Fault Diagnosis based on Semi-supervised Global LSSVM for Analog Circuit

摘要

Aiming at the analog circuit performance online evaluation demand of the largest interval principle and underlying geometric structure, two online methods of dimension reduction are proposed for analog circuit performance evaluation from the angle of feature extraction, First, a supervised method of dimension reduction based on Fisher's Linear Discriminant Analysis (LDA) is presented to increase the classification distance largely. This method is a well-known scheme for feature extraction and dimension reduction. However, the incomplete classification will lead to great influence on performance evaluation accuracy. Based on this, another feature extraction strategy by Locality Preserving Projections (LPP) is proposed. LPP should be seen as an alternative unsupervised approach to Principal Component Analysis (PCA). This method properly obtains a local space that best detects the essential manifold structure. In this paper, the fault diagnosis can be recognized via the Global and Local Preserving based Semi-supervised Support Vector Machine (semi-supervised Global LSSVM). The experiment takes a typical Sallen-key low-pass circuit as diagnosis object. In order to prove the effectiveness of the proposed method in this paper, the traditional fault diagnosis method based on standard support vector machine (SVM) is employed also. The diagnosis speed and accuracy are all proved via numerical simulation.