A Bridge Health Diagnosis Approach Base on Deep Neural Networks

Abstract

With the development of transportation, bridge health monitoring is of vital importance as damages in bridges lead to heavy casualties and property losses. We present a approach to assist bridge structure monitoring system to evaluate the health of bridge structure. In this approach, temperature and stress value of bridge status is initialized as input, and bridge deflection degree is set as output. As the bridge data obtained from the monitoring system, a deep neural networks (DNN) model is proposed to identify bridge healthy status while effective parameters of this model is optimized through parameter and hidden layer adjustments. In addition, the influence by multiple factors in the neural network of our approach is considered as only a factor is discussed in most models. We used multiple factors to get a better estimate. Experimental results successfully verify the effectiveness of the proposed method.