2020 年 8 巻 1 号 p. 59-70
Local governments manage 70% of all bridges in the Japan. One-half of the number of bridges will reach their service lifetime in the near future. However, local governments have insufficient financial resources, specialists, and technologies to maintain these bridges. Therefore, it is necessary to develop a practical method for the evaluation of bridge soundness to assess the degree of the structural health of a bridge’s superstructure. In this paper, the maintenance priority evaluation of bridge integrity of small and medium span bridges is examined using the support vector machines (SVM) of artificial intelligence (AI) techniques. Based on the model, an algorithm is proposed as a useful feature to provide the engineering expertise for the inspection of bridges. This proposed method was able to substitute engineer judgement to distinguish the health rating of I. The input data were the degrees of deterioration of the structural parts and the output data were the soundness of the structure. As inspection example, 971 inspection data on bridges in Gifu prefecture were used. The results showed that adding the inspection item of exposed direction of steel bars gives good assessment on whether bridge maintenance works are needed or not.