Analog Circuits Test by Using Principal Component Analysis

Abstract

In view of the difficulties caused by determining threshold for analog circuits test, a method based on principal component analysis (PCA) of node voltages was proposed to overcome these difficulties. At first, the principal component model of fault-free circuits was constructed. Then the circuits-under-test was compared with the principal component model to calculate the statistic for fault detection. The proposed method was used to test the output signal amplifying circuit, which is used in the ultrasonic liquid sensor. The testing results show that the PCA based method has a higher sensitivity than other test methods. And the proposed method can overcome the difficulties in determining threshold by the expert’s empirical knowledge. These make it a suitable candidate for analog circuits test.

What is a Correlation Machine?

“…Some of the applications of the auto-associative memory network include maximizing the number of correctly stored patterns (Masuda et al, 2012a), reconstructing color images (Valle and Vicente, 2012), in neural binding problem (Hayworth, 2012), for memorization (Masuda et al, 2012b) and for diagnosing power transformers (Miranda et al, 2012). Some of the applications of the PCA include in brand power index (Bei and Cheng, 2013), structural assessment of zone subway performance assessment (Yang, 2013), testing analog circuits (Zhang and Chang, 2013), validating metabolic syndrome (Dusseault-Belanger et al, 2013) and in canine hip dysplasia phenotypes (Duan et al, 2013). The EM technique has been used widely for missing data estimation and some of these examples include in modeling forest growth (Mustafa et al, 2012), evolving Electroencephalography (EEG) data (Kim et al, 2011), estimating speed (Ramezani et al, 2011), classifying volume (Yu et al, 2010) and in medical application (Nelwamondo et al, 2007).…”