Research On Intelligent Monitoring Method And Application Of The Surface Compaction Of Core-wall Rock-fill Dams

The construction of hydraulic engineering is related to economic development and social livelihood.Since the formulation of the “13th Five-Year Plan” hydraulic engineering reform and development plan,the scale and investment scale of hydraulic engineering has expanded year by year,and significant economic and social benefits have been achieved.The core-wall rockfill dam has become the preferred dam type due to its economic and safety advantages.With the rapid development of science and technology,adv anced technologies such as artificial intelligence,Internet of Thing and big data,the current compaction construction of core-wall rock-fill dam is in the digital stage and is moving towards the intelligent stage after experiencing the stage of artificialization,mechanization and automatic compaction construction.However,the digital compaction of core-wall rock-fill dams faces many challenges in the era of intelligent construction,including incomplete perception of compaction construction information,no depth of analysis,low degree of decision-making and control integration,and the lack of effective intelligent management platform.Therefore,it is urgent to summarize the existing engineering research results and experience,comprehensively carry out the intelligent compaction theory research of core-wall rock-fill dams,and focus on solving the key technologies.The aforementioned measures have important research significance for improving the construction level of core-wall rock-fill dams.Combining the current generation of "Internet of Things +" and "Artificial Intelligence +" and other intelligent technology concepts,the dissertation has carried out in-depth research on the above issues and achieved the following major achievements:(1)The basic concept of intelligent monitoring for the surface compaction of core-wall rockfill dam is proposed,enriching and developing the theory of digital dam.After four stages of construction,i.e.artificialization,mechanization,automatic and digital,the theory of construction management and control of corewall rock-fill dam has been developed to digital dam.On this basis,the basic concepts of construction information model,intelligent dam,intelligent monitoring and intelligent dam construction have emerged successively.These theories provide scientific guidance for compaction construction of core-wall rock-fill dam from the aspects of perception,analysis,feedback control and monitoring system.However,the surface compaction construction of core-wall rock-fill dam based on these theories is still faced with such major challenges as incomplete perception,lack of depth analysis,low level of intelligent decision-making and control.In order to solve the above problems,firstly,the dissertation review s the development process of roller compaction construction management of core-wall rockfill dam in detail,and deeply analyses the basic concepts and connotations of digital dam,construction information model,intelligent dam,intelligent monitoring and intelligent dam construction,and integrates many intelligent technology concepts such as "Internet of Things+" and "Artificial Intelligence+" to closely adhere to the compaction construction of core-wall rock-fill dam.The basic concept of intelligent monitoring for the surface compaction of core-wall rock-fill dam is put forward,and the basic characteristics and main research contents of intelligent compaction for core-wall rockfill dam are expounded.Secondly,the technical methods of intelligent perception,analysis and feedback control for core-wall rock-fill dam compaction construction are sorted out.Finally,the mathematical model of intelligent compaction for core-wall rockfill dam is established,and the aim of intelligent compaction for core-wall rock-fill dam is established.The research framework of intelligent compaction for core-wall rockfill dam is put forward based on labeling,information,method and constraint sets.(2)To counter the incomplete information perception of surface compaction of core-wall rockfill dam compaction construction,the intelligent perception method of the compaction construction information is proposed,which realizes the intelligent perception and integration of compaction construction information.At present,the surface compaction construction of core-wall rock-fill dam has not yet established a perfect information perception system,which cannot fully and thoroughly perceive the compaction construction information,such as the lack of processing the vibration signal noise of and the perception of the vibration characteristic parameters of the roller,the lack of recognition of multiple types of obstacles,the lack of perception and reconstruction of topographic and geomorphological information,and so on.At the same time,in the aspect of information integration of roller compaction construction,there are some problems such as lack of integration of multi-source heterogeneous compaction construction information and unbalanced data processing.In order to solve the above problems,the intelligent perception method of surface compaction construction information of core-wall rock-fill dam is proposed.Firstly,the content and of compaction construction information is analyzed,and the intelligent perception framework of compaction construction information is established.Secondly,the vibration signal of the steel wheel is sensed using the acceleration sensor,and the vibration characteristics of the steel wheel are extracted by a fast Fourier transform on the basis o f wavelet noise reduction.Then,the image information of the multi-type target of the compaction unit surface is obtained by the industrial camera,and the dilated convolution-based Faster-RCNN(a region with the convolutional neural network)is proposed to realize the recognition of different targets.Furthermore,the 3D modeling technology based on UAV tilt photography is used to realize the dam topography and landform.The perception and reconstruction of information is used to build the three-dimensional scene of intelligent compaction monitoring system.In addition,a multi-source heterogeneous compaction construction information integration framework is proposed,which integrates compaction parameters,material source characteristics,roller vibration characteristics and pit test data,and provides data basis for construction information analysis.Finally,a K-Meansbased downsampling method is proposed to process unbalanced data to ensure the reliability of data analysis.(3)To counter the insufficient depth of information analysis of the surface compaction construction information of the core-wall rock-fill dam,the intelligent analysis method of the surface compaction quality of the core-wall rock-fill dam is proposed,realizing the intelligent analysis of the compaction construction information.At present,the surface compaction quality analysis of core-wall rock-fill dam has not comprehensively considered the influence of compaction parameters,material source parameters and steel wheel vibration characteristics parameters on the quality of compaction construction;The accuracy,robustness and generalization ability of the compaction quality evaluation models established by multiple linear regression,back propagation neural network and support vec tor regression need to be improved;And the existing evaluation models fail to evaluate the quality of the roller compaction construction in real time.Furthermore,the research on the updating of compaction quality evaluation model of the lack in current studies.In order to solve the above problems,an intelligent analysis method for surface compaction construction information of core-wall rockfill dam is proposed.Firstly,based on the intelligent perception data of the compaction construction information,the compaction parameters,material source parameters,and vibration characteristic parameters are comprehensively considered,and the comprehensive evaluation model of compaction construction quality is established;Secondly,coupling the artificial intelligence,data mining and various intelligent components,the fuzzy logic algorithm based on combined kernel and adapti ve bacterial foraging algorithm,and the customized kernel support vector regression based on intelligent bacteria foraging algorithm are proposed to establish a high-precision,high-generalization and high-robust compaction quality evaluation model.Finally,based on the idea of concept drift detection in stream data,a concept drift detection based on enhanced probabilistic neural network and variable window technology is proposed.The algorithm realizes the detection of the concept drift phenomenon in the compaction construction stream data;and as the condition of the model update,the intelligent update of the compaction quality model is realized,which solves the problem of when the current compaction construction model is updated and how to update it.(4)To counter the problem of low level of intelligence in decision-making and control of the surface compaction of core-wall rockfill dam,the intelligent control method for the surface compaction construction of the core-wall rockfill dam is proposed,which enhances the intelligent control of ex-ante control,in-process control,and post-event control.Compaction construction control of core-wall rock-fill dam includes ex-ante control,in-process control,and post-event control.At present,ex-ante control mainly determines the compaction parameters through a compaction test,but this method does not comprehensively consider the construction quality,construction duration and construction cost of the compaction unit;The vehicle flat panel system monitors the compaction construction parameters in real time to realize feedback control.However,this method does not plan the construction path of the compaction unit and does not realize the intelligent evaluation of compaction quality and different target recognition.The post-event control is mainly realized by the test of the pit test,but this method has large dispersion and poor timelin ess.In order to solve the above problems,an intelligent control method for surface compaction construction process of core-wall rock-fill dam is proposed.Firstly,the intelligent control framework for the core-rockfill dam compaction construction was established.The contents of the control before,in and after the event and the corresponding control methods were discussed.Secondly,an optimization model with rolling construction parameters as decision variables,progress and cost as multi-objective and rolling quality as main constraints is established.The multi-objective intelligent bacterial foraging algorithm is used to solve the model.Finally,an intelligent control method for roller compaction based on AR(Augmented Reality)real-life guidance is proposed to realize the roller compaction construction path planning,the intelligent evaluation of compaction construction,the multi-target identification information are fused with the real-life construction scene,which is used for intelligent surface compaction control of the core-wall rock-fill dam during construction and after the event.(5)Based on Intelligent perception,intelligent analysis and intelligent control of core-wall rock-fill dam,an intelligent monitoring system for compaction construction information of core-wall rock-fill dam is developed.At present,the digital compaction of the core-wall rock-fill dam adopts the real-time monitoring system as an information management platform.The system realizes all-weather and fine-grained real-time monitoring of compaction construction parameters,but the system realizes information visualization by two-dimensional interface,and the display is poorly intuitive;at the same time,it is based on C/S framework(client/server),so it mainly runs in the Windows system and difficult to use across platforms.The client carries a large number of logical processing functions,so there are security risks such as data leakage.In order to solve the above problems,combined with the research results of intelligent perception,intelligent analysis and intelligent control of rolling construction information,and based on Browser/Server,object-oriented technology and Unity3 D development engine,an intelligent monitorin g system for core-wall rockfill dam rolling construction information is developed on the basis of three-dimensional model established by UAV tilt photography.The integration of rolling construction information and three-dimensional visualization,real-time intelligent analysis of rolling construction information and feedback control of rolling construction process are realized.The problems of difficult cross-platform use and system security are solved.