Aiming at estimating the road surface condition with improvement of the accuracy in spatial, this paper proposes a new method to classify road surface condition by considering identification interval based on vehicle system responses. First, the response signals in different vehicle speeds are decomposed by using both Wavelet Transform (WT) and Empirical Mode Decomposition (EMD) techniques. Then characteristics of the signals in both the time and decomposed frequency domain are subsequently extracted. An Improved Distance Evaluation Technique (IDET) is used to select superior features from the characteristics. Finally, a Support Vector Machine (SVM) classifier is applied to determine the road classification. The influences of identification intervals in spatial accuracy are discussed, and an adaptive classification interval was proposed to improve accuracy. The algorithm is validated by using both simulation and experimental results.

Based on the characteristics of GIS insulation and mechanical fault types and practical measurement experience, two typical vibration measurement methods, direct measurement method and natural frequency measurement method, are proposed in this paper. Then, based on the real data of experiment and field measurement, the vibration signals of typical insulation fault and mechanical fault such as spike discharge and end cover looseness are compared, analyzed and summarized, and their respective data characteristics are extracted, which can provide the basis for the determination of GIS vibration fault types.