• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.1-16
• /
• 2022

Despite recent breakthroughs in deep learning and computer vision fields, the pixel-wise identification of tiny objects in high-resolution images with complex disturbances remains challenging. This study proposes a modified U-net for tiny crack segmentation in real-world steel-box-girder bridges. The modified U-net adopts the common U-net framework and a novel Self-Attention-Self-Adaption (SASA) neuron as the fundamental computing element. The Self-Attention module applies softmax and gate operations to obtain the attention vector. It enables the neuron to focus on the most significant receptive fields when processing large-scale feature maps. The Self-Adaption module consists of a multiplayer perceptron subnet and achieves deeper feature extraction inside a single neuron. For data augmentation, a grid-based crack random elastic deformation (CRED) algorithm is designed to enrich the diversities and irregular shapes of distributed cracks. Grid-based uniform control nodes are first set on both input images and binary labels, random offsets are then employed on these control nodes, and bilinear interpolation is performed for the rest pixels. The proposed SASA neuron and CRED algorithm are simultaneously deployed to train the modified U-net. 200 raw images with a high resolution of 4928 × 3264 are collected, 160 for training and the rest 40 for the test. 512 × 512 patches are generated from the original images by a sliding window with an overlap of 256 as inputs. Results show that the average IoU between the recognized and ground-truth cracks reaches 0.409, which is 29.8% higher than the regular U-net. A five-fold cross-validation study is performed to verify that the proposed method is robust to different training and test images. Ablation experiments further demonstrate the effectiveness of the proposed SASA neuron and CRED algorithm. Promotions of the average IoU individually utilizing the SASA and CRED module add up to the final promotion of the full model, indicating that the SASA and CRED modules contribute to the different stages of model and data in the training process.

• Wind and Structures
• /
• 제28권2호
• /
• pp.129-140
• /
• 2019

Accurately simulating the wind field of large-scale region, for instant urban areas, the locations of large span bridges, wind farms and so on, is very difficult, due to the complicated terrains or land surfaces. Currently, the regional wind field can be simulated through the combination of observation data and numerical model using observation-nudging in the Weather Research and Forecasting model (WRF). However, the main drawback of original observation-nudging method in WRF is the effects of observation on the surrounding field is fully mathematical express in terms of temporal and spatial, and it ignores the effects of terrain, wind direction and atmospheric circulation, while these are physically unreasonable for the turbulence. For these reasons, a spatial correlation-based observation-nudging method, which can take account the influence of complicated terrain, is proposed in the paper. The validation and comparation results show that proposed method can obtain more reasonable and accurate result than original observation-nudging method. Finally, the discussion of wind field along bridge span obtained from the simulation with spatial correlation-based observation-nudging method was carried out.

• 한국정보시스템학회지:정보시스템연구
• /
• 제18권1호
• /
• pp.145-157
• /
• 2009

If industrial disasters happen at workplace, work's desires of workers and business activities may be dispirited. Especially, industrial disasters frequently happen in the field of heavy and chemical industry companies. According to report of the korea occupational safety and health agency(KOSHA), damage due to industrial disasters is more 5 times than losses due to industrial strifes. Thus, in these manufacturing companies, a safety management field for preventing industrial disasters is emerging as an important factor of business activities. This paper proposes an industrial disaster prediction model for safety management and also implements industrial safety management system(after this, is called as u-Safety system) based on the proposed model using ubiquitous computing technologies like USN(ubiquitous sensor networks) which are given much attention among recent information technologies. The implemented system is successfully operating in the shipbuilding division of 'H' companies.

• 대한안전경영과학회지
• /
• 제7권4호
• /
• pp.61-71
• /
• 2005

Many researches have been focused on the analysis of industry disasters in order to reduce them. As a similar endeavor, this paper provides a propensity analysis of injured people from various industries using classification and regression tree(CART), a data mining algorithm. The sample for this work was chosen from 25,157data related to various industries during one year ( $2003.2\sim2004.1$ ) at Kangwon-Do in Korea. For the purpose of this paper, eight independent variables (injured date, injured time, injured month, type of Injured person, continuous service period, sex, company size, age)are taken from injured person group. According to the analysis result, it is found that five out of the eight factors that are predicted as significant have salient effects. Factors of season, time/hour, day of the week, or month which disasters happened do not show any significant effect. This paper provides common features of injured people. The provided analysis result will be helpful as a starting point for root cause analysis and reduction of industry disasters and also for development of a guideline of safety management.

• 한국건축시공학회:학술대회논문집
• /
• 한국건축시공학회 2017년도 춘계 학술논문 발표대회
• /
• pp.187-188
• /
• 2017

Domestic industrial disasters are decreasing, but construction industrial disasters are increasing every year. So this study draw a conclusions from the major types of safety accidents based on disaster intensity analysis to solve the problems caused by increasing construction industry disasters. Also figure out a risk about original cause material to establish management directions which is significant manage things.

• 한국산업정보학회논문지
• /
• 제19권3호
• /
• pp.67-74
• /
• 2014

Over the last several decades, the United States has experienced a great number of natural disasters. To minimize the impact of the natural hazard events, the U.S. government spent a tremendous amount of money through federal assistant programs. To be eligible for the programs, a mitigation project must be cost effective (more benefits compared to project costs). However, the state and local communities suffering from the natural disasters generally have difficulty in collecting reliable evidence for their damages which can be converted later into benefits when a mitigation project is implemented. Therefore, this paper shows the process of conducting a benefit cost analysis with limited data. Besides, it also provides how to apply the limited data to the analysis through a case study. Consequently, this paper help state and local communities get funding from the federal government, which in turns will improve the image of construction industry by preventing people from natural disasters.

• 환경영향평가
• /
• 제16권5호
• /
• pp.319-325
• /
• 2007

Recently, since the risk on natural disasters is increasing due to abnormal weather such as the global warming, a need for a system on prior review on the influence of disasters has emerged in order to establish a solution by analyzing elements of disaster in advance. However, since the inherently destroying business namely the quarrying business is excluded from the range of subject business of Prior Review System on the influence of disasters, a correction for this is required. In order to actually explore how much risk it contains, actual outflow of soil and flood in the quarrying block where quarrying is being currently carried out was examined and the required undercurrent facility capacity which is also used as a grit chamber was investigated. In addition, by comparing the soil outflow of industrial complexes and golf courses which are current subject businesses of Prior Review on the Influence of Disasters and that of rock mountains relative risk level was examined. After investigation, it was found that the risk on occurrence of disasters was increased due to increase in outflow of soil and flood because of the change of land condition during and after development thus an adequate solution to decrease is required. In addition, after comparison with other business groups it was found that a significantly higher amount of soil is outflown in case of rock mountains thus it was analyzed that a solution to decrease is required. Therefore, a correction is immediately required in order to include quarrying business in the subject business of Prior Review System on the Influence of Disasters.

• 한국안전학회지
• /
• 제15권4호
• /
• pp.47-55
• /
• 2000

In this modern age which electric energy is the fundamental source of energy in the industry, electric shock disasters are unavoidable. Although numerous efforts, time, and money have been invested to prevent such electric shock disasters, the number of electric shock disasters are on the increase. In this study, models for equivalent electric circuit are developed for the different types of electric shock which are classified into three groups. The objective of these models is to calculate the electric current flowing through the body at the time of the shock. Based on the analysis, countermeasures to prevent the shock are suggested. The data used in this study are based on 28 actual incidents which occurred in the Daegu area during January of 1995 through June of 1999. The results of this study can be used as a technical manual for workers treating electrical facilities.

• 대한안전경영과학회지
• /
• 제20권1호
• /
• pp.23-31
• /
• 2018

Industrial disasters cause lots of damages from individuals to companies and country. Especially, damages caused by constructional disasters take very large portion in the whole industry and are accompanied by huge personal physical damages. For the prevention of disasters, roles of safety manager are very important and especially, effects of disaster prevention can be changed by occupational ability of construction safety manager. To improve job abilities of safety manager, job education is very important. This research suggested a model for job education management of construction safety manager and proposed improvements as a study on improvement of curriculum for vitalization of job education of construction safety manager so as to improve occupational ability of safety manager. To achieve them, this research examined standard and operation status of current job education of safety manager and the problems and suggested a model materializing job educational contents as a measure to vitalize job education system of safety manager.

• 수산해양교육연구
• /
• 제26권5호
• /
• pp.1044-1057
• /
• 2014

This study has seen about a indemnity system of fishery damage by natural disasters such as a problem and improvement methods of government aid system and system on accident insurance for cultured fishery products. Recently, in the situation that the demage of aquaculture industry caused by frequent typhoon resulted from global warming and abnormal changed of weather is nasty, the accident insurance for cultured fishery products is necessary to show its true quality and to protect fishers against natural disasters owing to the limitation of government's aid for them. However, The objects of accident insurance for cultured fishery products which is progress on, is too short to apply, so that it is absolutely insufficient to fulfill the demands of culturing fishermen. Therefore, It could be a certain preparation to magnify the range of object items of it and to convert the trial industry being adjusted to limited area into full scale industry to adjust over all nationally. Furthermore, This insurance is complicated and strict to join rather than other insurances. As it can be seen by examples that got in trouble, despite culturing fishermen applied to join the insurance, they took all demage because the insurance was not realized. So, It is hard to say that causes impute the responsibility of it to the authority of insurance, not culturing fishery. They should simplify the registration process, limit the period each registration process and consider a countermeasure to complete it. Concerning compensation for the loss, agriculture part is easier to investigate the loss due to remained dead crop damaged by natural disaster, meanwhile, in fishing part, especially in case of cultivation of fish, it is difficult to investigate the loss and demage because crops are blown all together with typhoon when it comes plus the facilities of them are also very old. Consequently to solve the problem needs more positive attitude as it is policy insurance.