Design And Implementation Of Flight Delay Analysis System For Beijing Tianjin Hebei Airport Group

In view of the unbalanced development and the high flight delay rate of airport cluster,a flight delay analysis system for Beijing-Tianjin-Hebei Airport Group is designed in this paper.The system makes full use of the flight information and meteorological data of the airport group,establishes a prediction model based on deep learning to evaluate the severity of flight delay,and provides the function of historical flight data statistics and visualization,which provides a reference for the operation decision-making and collaborative development of the Beijing-Tianjin-Hebei airport group,and reducing the overall delay rate of airport group.First of all,a detailed demand analysis for the flight delay analysis system of Beijing-Tianjin-Hebei airport group is proposed,the system is designed according to the demand,and the key technology implementation is mainly introduced.The system is developed for the air traffic control department,and installed in the air traffic control center flow management room,tower and traffic management center of North China Air Traffic Administration of Civil Aviation of China.The system is adopted using the Browser/Server architecture.The system can be divided into four parts,which are flight data retrieval,delay analysis,delay prediction and data management.According to the system functions,the storage structure and optimization method based on My SQL,delay statistical analysis based on Echarts,delay prediction algorithm call based on Socket are introduced in detail,The Big Data Development Methods to improve the fluency of flight delay analysis system,including Elastic Search Large Data retrieval engine and Redis cache system,are emphasized.Then,a data retrieval algorithm based on ES and Redis is proposed.Considering that the replacement principle of Redis native replacement algorithm only considers the replacement of the least recently used data,which has certain limitations.On this foundation,the Redis native replacement algorithm is improved and a new cache replacement algorithm(Hybrid)is developed,which measures the cache value of data by the size of the cache data,the most recently used time,the frequency of use and other factors.The data retrieval speed of ES-Redis is about 10 times higher than that of traditional database search algorithms.Next,an airport group delay prediction algorithm based on GRU network is presented.Considering that the flight take-off and landing in the airport is a continuous process with time sequence,long short term memory(LSTM)is used to extract flight features.The LSTM unit consists of a forgetting gate,an input gate and an output gate,which solves the gradient disappearance and explosion phenomenon of the recurrent neural network to a certain extent.The GRU network only keeps the update and reset gates.GRU not only inherits the excellent characteristics of LSTM network,but also gets faster training speed by simplifying the model.After data preprocessing,encoding and network training,the prediction accuracy is 87.42%.Finally,the flight delay analysis system is tested in detail by means of black box test,and the specific test cases and system test results are given.The test mainly includes the functional test of flight data retrieval,flight delay statistical analysis and flight delay prediction,as well as the system performance test of page response time,system robustness and delay prediction accuracy.