LSTM combined with BIM technology in the management of small and medium-sized span highway concrete beam bridges

Abstract

To better carry out maintenance management for medium and small span highway concrete beam bridges, this bridge is taken as the research object. Building information modeling technology is applied to its maintenance management. The corresponding management system is constructed. Through this system, the distribution of diseases and other information can be viewed. Long short-term memory neural network is used to build the corresponding prediction model to predict the technical condition score of bridge components. The results show that the prediction error is small. The minimum error rate in component technical condition scoring prediction is 0.00 %. In the full bridge technology scoring prediction, the minimum scoring error is 2.8 points, and the maximum scoring error is 6 points. The minimum average error rate of this model in predicting component technical condition scores is 0.15 %. Except for individual components, the average error rate of all other components is less than 5 %. The accuracy of the prediction model is 95 %. The performance is superior to linear regression models. The average error rate in the abutment is 7.45 % lower than the linear regression model. This method can effectively predict the degradation of small and medium-sized span highway concrete beam bridges and achieve three-dimensional visualization of diseases.