Research And Implementation Of Bridge Maintenance Management System Based On Image Recognition And Machine Learning

Since the founding of China,the transportation of our country made great achievements,and the development of bridges has also been further expanded.Prolonged exposure to sunlight and rain,overloading and other irrational use may affect the structural safety of bridges,resulting in peeling,rust,crack and other diseases.If we do not repair and maintain it in time,it may even lead to further deterioration.Effective bridge maintenance and management can not only reduce the frequency and cost of maintenance,increase the service life of bridge,but also guarantee public safety.In this paper,a bridge maintenance management system based on web application frameworks including Spring,Struts2 and Mybatis is proposed.The various kinds of detection data and calculation data of the bridge are classified and managed effectively.In the system,the hierarchical weighting method and fuzzy comprehensive evaluation method are used to evaluate the reliability and rating,and the corresponding proposal maintenance scheme is given,which is convenient for maintenance personnel to repair in time.This paper not only pays attention to the evaluation,calculation and maintenance decision of the bridge,but also makes further research on the detection of bridge disease pictures in combination with image recognition and machine learning.For disease pictures,the local binary patterns algorithm and the rotation invariant feature extraction algorithm of gray level co-occurrence matrix are used to extract texture features of disease together.The random forest classification algorithm is used to classify and train pictures of various diseases,and generate corresponding models.The real-time prediction classification of the disease is realized by calling the relevant interface of Weka,and the whole prediction process is encapsulated into an interface.Wechat Mini Program is used to shoot the picture of the bridge diseases,calling the encapsulated interface to analyze the categories of diseases and returns the classification results,and achieve the direct correlation between the pictures taken and the statistics of diseases in bridges,and dispense manual complex operations.