Deep convolutional neural network with new training method and transfer learning for structural fault classification of vehicle instrument panel structure

Abstract

Structural defect have been detected by attaching sensors to all possible defect locations. A new method is proposed to enable the identification of structural defect locations with minimal data collection points using a deep convolutional neural network. Transfer learning was used to improve the accuracy of a hard-to-classify task by using a pre-trained model from an easy-to-classify task. To reduce the number of data collection points, it is necessary to learn the spatial information of the structure. To this end, a structure fault classification-deep convolutional neural network (SFC-DCNN) is proposed. It is an end-to-end convolutional neural network. The time-domain input data and convolutional neural network filter have 2 dimensions. With the proposed method, the accuracy of classifying the location of structural defects in a vehicle’s instrument panel structure was verified with a single vibration measurement point where the location is independent of the structural fault location.