Applications of Semi-Supervised Support Vector Machines in Data Separation Methods in Structural Health Monitoring

Abstract

Structural health monitoring has been aided from the adaptability of statistical damage-detection techniques in this area. We propose using efficient semi-supervised support vector machines to use unlabeled data for classifying between healthy and unhealthy stages. Since support vector machines are a very popular in this area, the semi-supervised SVM s are used to do so. For this reason, a combined model-based and data-based approach is taken to determine the damage sensitive features. To evaluate the performance of classification algorithm, the Precision and Recall criteria for the mentioned algorithms are presented. To compare the effectiveness of the proposed algorithm, different states of the structural response is determined by the labeled and unlabeled data It can be seen that the use of unlabeled data will enhance the effectiveness of the classification methods especially in the lack labeled data. We demonstrate the improved performance of these methods over currently used supervised algorithms in low labeled data situations but, their results are approximately the same with SVM when large labeled data is accessible.