A semi-supervised support vector machines approach for condition monitoring of construction equipment

In this paper, a semi-supervised learning-based method for condition monitoring of construction equipment is developed. The method is suitable for vibration datasets collected from mechanical equipment on the construction site, for which class definitions are difficult to obtain. The collected vibration signals are analyzed in the time and frequency domain, respectively. Combining the statistical features of the vibration data and some expert information to obtain the category labels of extremely few data, the Fast Fourier transform (FFT) of the vibration signal is used for feature extraction to increase the ability of the classifier. Finally, the limited labeled samples and a large number of unlabeled samples are used as training sets to establish a condition monitoring model based on semi-supervised support vector machines. The performance of the proposed method is evaluated on the real datasets which collected on three different mechanical devices. The result shows that the correct classification rates of the method is 98.87%, 97.37% and 95.33% respectively, which proves that the proposed method is suitable for the condition monitoring of multiple mechanical equipment.