Research On Vehicle Ground Wireless Communication Fault Early Warning Model Basedon Machine Learning

Aiming at the communication abnormal and failure of DCS in the subway wireless communication system CBTC system and the problems that the existing communication system can not display hidden faults,the paper analyzes the communication status data of AP log,establishes the vehicle ground wireless communication fault warning model based on machine learning,and designs the visual display page.The early warning model is as follows:Firstly,the AP log data of Shanghai Metro Line 11 was collected,stored and read from January to February 2019,which is divided into training data and test data.The data processing database of Python language is used to preprocess AP log data,which can be divided into data cleaning,data transformation,data integration and merging.The key words of text column data are extracted,and the extracted communication status description words and AP device location columns are coded to facilitate subsequent modeling.Then,the clustering k-means algorithm and ARIMA algorithm of machine learning are used to analyze the time and space dimensions respectively.On the one hand,the correlation between time dimension and AP failure is analyzed by ARIMA model,that is,time and space have certain influence on the occurrence of AP equipment failure,On the other hand,kmeans algorithm is used to extract the time period of frequent failure and the points of subway station and AP equipment to form characteristic serial data.Finally,the log prediction model is established by using LSTM NN algorithm.The sequence data analyzed in the early stage is taken as the input value,and the information feature is automatically extracted under the control of update gate and forgetting gate.The historical information state before the current time is analyzed,the status information at the moment is predicted,and the final output prediction results are obtained after learning the complete sequence.The whole early warning model aims to dig the deep reasons behind the fault,such as the influence relationship between the time period of frequent fault occurrence and the flow of people in the site,or the relationship between the AP equipment with frequent failure and the system environment(under the hole,elevated).That is,the whole early warning system is used to analyze all internal and external causes that affect the frequency of failure and predict in advance,which provides some guidance for the detection of subway personnel.