Neural Network Prediction And Numerical Simulation Analysis Of Subway Thermal Environment

As a transportation system widely used to solve the problem of urban congestion,the subway plays an important role in the current urban development.Safety and comfort are the basic requirements for the operation of the subway system.In line with the concept of people-oriented,efficient energy-saving and refined management,the increasingly prominent thermal environment problems inside the subway must be solved.The effective way to solve this problem is to regulate the internal thermal environment of the subway scientifically.Therefore,it is necessary to continuously monitor and analyze the thermal environment state of the subway,and to predict and grasp the changing rules of the thermal environment state in combination with the simulation of the whole-field thermal environment of the subway station,so as to provide a reliable basis for the internal environment regulation and system operation optimization of the subway.This thesis takes a typical transfer station in the Beijing subway as the research object,and the thermal environment of the station is studied by means of on-site testing,continuous monitoring,time series data mining and numerical simulation analysis.The main research contents include:(1)the dynamic change characteristics of the subway thermal environment state parameters;(2)the time-series wavelet analysis and neural network prediction of the subway thermal environment state parameters;(3)the numerical simulation analysis of the spatial and temporal changing rules of the subway thermal environment;(4)Preliminary study on the analysis method of the subway thermal environment based on time-series neural network prediction and full-field numerical simulation.Furthermore,a neural network prediction program for subway environmental state parameters is compiled.The research results show that: for the subway station with non-shielded door system,under the influence of many factors such as the interval of travel and train movement,the thermal environment state parameters of the platform have obvious fluctuation and non-uniformity in spatial distribution,and the data continuously monitored on-site presents strong nonlinear noise;wavelet systems such as db N,ciof N and sym N are used as wavelet basis function and the three-layer decomposition layers are used to analyze the monitoring data with obvious fluctuation,its noise reduction effect is better.In particular,the curve fluctuation trend after db2 and sym2 wavelet denoising is most consistent with the monitoring results,which can effectively solve the problem of data mutation and feature loss and ensure smooth data transition.The monitoring data after noise reduction and reconstruction can be used for neural network prediction;based on the neural network prediction model of the Bi LSTM unit has additional training capabilities,and the prediction effect and accuracy of thermal environment state parameters are better than the conventional model based on the LSTM unit.The prediction accuracy of the Bi LSTM combined model based on wavelet analysis can reach 0.9;Numerical simulation of the thermal environment in the whole process of train running at a constant speed,decelerating into the station,stopping and accelerating away from the station can be numerically simulated to obtain the dynamic distribution of the parameters of the entire field in a short time,and the boundary conditions of the future numerical simulation of the entire field and verification data can be provided by the results predicted by the time-series neural network.The combined analysis method based on deep learning time series prediction and CFD numerical simulation proposed in this thesis will facilitate a more comprehensive and accurate analysis of the future changes of subway thermal environment,and has certain innovation and application value.The research work of this thesis is supported by the project "Research on the Overall Thermal Environment System Control Scheme of Beijing Metro Lines 1 and 2." The research results have strong practical significance for optimizing subway thermal environment regulation,improving passenger comfort,saving energy and reducing emissions,and improving subway operation management level and service quality.