board-level analog circuits fault localization, this paper presents an extended ICD method to solve the fault localization problem in SMPS. An optimal boundary determination method is adopted in the extended ICD, which can improve the separating ability of each feature. In the paper, faults in SMPS, the available test points as well as the experimental setup are introduced first. Then, the extended ICD method is presented. Finally, the developed method is applied in the SMPS, and the test result shows eight fault states can be isolated only using six features with an accuracy of 92.5%. Highlights: An extended integer-coded dictionary method is used for SMPS fault localization. The core of the method is optimized boundaries derived from the genetic algorithm. Gaussian mixture model is used to represent feature distributions in the method. The experiment test is implemented to show the efficacy of the developed method.