Research On Online Monitoring And Prediction Analysis Of Dam Safety Based On Internet Of Things And Extreme Learning Machine

In recent years,due to the increasing size of the dam body of the reservoir and the increasingly complex structure of the dam body,how to ensure safety during the operation of the dam has become one of the topics of increasing concern.Effective monitoring methods and reliable data analysis methods,as the two most effective means of dam safety protection,play a huge role together.The thesis starts from the two aspects of dam monitoring and dam deformation prediction,and combines information technology to study dam safety protection methods in the new era.For dam monitoring,the traditional method relies on manual methods to complete some simple monitoring,which not only consumes a lot of manpower and material resources,but also has the problems of low monitoring efficiency and poor monitoring effect.In the past ten years,the Internet of Things technology and wireless communication technology have become more and more mature.Using the Internet of Things to achieve dam safety monitoring has become a new method.In terms of dam deformation prediction,because the dam deformation is affected by many factors,and the deformation process is random and highly nonlinear,the traditional prediction model is difficult to adapt.With the continuous improvement of computer hardware,based on the machine under limited samples The learning method has achieved better and better results in deformation prediction.Therefore,the paper based on the Internet of Things and extreme learning machine to achieve online monitoring and prediction analysis of dam safety,the main research content is as follows.1)Research on dam safety data collection methods.With the help of Internet of Things technology,dam data collection is realized,and the whole collection scheme is divided into three layers.The perception layer uses small and modular sensors as the data collection unit to complete the data acquisition;the transmission layer uses the DTU device as the on-site data transmission center to achieve remote data interaction;the application layer is also called the user layer and is responsible for visual interaction.2)Optimize the extreme learning machine to achieve high-precision dam deformation prediction.Take the extreme learning machine(ELM)as the initial model and gradually optimize its performance.For the instability problem caused by the extreme given machine's random given hidden layer parameters,kernel function mapping is used to replace the random map;for the single-core extreme learning machine(KELM),there are strong local kernel learning capabilities but weak generalization performance,and global kernel panning The lack of strong performance but weak generalization performance,construct a PG_KELM model composed of a global kernel of polynomial kernels and a local kernel of Gaussian RBF kernels;in order to solve the optimization problem of the PG_KELM model,the bat algorithm is introduced to seek parameters,and the BA_PG_KELM model is proposed;Due to the long training time of the bat algorithm and easy to fall into the local optimal solution,the Levi flight optimization improvement that can enhance the local search and also have the ability to jump out of the local optimal is adopted,and the final LBA_PG_KELM model is proposed.3)Design and implement an online monitoring and analysis system for dam safety.In order to improve the flexibility of the system,the system is divided into two subsystems: data acquisition service and remote monitoring platform.The data collection service uses Socket socket for communication.First,a universal remote communication library is designed based on Socket,and then the secondary development is implemented on this basis to realize the dam data collection service.The remote monitoring platform is released as a web site,and the visual user interaction function is realized with the help of the Asp.net platform.4)Taking Lishan Reservoir as an engineering example,the actual process of dam deformation monitoring is explained,and the feasibility of dam safety online monitoring and analysis system is proved.At the same time,the collected data is used to verify the superiority of LBA_PG_KELM model.It can be known from the engineering application that the data collection scheme designed in this paper has practical feasibility and meets a series of requirements for data collection and data transmission in the dam monitoring process.The monitoring system developed based on this is highly scalable and can not only be easily expanded The new acquisition equipment can also apply the system to different fields without modifying the logic code.It can be seen from the verification of the measured data that the LBA algorithm based on Levi's flight optimization improves the BA algorithm's long training time and is easy to fall into the local optimal defect.Compared with the single-core KELM,the KELM based on the mixed core realizes the overtaking of the LBA_SVM algorithm,which explains the paper The proposed LBA_PG_KELM model can effectively improve the accuracy of dam deformation prediction.