| With the continuous development of the scale of construction projects,the problems of complex construction site personnel,broken terrain environment and cumbersome management process are more and more obvious,which makes the monitoring and management of site personnel behavior more and more difficult.On the other hand,the unsafe behavior of personnel in the construction site is a relatively common phenomenon,and the occurrence of construction accidents is mainly caused by the unsafe behavior of people or the unsafe state of objects,and the proportion of the accidents caused by the unsafe behavior of people is up to more than 80%.Therefore,how to improve the construction site image information collection ability,to effectively identify and warn the unsafe behavior of personnel in the construction area,so as to reduce the occurrence of safety accidents,improve the construction site construction efficiency and construction quality has become the focus of engineering practitioners.On this basis,as a new type of environmental restoration construction project in recent years,large abandoned mine hotel project is characterized by broken terrain and large ups and downs,which makes it more difficult to lay out traditional measuring points and image acquisition.Moreover,existing studies lack application cases for monitoring the behavior of construction personnel in this project.Based on the above problems,this thesis starts from the problem that construction site personnel behavior monitoring is easily affected by construction terrain and other factors,carries out research from the aspects of statistical analysis of unsafe behaviors of construction personnel,optimization principle of measuring points,identification algorithm of unsafe behaviors,UAV image acquisition and construction of unsafe behaviors monitoring platform,and takes Tangshan Pit Park as a specific case.Optimize the application for the large abandoned mine hotel project.The specific research content and relevant conclusions are as follows:(1)In view of the problem that the behavior monitoring of construction personnel is affected by site factors at present,the optimal arrangement of unsafe behavior monitoring and measuring points based on statistical analysis is proposed,and based on this optimization arrangement,the production safety accidents of housing and municipal engineering in China during the nearly 8 years from 2014 to2021 are analyzed statistically.The monitoring level of the six common unsafe behaviors of construction workers is,from high to low,1.High work without a safety belt 2.In the edge of the unprotected high work 3.In the construction area without a helmet 4.Do not take the safe passage(short road,path,over the railing)at the operation site.When walking and standing under lifting objects,six types of unsafe behavior monitoring are obtained.The scope of monitoring mainly includes places with large topographies,roads outside the construction area and areas with construction lifting devices.Based on the results of statistical analysis,a monitoring scheme based on statistical analysis of the optimal arrangement of unsafe behavior detection points and UAV image acquisition is determined.(2)In view of the problem of how to identify unsafe behaviors of construction personnel,CNN-LSTM human movement behavior recognition and YOLOv5 object detection are adopted for identification.The two algorithms were verified by jumping over the railings and whether to wear the safety hat.The test results show that the recognition accuracy of the human action behavior recognition algorithm based on CNN-LSTM is up to 95.8%,and the continuity recognition effect of the action is good,and the probability of misjudgment is low for similar actions.In addition,the recognition efficiency of CNN-LSTM recognition algorithm can reach 30 frames /s,which can meet the requirements of real-time recognition in construction engineering monitoring platform.The object detection and recognition method based on YOLOv5 algorithm can successfully recognize human bodies wearing and not wearing helmets,and has good judgment ability for overlapping images,and can accurately recognize,classify and track marks in continuous videos.In this experiment,the accuracy rate is97.2%,which can meet the requirements of application of recognition algorithm in construction engineering monitoring.(3)As for the remaining monitoring blind areas caused by height difference after the monitoring layout of the construction site,UAV image information acquisition means should be adopted to supplement them.Taking Erudite Building in Nanhu Campus as an example,the feasibility of UAV image information acquisition scheme was verified.The site 3D model and site elevation information files in TIFF format were successfully obtained through oblique photography image acquisition.Then the altitude information file is imported into the remote control of UAV,and the simulated flight experiment is carried out,and the image taken according to the altitude information is obtained successfully.The accuracy of the altitude information and the feasibility of the acquisition scheme of the flight image information are proved by the flight altitude data displayed on the screen of the UAV remote control and the information of the shot image.(4)The large abandoned mine hotel project is formed on the basis of the abandoned mine rehabilitation project,with typical construction environment features of large height drop and broken terrain,leading to difficulties in personnel behavior monitoring.This thesis takes Tangshan Mine Park as an example to study the application of monitoring scheme,and uses the principle of optimal arrangement of unsafe behavior measuring points based on statistical analysis to arrange measuring points,and compares the monitoring effect with the original scheme.By comparison,it can be seen that the optimal arrangement has an obvious optimization effect on the monitoring length of the cliff wall,increasing by 40.33%,which enhances the monitoring ability of unsafe behaviors at large height difference,indicating that the optimal arrangement has a good monitoring coverage effect,and it also proves that the optimal arrangement principle of unsafe behavior measuring points based on statistical analysis has an optimization effect.(5)In order to further improve the information collection function in the practical application of engineering,this thesis introduces the optimized design of "Intelligent Helmet" in the Tangshan Mine Park project to further supplement and optimize the identification platform of unsafe behaviors of construction personnel.Through the data call test of "Intelligent Helmet",it is proved that it has a good optimization effect on the information acquisition front end of monitoring platform.At the same time,the comparison between the lightweight identification model and the original model puts forward some suggestions for the improvement of the monitoring scheme: in the specific application,the resolution of the acquisition equipment is still 640,and the original model and the lightweight model can be used together to further improve the monitoring and early warning effect.This thesis has 59 figures,13 tables and 98 references. |
PDF Full Text Request