Research On Optimization Modeling Of Aircraft Fuel Consumption In Terminal Descent

Currently, the global aviations are trying their best to save oil according to the rising fuelprice and environmental pressures, avoiding unnecessary oil burn is one of the important means,which requires accurate estimations of fuel consumption when developing fuel plan. Thefuel consumption model is very important to estimate fuel consumption during the flight and accuratemodel can minimize the waste oil. However, the theoretical system of fuel consumptionestimations has not been established in Chinese aviation and the high precision fuel estimationtechniques are controlled by United States and European countries.Compared with cruise stage, the environmental change of terminal descent is more complex, aswell as the factors which affect the aircraft fuel consumption. Analyzing the fuel consumption of theterminal descent and creating an accurate model of fuel consumption play important parts for accurateestimations of fuel consumption.During the analysis and comparison of a large number of QAR data, the paper finds that the fuelconsumption values are repeatable for the same aircraft type in the same environment of theaircraft’s terminal descent. So the sub-match fuel consumption model has been proposed for aircraft’sterminal descent. The model has classified the flight environment of the terminal descent based onthe range of each factor which affects fuel consumption and the fuel consumption estimation modelsare established depending on historical fuel consumption data in different environments. Theaccuracy of fuel consumption estimation depends on the environmental coverage of the historicaldata in terminal descent, once the coverage is inadequate, the continuity of the model is not ideal. Sothe paper applies the neural network modeling approach to make the model achieveany flight environment of the terminal descent. In fact, different factors have different impacts onfuel consumption and determining influence level of each factor is benefit for model optimization.So, the paper has calculated the correlations between each factor and fuel consumption and theparameters have been optimized by neural network model. Through simplifying the numberof model input parameters, the fuel consumption based on neural network optimization model hasbeen established without reducing the fuel prediction accuracy greatly.The experiment reduces the factors from ten factors （pressure altitude, airspeed,vertical acceleration, air temperature, wind speed, wind direction, title angle, vertical acceleration,pitch angle and weight of aircraft）which affect the fuel consumption in terminal descent of the aircraft to four ones（airspeed, vertical acceleration, pitch angle and weight of aircraft），and experimentresults have demonstrated that the prediction error rate of the optimized BP networks just raises from4.02%to4.38%, but the optimized model is more simple, practical and the amount of computationhas been greatly reduced.