Study On The Estimation Model And Prediction Method Of Aircraft Fuel Consumption

For airlines,fuel costs are the biggest expense,accounting for about 40% of total costs.If the fuel quantity on the route can be predicted more accurately,the contingency fuel can be accurately managed.Therefore,accurate estimation the fuel quantity is an important way to save fuel.First,collect the fuel data and find out the relevant parameters that affect the amount of fuel.Including total fuel volume of flight plan,fuel consumption of flight plan route,flight plan flight time,planned takeoff weight,flight plan altitude,flight plan route wind,and flight plan temperature.Analyze the influence of these parameters on the amount of fuel in the route,and extract the model parameters at different quantiles regressions.Then establish a route fuel model and calculate contingency fuel standards.The experimental results show that the fuel quantity prediction model established by quantile regression can accurately predict the fuel quantity of airlines.This makes it possible to carry contingency fuel more flexibly while ensuring safety.Then build a BP neural network.After the flight plan parameters are input to the BP neural network,the fuel quantity is predicted to determine the contingency fuel.The experimental results show that by using the BP neural network to predict the fuel quantity of the route,and comparing with the actual fuel quantity of the route,model error is within 5%.Contingency fuel is determined on the basis of predicting the fuel quantity of the route through the BP neural network.Finally,compare the two models.BP neural network model is relatively flexible in input parameter selection.Through further optimization,the input parameters of the model can be extended,thereby improving the quality of the model.However,the disadvantage of the BP neural network model is that the initial weight and threshold of the BP neural network are random each time it is used.This also caused the model quality to be unstable.Quantile regression performs well in stability.But the quantile regression model belongs to the traditional statistical model and is not an artificial intelligence.