9:15 AM - 9:30 AM
[HTT15-02] On Estimation of Bridge Deterioration based on Natural Environment
Keywords:GIS, Random Forest, Operation and Maintenance, Asset Management
1. Introductioin
In Japan, a significant portion of social infrastructure was intensively constructed during the period of rapid economic growth. These structures are now aging rapidly, with the number of deteriorating assets expected to increase significantly over the next 20 years. In particular, the aging of bridges presents a major challenge in Japan. The percentage of bridges over 50 years old is projected to rise from approximately 37% in March 2023 to 54% in March 2030 and 75% in March 2040.
Additionally, Japan faces the problem of a decreasing birthrate and aging population. The number of engineers needed to maintain the infrastructure of Japan is insufficient. In response, the Japanese government is promoting some countermeasures such as digital transformation and asset management, which can help reduce maintenance costs. However, implementing these policies will require approximately 20 years. Therefore, it is important to prioritize bridge maintenance. The goal of this study is to develop a method for assessing bridge deterioration more efficiently.
2. Purpose and method
The purpose of this study is to estimate the degree of bridge deterioration based on some environmental factors, with the aim of creating a simplified assessment method.
In this study, we conducted an analysis using the Random Forest algorithm, a machine learning technique available in Python. We chose Random Forest due to its ability to identify key influencing factors and because of its ease of interpretation.
In this analysis, we researched bridges in Japan. The object variable is the degree of deterioration. In Japan, the degree of deterioration of bridges is rated on four ranking scale, which was used in this study. Explanatory variables included bridge characteristics and environmental factors. For instance, the length of the bridge, year of construction and distance from coastline.
We specifically sanalyzed bridge in Fukushima Prefecture because Fukushima has various degrees of bridge deterioration. The dataset included 13,563 bridges and was formatted as a CSV file. For creating the dataset, we integrated the information of environmental factors and bridges using GIS. The dataset was split into training data and test data. From the results of the analysis, we created a confusion matrix for evaluating the accuracy of the model and feature importance analysis was conducted to determine the most influential factors affecting bridge deterioration.
3. Result of analysis
This study revealed important factors for estimating the degree of deterioration. Environmental factors, particularly distance from the coastline and wind speed, are important to estimate the degree of deterioration. Additionally, structural attributes such as bridge lengths and year of construction are important in estimating deterioration levels.
The confusion matrix provided information on the model’s accuracy and we considered the relationship between the degree of deterioration and environmental factors.
4. In conclusion
In this study, we attempted to estimate the degree of deterioration of bridges by applying the Random Forest algorithm. This study clarifies the importance of variables such as distance from the coastline, precipitation, and wind speed for estimating the degree of deterioration.
Future research will focus on improving the accuracy of this model by using aerial photography, population data and other environmental factors. In addition, this study has the potential to predict the environmental impacts in the early design stages of infrastructure projects.
In Japan, a significant portion of social infrastructure was intensively constructed during the period of rapid economic growth. These structures are now aging rapidly, with the number of deteriorating assets expected to increase significantly over the next 20 years. In particular, the aging of bridges presents a major challenge in Japan. The percentage of bridges over 50 years old is projected to rise from approximately 37% in March 2023 to 54% in March 2030 and 75% in March 2040.
Additionally, Japan faces the problem of a decreasing birthrate and aging population. The number of engineers needed to maintain the infrastructure of Japan is insufficient. In response, the Japanese government is promoting some countermeasures such as digital transformation and asset management, which can help reduce maintenance costs. However, implementing these policies will require approximately 20 years. Therefore, it is important to prioritize bridge maintenance. The goal of this study is to develop a method for assessing bridge deterioration more efficiently.
2. Purpose and method
The purpose of this study is to estimate the degree of bridge deterioration based on some environmental factors, with the aim of creating a simplified assessment method.
In this study, we conducted an analysis using the Random Forest algorithm, a machine learning technique available in Python. We chose Random Forest due to its ability to identify key influencing factors and because of its ease of interpretation.
In this analysis, we researched bridges in Japan. The object variable is the degree of deterioration. In Japan, the degree of deterioration of bridges is rated on four ranking scale, which was used in this study. Explanatory variables included bridge characteristics and environmental factors. For instance, the length of the bridge, year of construction and distance from coastline.
We specifically sanalyzed bridge in Fukushima Prefecture because Fukushima has various degrees of bridge deterioration. The dataset included 13,563 bridges and was formatted as a CSV file. For creating the dataset, we integrated the information of environmental factors and bridges using GIS. The dataset was split into training data and test data. From the results of the analysis, we created a confusion matrix for evaluating the accuracy of the model and feature importance analysis was conducted to determine the most influential factors affecting bridge deterioration.
3. Result of analysis
This study revealed important factors for estimating the degree of deterioration. Environmental factors, particularly distance from the coastline and wind speed, are important to estimate the degree of deterioration. Additionally, structural attributes such as bridge lengths and year of construction are important in estimating deterioration levels.
The confusion matrix provided information on the model’s accuracy and we considered the relationship between the degree of deterioration and environmental factors.
4. In conclusion
In this study, we attempted to estimate the degree of deterioration of bridges by applying the Random Forest algorithm. This study clarifies the importance of variables such as distance from the coastline, precipitation, and wind speed for estimating the degree of deterioration.
Future research will focus on improving the accuracy of this model by using aerial photography, population data and other environmental factors. In addition, this study has the potential to predict the environmental impacts in the early design stages of infrastructure projects.