The importance of ultrasonic nondestructive testing has been increasing in recent years, and there are high expectations for the potential of laser ultrasonic visualization testing, which combines laser ultrasonic testing with scattered wave visualization technology. Even if scattered waves are visualized, inspectors still need to carefully inspect the images. To automate this, this paper proposes a deep neural network for automatic defect detection and localization in LUVT images. To explore the structure of a neural network suitable to this task, we compared the LUVT image analysis problem with the generic object detection problem. Numerical experiments using real-world data from a SUS304 flat plate showed that the proposed method is more effective than the general object detection model in terms of prediction performance. We also show that the computational time required for prediction is faster than that of the general object detection model.

In order to allocate appropriate budgets and formulate plans to extend the service life of bridges, it is important to improve the quality of inspection data, which is the basis for understanding the condition of bridges, and the use of CIM models is expected. Since the judgment results are based on 2D drawings that show the locations of damage, the drawings do not show the complicated parts of the building. In addition, some of these drawings do not have dimensions, which causes a problem that the checking process takes more time than usual. In this study, a 3D damage drawing support system was developed to solve these problems. The issues and problems in introducing the system to the practice of periodic bridge inspections are summarized through demonstration experiments, and the issues of periodic bridge inspections using CIM models are discussed.

In the field of non-destructive evaluation using ultrasonic and electromagnetic waves, several researches on inverse scattering analysis to reconstruct a defect in materials have been studied since early times. In general, inverse scattering analysis methods using the Born and Kirchhoff approximations have been formulated using difficult mathematical theories. However, there are still many problems to be improved from the viewpoint of practical application in actual nondestructive evaluation. In this study, we attempt to develop a method to reconstruct the shape and location of a defect from received scattered waves using deep learning, which has been attracting attention as a basis for AI(Artificial Inteligence) in recent years. However, in this paper, as a first step in this type of study, the received scattered waves are simulated waveforms created by the time domain boundary element method. The effectiveness of the proposed method is discussed by showing the results of estimating the location and size of a defect in solids by deep learning using the scattered waves from a defect as supervising learning data. The deep learning results indicate that the accuracy of defect shape reconstruction decreases as the defects move away from the received elements.

Regarding social infrastructuresin Japan, the ratio of facilities constructed 50 years ago or more is on the rise. Inaddition, since it is estimated that the number of employed persons in the construction industry will decrease in the future, it will be difficult to properly maintain the social infrastructure. For this subject, methods for improving maintenance and efficiency by utilizing BIM / CIM and point cloud data are becoming common. In addition, a method for product model using point cloud data, in which point cloud data is divided for all of the features and parts, has also been proposed. Therefore, in this research, the method of product model using point cloud data is compared withother data management methods from the viewpoint of the accuracyand freshnessof the data to be maintained. We verified the effectiveness of the point cloud data as a product model method.

The total length of sewer culverts in Japan will be approximately 490,000 km by the end of FY2020, and the country has a huge stock of sewer pipes. In addition, the proportion of sewer culverts with a standard service life of 50 years will increase rapidly. Currently, inspections of sewer culverts are conducted by visual and TV camera surveys, but the total survey length in FY 2018 was only 6686 km, including both visual and TV camera surveys. In this study, we constructed an urgency classification model using a onedimensional convolutional neural network, which is one of the deep learning methods, utilizing the database of culvert deterioration published by the National Institute for Land and Infrastructure Management of the Ministry of Land, Infrastructure, Transport and Tourism (MLIT).

In recent years, with the declining birthrate and aging population and the declining population, the number of vacant houses is increasing everywhere regardless of the region. As a measure to increase the number of vacant houses, the national and local governments recommend the utilization and solve the problem, but the importance of the analysis is also required for the demand and sustainability after the utilization. Therefore, in this study, we will understand the utilization of vacant houses by analyzing the population composition of Hatoyama Town, Saitama Prefecture and the existing facilities in the area. After that, using the modified Huff model, the absorption rate in the target area of the existing facilities in the town is calculated, and the absorption rate in the target area when the vacant house in the area with low absorption rate is utilized for appropriate purposes is calculated, and each GIS By showing above, we succeeded in visualizing the absorption rate for each area for each utilization purpose, and found the future in the utilization of vacant houses in the area.

Pharmaceutical users, such as chronically ill patients, often experience a worsening of their condition when they are unable to use pharmaceuticals. In particular, in the event of a large-scale earthquake disaster, in addition to the damage to the family hospital, the need for medicines in the community increases, and the likelihood of deterioration of the condition of chronic drug users is high. Therefore, the purpose of this study was to provide need-based pharmaceutical assistance to chronic users of pharmaceuticals in the event of a disaster. In Hakui City, Ishikawa Prefecture, pharmaceutical prescription status in the community was ascertained, and by assuming the evacuation of pharmaceutical users in the event of an earthquake disaster, it was determined where and what kind of pharmaceutical needs would exist in the event of a disaster. As a result, many people in Hakui chronically use antihypertensive agents and hyperlipidemia drugs, and we found that demand for these drugs is expected to be high at evacuation centers in the event of an earthquake or disaster.

In Kanazawa City, "Introduction of public bicycle rental system" was mentioned as a concrete initiative to promote the use of bicycles in the "Kanazawa City Machinaka Bicycle Use Environment Improvement Plan" (2010-31). As part of this, after conducting a social experiment (commonly known as Kanazawa Rent-A-Cycle "Machinori"), "Machinori" started operation in March 2012. As a result, it functions as a means of transportation for many citizens and visitors, and the migration of many users has improved. In this paper, we evaluate and consider the port layout of "Machinaka" from the number of employees and the number of accommodation rooms for each town in Kanazawa.

A variety of organisms live on the earth. Scientifically,it is said that there are about 1.75 million species,but it is estimated that 3 million to 111 million species are actually alive. And these organisms live by directly and indirectly supporting other organisms. In addition,many benefits from organisms such as daily diet,medical care,industry,and culture are "resources". In recent years,this biodiversity is being lost due to environmental destruction,etc.,and it is extremely important to analyze geographical information in order to grasp the actual situation. Therefore,in this study,we will analyze the relationship between land use and human activity and the distribution of organisms,and clarify the impact of human activity on the habitat of each species.

Visual inspection is important for the maintenance of bridges. However, the decrease in the working population in the construction industry has become an issue in Japan. In addition, visual inspection is time consuming and dangerous in some cases. Therefore, the efficiency, rationalization, and safety of inspection work are required. The current inspection content need decision making based on the experience of the inspector. This can lead to serious accidents due to human mistakes. Inspection methods that utilize AI and UAV can solve these problems. In this study, we performed automatic damage detection of UAV images using the U-Net model. The balance of the dataset was ensured by focusing only on corrosion. The problem that UAV images have a large proportion of background and are prone to false positives was improved by background reinforced training. This method is to train a U-Net or other semantic segmentation models by standard damage annotated image data, and training it again before use it to real bridge UAV videos by a few background non annotated images to let the background looked familiar to the model. The background reinforced training of UAV images resulted in improved detection accuracy. It is considered that this is because the model learned the characteristics of the bridge and the information around the bridge from the UAV image.

The total length of sewer culverts in Japan is currently about 490,000 km, of which about 25,000 km have a standard service life of 50 years, and this number is expected to increase rapidly in the future. However, the current inspection methods of visual inspection and television cameras require an enormous amount of time to inspect all sewer culverts. To solve this problem, it is necessary to determine the necessity and priority of inspection. In this study, we constructed models to predict the deterioration status of sewer culverts using only specification data and surrounding environmental data, without the need to survey the inside of the culvert to obtain ANN data, using sewer pipe data made available by the National Institute for Land Infrastructure Management of Sewerage Research. Several models with different hyperparameters (e.g., optimization algorithm) and number of hidden layers were created, and their classification performance was compared to determine the appropriate hyperparameters.

This study develops a multilayer neural network model (hereinafter referred to as an "AI model") to predict the deterioration of light fixtures installed in highway tunnels. The AI model is based on the results of an inspection of light fixture installations classified in the "C" category, and uses 15 environmental factors that are considered to affect the degradation rate of light fixtures as inputs, and outputs the progress of degradation at the time of the next inspection. A logistic regression model was also developed. Then, test data that had not been used for training were input to the constructed model to make predictions, and the prediction reproducibility was evaluated. The results showed that the AI model was able to predict lamp deterioration more accurately than the logistic regression model. Furthermore, a sensitivity analysis of the input data was conducted to identify the variables that are important for improving the accuracy of the model.

Gradient Boosting Decision Tree (GBDT), a machine learning technique, is widely used in both practice and research because of its superior accuracy and computation speed. In this research, with the aim of improving the accuracy of repair plans for bridges with longer service lives, the specifications and inspection data of bridges managed by Tochigi Prefecture and GIS data such as climate and topography (GIS data: National Land Information) are combined using convolutional neural networks (CNNs) and GIS as teacher data. The model to determine which bridges are likely to develop deterioration was developed using GBDT by teacher data. As a result of the verification, we were able to construct a GBDT that accurately estimates the presence or absence of damage progression. Based on the estimation results of the constructed GBDT, it was confirmed that grouping bridges and deriving a deterioration curve for each group improves the accuracy compared to the conventional method.

This paper proposes a method for evaluating particle size distribution by excavation sound using machine learning. First, the mass content of artificial beads of different particle sizes was estimated from the excavation sound of the model ground mixed with the beads of different particle sizes. Second, we conducted the same tests with a silica and other sands. The test results showed that the proposed method using machine learning has the potential to estimate particle size distribution of geomaterials.

Our country also increases many practical application examples of deep neural networks. Many research and application examples exist in the social capital infrastructure field. On the other side, we cannot improve to prepare the big data, which is indispensable for deep Learning. Unlike data under certain indoor conditions such as factories, data related to social capital infrastructure outdoors is diverse, and it is expected that database development that expresses that diversity will progress in the future. In this paper, based on such a situation, we applied reservoir computing, which has fewer parameters than deep neural networks, to dam inflow prediction and confirmed its practicality. Although it is possible to secure a certain degree of prediction accuracy, the accuracy is inferior to that of deep neural networks. It was found that it is necessary to devise networks in the future.

In this paper we propose an AI-based method for soil classification from images of soil particles. The AI model was trained by virtual soil particle images created on a computer. Under the condition that soil particles were not in contact with each other, one hundred images of real crushed stones with different grain size distribution were classified into simulated seven categories and the 78% estimate was correct. Although further improvement is needed for practical use, this feasibility study showed the possibility of substituting computer-generated images for training data of actual soils.

This study aims to improve the accuracy of the proposed method to visualize cracks in concrete using pix2pix from radar images by adding quasi-3D information. In a previous study, ground-penetrating radar tests were conducted on concrete specimens in which artificial defects were embedded in varying positions, angles, and sizes, and cross-sectional images including geometric information of the defects and corresponding radar image pairs were obtained. The dataset was applied to pix2pix to construct a model that outputs the cross-sectional image from the radar image. In this study, we further attempt to apply a method to output the cross-section image by applying two radar images, in which the radar scanning positions are shifted by about the maximum size of coarse aggregate in concrete. The proposed method can visualize defects with high accuracy, even in cases where the reflection intensity of electromagnetic waves becomes small, and the accuracy of conventional methods decreases.

In recent years, the damage to bridge piers caused by the local scouring and the resulting bridge failures have been frequent in many parts of Japan as one of the torrential disasters that occur due to heavy rainfall. In order to prevent such disasters, it is necessary to identify bridge piers at high risk of damage, systematically implement preventive maintenance measures, and evaluate the residual capacity of the piers appropriately in order to determine their soundness. In this study, we have proposed a model for predicting river water level using adequate rainfall as an explanatory variable with machine learning and a model for evaluating the residual bearing capacity of bridge piers according to the progress of local scouring. By combining these models, it is possible to evaluate the residual bearing capacity of bridge piers in fluctuation in river water level and the progress of local scouring.

In Japan, a large-scale earthquake disaster is expected to occur enormous human damage in the near future. Among them, damages to vulnerable people will be significant. Therefore, it is necessary to develop a support system that focuses on vulnerable people. Since support for disaster victims begins with collecting the needs of the affected areas, Information transmission system for vulnerable people is needed to develop in recent years. In this study, we propose a method developed by the authors to identify areas where LANDED can be installed, which is the key to understanding needs. In addition, a web-based questionnaire survey was conducted to summarize the usefulness and issues of LANDED.

During the Great East Japan Earthquake and the Kumamoto Earthquake, valunable people such as infants, the elderly, and people with disabilities, were affected. In addition, many people struggles in shelters because relief supplies were delayed. The purpose of this study is to estimate the number of infants evacuated to shelters and the amount of food, diapers, and other necessary goods. Through the analysis in this study, we estimated the number of infants evacuated from each shelters based on the census and administrative data maintained by local governments. In addition, the quantity and quality of supplies for infants needed at each shelter were clarified.

Japan has now entered a super-aging society, and this has led to various problems such as increased medical and long-term care costs, a shortage of hospital beds, elderly care, and long-term care refugees. In response to these problems, there is an urgent need not only to review themedical and long-term care systems, butalso to create communities where elderly people cancontinue to live in the community even if they require long-term care. This study contributes to the construction of an efficient community comprehensive care system by utilizing data from the National Health Insurance Database, which is currently underutilized medical big data, to understand the local environment in which people certified as requiring support live in each detailed community unit, such as a town or district. We calculated the number of persons certified for support by region for persons aged 75 and over, and examined the relationship between the number of persons certified for support and regional characteristics that were quantitatively evaluated. As a result, the number of certified persons varied greatly from region to region, and regions withalarge number of persons certified as requiring support and poor living convenience, which is one of the regional characteristics, were identified.

Several big earthquakes are expected to occur in near future. Then, supporting for victims suffered from earthquake disaster is very important to recover from the damage, and investigation for victim supports has to execute immediately when the earthquake occurred. Authors analyzed the overwrap ratio of the four disaster investigations which are executed after the earthquake disaster. The four investigations are quick inspection, building damage assessment, damage assessment for earthquake insurance and damage assessment for recovery which were executed past several earthquakes. As a result of this analysis, it become clear that quick inspection and building damage assessment has some same inspection contents, and building damage assessment, damage assessment for earthquake insurance has some same inspection contents. Moreover, some inspections will execute under information sharing condition, it is possible to execute the disaster inspection efficiency and rapidly.

In recent years, the aging of Japan's population has made it a major challenge to secure means of transportation in areas where public transportation is not available. Demand-responsive transportation, which is more responsive to users' needs than local buses and community buses, is expected to become increasingly important in the future to solve this problem. Higashimatsuyama City in Saitama Prefecture introduced a demand cab service in December 2015 to provide a means of transportation in areas with public transportation vacancies. Six years have passed since the start of the service, and while demand for the service has taken root, there are also issues such as requests for service improvement and increased project costs.

In this study, a demand heat map was created using the accumulated usage data, and a model was developed to predict demand spots in space and time by applying deep learning image recognition technology using a neural network.