Airport Ground Support Data Quality Check And Anomaly Detection Research

Massive ground support data has been produced in the operation process of airport.Due to the imperfect data quality management,there are many data quality problems such as data value missing and incorrect.These data quality problems affect the prediction accuracy of A-CDM,and are not helpful to the mining of data,so it's necessary to build data quality check system to improve data quality.Besides,it is important for airport to identify abnormal data which deviates too much from the other data,so anomaly detection technology is used to distinguish the normal and abnormal data.Then new ground support mode and rule can be discovered through the mining of data.And the new rules can be used to better monitor the ground support process.According to above problems,airport ground support data quality check and anomaly detection research are carried out.Firstly,data quality check indicators and the expression of each indicator are designed;then data quality check system is built by using IBM ODM and Websphere,and the system can realize the real-time check and feedback of data quality problem.Secondly,anomaly detection method is improved to build the normal and abnormal data subdivision model which can distinguish normal and abnormal data.After that,the indicator is selected and the k-fold cross validation is used to evaluate the algorithm.The validity of the algorithm is verified by comparing with other two algorithms.Then the anomaly data is analyzed.Finally,the normal data identified is analyzed.One case is analysis of the univariate distribution pattern.Then an association rules based mining model is built to mine the ground support rules.The model can consider the relationship between different attributes.And the support,confidence and improvement of each rule are calculated.The generated rules can be added into the airport ground support rule set,which can be used to better monitor the ground support process.