Fuel Flow Precision Estimation Model Of Civil Aviation Airplane Based On Deep Learning

The quantitative measurement of fuel consumption is the basis for evaluating the efficiency of air traffic operation and the effect of energy saving and emission reduction of air traffic control new technologies,and it is a more comprehensive and feasible way to estimate air traffic fuel consumption based on radar trajectory data.The fuel flow rate estimation was influenced by many factors,it was difficult to characterize the complex relationship between multivariate variables using current models.The deep learning model had unique advantages in mining the complex internal relationship between multivariable variables and fuel flow rate,and can improve estimation accuracy of the model more effectively.First,the identification of the key influencing factors of the fuel flow rate based on air traffic trajectory data.Combining QAR data with trajectory data,the performance parameters influencing fuel flow rate were analyzed and extracted from trajectory data.The influence level of each parameter on fuel flow rate was analyzed and the features were selected using mutual information method according to the forward selection process.Secondly,aiming at the lack of true airspeed in integrated trajectory data,the relationship between altitude,ground speed and true airspeed was explored,then the true airspeed estimation model was built using deep belief network.The results show that fitting degree of the model was 95.92%,which can be used to estimate the true airspeed in the whole flight trajectory.The fuel flow rate estimation model of civil aviation aircraft was further constructed using deep belief network.The results show that the model had a better fitting effect,and can estimate the fuel flow rate based on the integrated trajectory data.Thirdly,aiming at the unsatisfactory accuracy of the fuel flow rate estimation model,the influence of network parameters such as hidden layer number,hidden node number and learning rate on the model was analyzed,and then the optimal value of each parameter was selected to train the model to optimize the structure of the fuel flow rate model.The results show that the accuracy of the model can reach 93.17%,which was 3.6% higher than before optimization model.Then,it was found that accuracy and stability of the deep belief networkfuel flow rate model were obviously better than other models through experimental comparison.Finally,the fuel flow rate model was applied to the calculation of air traffic fuel consumption,and the reliability of the model was verified.Comparing with trajectory data of actual flight operation,it was found that the fuel flow rate model can accurately estimate the fuel flow rate of civil aircraft,and thus achieve the calculation of air traffic fuel consumption,which laid the theoretical foundation of fuel consumption measurement for large-scale air traffic trajectory data.