Heterogeneous Data Mixed Modeling Methods And Its Application For Tunneling Machine

Tunneling machine is a kind of large construction machinery which is specially used for the full section excavation of tunnel engineering.It is an intelligent tunnel construction equipment which integrates machinery,electricity,hydraulics,information and control,which plays a very important role in urban underground engineering construction.Its intelligent operation and maintenance system is driven by big data technology,which uses artificial intelligence technology for data mining,analysis and processing on the data collected at the work site,and then to ensure the tunneling machine efficient,stable,safe and scientific operation.In this way it predicts and gets the required information which provides an important reference for the operation and maintenance decision of tunneling machine.The data modeling algorithms,which are the cores of the intelligent operation and maintenance,is of great importance in predicting the driving conditions.However,the data obtained in the process of operation are multi-source and heterogeneous,the existing artificial intelligence algorithms lack of research on mixed modeling of the heterogeneous data,which makes negative impacts on the operation system of tunnel excavation projects,indirectly affects the progress of the project and causes engineering accidents.To solve the above situation,this issue designs methods such as improving machine learning algorithms and designing structured coding to realize heterogeneous data mixed modeling,which can be used to predict the working conditions.Based on the tunnel construction data of Shenzhen Metro,the prediction models of tunneling condition parameters is established to verify the performance of the proposed algorithms,which provides algorithm support for the intelligent operation of the tunnel machine.The main research results of this thesis can be listed as follows:1)Investigate the research status of data modeling and working condition prediction of tunneling machine.Discuss the significance of the research.Design the technology route and methods of heterogeneous data mixed modeling.2)Data preprocessing for tunneling machine working data.Including: data merging,outlier test,and feature engineering.It makes a basis for subsequent data modeling and prediction.3)Based on support vector regression,the hypersphere parameterization is introduced to improve the kernel function to mixed model the numerical data and classified data.Gray wolf optimizer is used to optimize the parameters.Several groups of numerical experiments were designed to verify the performance of the proposed algorithm.Based on the actual engineering data,the prediction model for working condition is established.The proposed method performs better than other existing algorithms.4)The geological data collected from drilling sampling were interpolated in order to analyze the stratigraphic distribution of the tunneling section in each tunneling ring section.Structured coding method is designed to make geological data suitable for mixed data modeling.Based on the operating data and geological data,the prediction models for different ring sections in tunneling are established.The proposed method performs better than other existing algorithms.5)The method of curve fitting in functional data analysis is introduced,and an improved algorithm based on the numerical data and classified data mixed modeling is proposed,which realized the prediction of curves.The engineering value of the proposed algorithm is verified by an engineering example where the cutter head speed response curve predicting experiment under sudden load.6)Based on MATLAB GUI,the heterogeneous data mixed modeling system of tunneling machine is designed.