A Study On Safety Performance Evaluation Of Large Airport Apron Control System

With the issuance of the Civil Aviation Administration of China’s "Notice on Promoting the Transfer of Aerodrome Operation Management to Airport Management Organizations"(CAAC Letter [2013] No.75)and the "Notice on Issuing the Overall Plan for the Transfer of Aerodrome Control Management"(CAAC Letter [2018] No.38),the transfer of aerodrome operations management at large airports has been constantly promoted,leading to increasingly concrete safety issues.As an important element of the Civil Aviation Safety Management System(SMS),safety performance is crucial for improving aerodrome safety management and enhancing its quality.In this article,a corresponding safety performance evaluation index system and model for aerodrome control safety performance evaluation at large airports have been established,and the model has been validated and applied through example.Firstly,the current research status of safety performance both domestically and internationally has been systematically reviewed.The characteristics of aerodrome control operations have been described in detail,and relevant theoretical methods for aerodrome safety performance have been analyzed.The advantages and disadvantages of BP neural networks and other safety performance evaluation methods have been compared,as well as the advantages and disadvantages of the DPSIR model and other models in establishing safety performance indicators.Using the DPSIR model and combining previous research on safety performance indicators,a corresponding index system for aerodrome control safety performance evaluation has been established,with five sub-indicators under driving force,pressure,state,and response,and six sub-indicators under impact.These 26 indicators interact with each other and serve as the basis for determining aerodrome control safety performance levels.Secondly,the traditional BP neural network has been analyzed,and it has been found that it cannot extract specific features for each indicator,which is crucial in actual safety performance evaluation,where some indicators have a significant impact on safety performance levels.Therefore,this article proposes an optimized BP neural network model for safety performance evaluation,named the MIBP model,which is sensitive to characteristic indicators.Finally,the constructed MIBP model has been used to evaluate the safety performance of aerodrome control at a large C airport.The model has been trained using a large amount of data,and its accuracy and sensitivity have been tested using test sample data.The loss curve shows that the model has reached the best results after about 10 epochs of training.By analyzing the evaluation results using a confusion matrix,the MIBP model achieved a evaluation accuracy of98% in the training set and 100% in the validation set.When applied to the example,the model accurately estimated the safety performance level.This model provides a quick and convenient tool for ensuring the safety of aerodrome control,and has good application value in improving the safety level of aerodrome control.