Abstract

Assuring the efficiency and reliability of electronic systems relies heavily on finding parametric faults in analog circuits, which is especially important in highly precise sectors. A typical problem with conventional fault detection algorithms when working with high-dimensional feature sets is the risk of over fitting, which can lead to higher computing costs and more acute accuracy. In response to these difficulties, this research presents a method, the High-Dimensional Parametric Optimized Approach (H-DPOA), which combines a Support Vector Machine model with a feature selection based on feature importance measure. By lowering the dataset’s dimensionality and making features more significant in fault classification, the H-DPOA improves the model’s performance and interpretability. The features that have been selected are subsequently utilized to sequence an improved support vector machine model, which aims to improve the detection accuracy of minor parametric errors while reducing the number of false positives. By analyzing benchmark analog circuit datasets extensively in simulation, the suggested method is proven to be valid. With less computing time required than conventional approaches, the results show that H-DPOA greatly improves fault identification rates. In areas where accurate fault identification is vital for preserving system integrity, such as telecommunications, industrial automation systems, and automotive electronics, this approach finds key application in fault diagnosis. The findings suggest that H-DPOA is a viable and scalable approach to diagnosing analog circuits, which could lead to research into automated fault detection and predictive maintenance for electronic systems.

Authors

G. Puvaneswari
Coimbatore Institute of Technology, India

Keywords

Fault Diagnosis, Analog Circuits, Feature Importance, Feature Selection, Optimization, Support Vector Machine