Aircraft fuel consumption calculation method based on meteorological data driving

Abstract

The invention relates to an aircraft fuel consumption calculation method based on meteorological data driving. The aircraft fuel consumption calculation method comprises the following steps: 1, collecting basic data influencing aircraft fuel consumption; 2, performing data preprocessing to form a data set; 3, dividing the data set into a training set and a test set; 4, constructing a fuel consumption evaluation model; 5, calculating a residual error between a predicted value and an actual value in the training set, and calculating a loss function; in the iteration process, a new regression tree is generated according to the residual error to minimize the current residual error, and iteration is stopped until the change of the loss function is smaller than a set threshold value; 6, performing K-fold cross validation on the model; and 7, inputting flight data into the model, and calculating and outputting fuel consumption. According to the method, aircraft fuel consumption calculation is carried out through the gradient boosting regression tree, the accuracy of a model calculation result is improved by introducing aircraft parameter data, navigation data and meteorological data, and the calculation result is more scientific and reasonable through model optimization, K fold cross validation and parameter tuning.

Description

Technical Field

[0001] The present invention relates to the field of aviation data technology, and in particular to a method for calculating aircraft fuel consumption driven by meteorological data. Background Art

[0002] In the aviation industry, fuel consumption has become an important part of airline operating costs. Accurately calculating aircraft fuel consumption is of great significance for improving fuel efficiency and reducing operating costs. Aircraft fuel consumption is affected by many factors, including engine model, aircraft weight, flight altitude, etc. The interaction between these factors makes fuel consumption calculation complicated and difficult.

[0003] Traditional fuel consumption calculation methods use empirical formulas or statistical models, such as the BADA model, which is an aircraft performance modeling tool developed by Eurocontrol. The BADA model builds an aircraft performance database based on BADA, which can realize functions such as flight simulation, trajectory prediction, emission assessment, and fuel consumption calculation. However, it covers a limited number of aircraft models and is restricted in use.

[0004] The prior art also discloses a method for estimating fuel emissions of civil aviation aircraft, such as a patent publication document with publication number CN116126888A, which uses QAR data to establish a relationship function between the fuel consumption of each aircraft model in the CCD stage and the flight time, and calculates the fuel consumption in the CCD stage through the relationship function. However, due to changes in environmental factors during the flight, it is impossible to accurately calculate the fuel consumption of the aircraft. Summary of the invention

[0005] In order to reduce the difficulty of calculating aircraft fuel consumption, the present invention provides an aircraft fuel consumption calculation method driven by meteorological data, which introduces the influence of meteorological data, improves the accuracy of calculation, optimizes operating costs, improves the fuel utilization efficiency of aircraft, and reduces the negative impact on the environment.

[0006] The technical objectives of the present invention are achieved through the following technical solutions:

[0007] A method for calculating aircraft fuel consumption based on meteorological data, the method comprising:

[0008] Step 1: Collect basic data that affect aircraft fuel consumption, including aircraft parameter data, navigation data and meteorological data. Aircraft parameter data includes aircraft engine model and aircraft weight. Meteorological data includes temperature, humidity, wind speed and wind direction. Navigation data includes operating altitude, operating speed and actual fuel consumption.

[0009] Step 2: preprocess the basic data, eliminate noise and abnormal data in the basic data, and standardize the basic data to form a data set;

[0010] Step 3: Divide the preprocessed data set into a training set and a test set, wherein the training set is used to construct a fuel consumption evaluation model, and the test set is used to evaluate the test of the fuel consumption evaluation model;

[0011] Step 4: Construct a GBDT model for fuel consumption evaluation;

[0012]

[0013] Where: F(x) is the predicted value of fuel consumption; F0(x) is the initial value of the model; β is the weight coefficient; h m (x) is the regression tree generated by the mth round of iterative training, where m is the number of iterations. The goal of each round of iteration is to train the regression tree h(x) so that the predicted value of this regression tree is as close to r as possible. i (m);

[0014] Step 5: Take the average fuel consumption of the training set data as the initial value of the model, calculate the residual between the predicted value of the current model and the actual value in the training set, and calculate the current loss function; in each round of iteration, generate a new regression tree based on the residual to minimize the current residual, and stop the iteration when the change of the loss function is less than the set threshold, and obtain the optimized fuel consumption evaluation model;

[0015] Step 6: Perform K-fold cross validation on the optimized fuel consumption evaluation model based on the data of the test set. If the validation fails, adjust the hyperparameters and execute step 5 until the K-fold cross validation passes.

[0016] Step 7: Input the aircraft engine model, aircraft weight, temperature, humidity, wind speed, wind direction, operating altitude, and operating speed into the fuel consumption evaluation model through K-fold cross validation, and output the fuel consumption through model calculation.

[0017] Furthermore, when performing data standardization processing in step 2, it includes:

[0018] S21, performing time matching and space matching of navigation data and meteorological data;

[0019] S22, integrating and fusing the aircraft parameter data, navigation data and meteorological data to form a two-dimensional matrix including time, elements, latitude and longitude, altitude, value and unit;

[0020] S23. Standardize the data according to its mean and standard deviation and convert it into normally distributed data with a mean of 0 and a variance of 1.

[0021] Furthermore, in step S21, when performing time matching, the time matching of the navigation data and the meteorological data is performed based on the standard that the time of the navigation data and the meteorological data is closest and the time difference does not exceed 30 minutes, and the occurrence event of the matched meteorological data is converted into the time of the navigation data;

[0022] When performing spatial matching, the distance between the navigation data and the meteorological data is the closest, and the horizontal distance is ≤0.04° and the height distance is ≤500m. The matched meteorological data is converted into the latitude, longitude and altitude of the navigation data according to the latitude, longitude and altitude.

[0023] Furthermore, in step 5, the following steps are included:

[0024] S51, model initialization, using the average fuel consumption of the training set data as the model initial value:

[0025]

[0026] Where: y i is the fuel consumption value of each group of data in the training set;

[0027] S52, calculate the residual r between the predicted value of the current model in the mth iteration and the actual value in the training set i (m), train the new regression tree h through the residual m (x);

[0028] r i (m) = y i -F m-1 (x i )

[0029] S53, adding the prediction result of the new regression tree to the existing model according to the weight to update the model, and obtaining the updated model as follows:

[0030] F m (x) = F m-1 (x)+βh m (x)

[0031] S54. Use mean square error to construct loss function:

[0032]

[0033] Where: y i is the actual fuel consumption, To predict fuel consumption;

[0034] When the loss function changes less than the set threshold, the iteration is stopped to obtain the optimized fuel consumption evaluation model. If the loss function changes greater than or equal to the set threshold, step S52 is executed.

[0035] Furthermore, in step 6, when performing K-fold cross validation, it includes:

[0036] S61, dividing the entire data set into K subsets of equal size;

[0037] S62, using one of the subsets as a validation set and the remaining K-1 subsets as training sets, and iteratively training and validating the optimized fuel consumption evaluation model;

[0038] S63. Record the mean square error of the model on each validation set;

[0039] S64. Calculate the average of the mean square errors of K validations to obtain the final performance verification result of the model.

[0040] Furthermore, in step 6, hyperparameter adjustment includes weight adjustment, adjustment of the number of regression trees, and adjustment of the depth of regression trees; the smaller the weight, the slower the convergence speed of the model; the more the number of regression trees, the higher the model complexity; the deeper the depth of the regression tree, the stronger the fitting ability of the model.

[0041] Furthermore, in step 3, the data volume ratio of the training set and the test set is 7:3.

[0042] Furthermore, the weight coefficient β is set between 0.01 and 0.1.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention calculates aircraft fuel consumption through a gradient boosting regression tree, improves the accuracy of model calculation results by introducing aircraft parameter data, navigation data and meteorological data, and makes the calculation results more scientific and reasonable through model optimization, K-fold cross validation and parameter tuning. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic flow chart of the aircraft fuel consumption calculation method driven by meteorological data of the present invention. DETAILED DESCRIPTION

[0046] The technical solution of the present invention is further described below in conjunction with specific embodiments:

[0047] A method for calculating aircraft fuel consumption based on meteorological data, the method comprising:

[0048] Step 1: Collect basic data that affect aircraft fuel consumption, including aircraft parameter data, navigation data and meteorological data. Aircraft parameter data includes aircraft engine model and aircraft weight. Meteorological data includes temperature, humidity, wind speed and wind direction. Navigation data includes operating altitude, operating speed and actual fuel consumption.

[0049] Among them, meteorological data can be collected and downloaded from the official website of meteorological data and the official aviation website, and the operating altitude, operating speed and actual fuel consumption can be collected and downloaded from the official aviation website.

[0050] Step 2: preprocess the basic data, eliminate noise and abnormal data in the basic data, and standardize the basic data to form a data set;

[0051] The collected raw data are cleaned by writing Python scripts to eliminate noise and abnormal data in the basic data, delete outliers, missing values, duplicate values ​​and erroneous values ​​in the data, and fill in missing values ​​using Newton interpolation.

[0052] When the data is standardized in step 2, it includes:

[0053] S21. Since the navigation data and the meteorological data are changing, time matching and space matching of the navigation data and the meteorological data are performed;

[0054] When performing time matching, the time matching of the navigation data and the meteorological data is carried out based on the standard that the time of the navigation data and the meteorological data is closest and the time difference does not exceed 30 minutes, and the occurrence events of the matched meteorological data are converted into the navigation data time;

[0055] When performing spatial matching, the distance between the navigation data and the meteorological data is the closest, and the horizontal distance is ≤0.04° and the height distance is ≤500m. The matched meteorological data is converted into the latitude, longitude and altitude of the navigation data according to the latitude, longitude and altitude.

[0056] S22, integrating and fusing the aircraft parameter data, navigation data and meteorological data to form a two-dimensional matrix including time, elements, latitude and longitude, altitude, value and unit, so as to facilitate analysis;

[0057] S23. Scale the data so that the data falls into a relatively small specific interval. In this embodiment, the z-score standardization method is used to standardize the data according to the mean and standard deviation of the data, and convert it into normally distributed data with a mean of 0 and a variance of 1.

[0058] It is expressed as follows:

[0059] In the formula, μ is the mean of the data, σ is the standard deviation of the data, x is the data before standardization, and x' is the data after standardization.

[0060] Step 3: Divide the preprocessed data set into a training set and a test set. As a preferred method, 70% of the preprocessed data is used as the training set, and the remaining 30% is used as the test set. The training set is used to build the fuel consumption evaluation model, and the test set is used to evaluate the fuel consumption evaluation model, specifically the performance evaluation and generalization ability evaluation of the model. When dividing the data set, it is necessary to ensure that the data between the training set and the test set are independent and random to avoid information leakage between the training set and the test set, thereby ensuring the objectivity and reliability of the model.

[0061] Step 4: Use the gradient boosting regression tree method (GBRT) to build a GBDT model for fuel consumption evaluation;

[0062]

[0063] Where: F(x) is the predicted value of fuel consumption; F0(x) is the initial value of the model; β is the weight coefficient, the value range is 0.01-0.1, and the value of β in this embodiment is 0.1; h m (x) is the regression tree generated by the mth round of iterative training, where m is the number of iterations;

[0064] Step 5: Take the average fuel consumption of the training set data as the initial value of the model, calculate the residual between the predicted value of the current model and the actual value in the training set, and calculate the current loss function; in each round of iteration, generate a new regression tree based on the residual to minimize the current residual, and stop the iteration when the change of the loss function is less than the set threshold, and obtain the optimized fuel consumption evaluation model;

[0065] In step 5, the following steps are specifically included:

[0066] S51, model initialization, using the average fuel consumption of the training set data as the model initial value:

[0067]

[0068] Where: y i is the fuel consumption value of each set of data in the training set, and x represents the corresponding flight data;

[0069] S52, calculate the residual r between the predicted value of the current model in the mth iteration and the actual value in the training set i (m), train the new regression tree h through the residual m (x), the goal of each iteration is to train the regression tree h(x) so that the predicted value of this regression tree is as close to r as possible. i (m);

[0070] r i (m) = y i -F m-1 (xi )

[0071] S53, adding the prediction result of the new regression tree to the existing model according to the weight to update the model, and obtaining the updated model as follows:

[0072] F m (x) = F m-1 (x)+βh m (x)

[0073] S54. Use mean square error to construct loss function:

[0074]

[0075] Where: y i is the actual fuel consumption, To predict fuel consumption;

[0076] When the loss function changes less than the set threshold, the iteration is stopped to obtain the optimized fuel consumption evaluation model. If the loss function changes greater than or equal to the set threshold, step S52 is executed.

[0077] Step 6: Perform K-fold cross validation on the optimized fuel consumption evaluation model based on the data of the test set. If the validation fails, adjust the hyperparameters and execute step 5 until the K-fold cross validation passes.

[0078] K-fold cross validation can be used to evaluate the stability and generalization ability of the model. It mainly verifies the effect of the model by splitting the data multiple times. When performing K-fold cross validation, it includes:

[0079] S61, dividing the entire data set into K subsets of equal size;

[0080] S62, using one of the subsets as a validation set and the remaining K-1 subsets as training sets, and iteratively training and validating the optimized fuel consumption evaluation model;

[0081] S63. Record the mean square error of the model on each validation set. The mean square error can be used to measure the difference between the predicted result and the actual result, so as to evaluate the prediction ability of the model.

[0082] S64. Calculate the average of the mean square errors of K validations to obtain the final performance verification result of the model.

[0083] If the final performance verification result of the model does not meet expectations, the hyperparameters can be further optimized. Hyperparameter adjustments include weight adjustment, number of regression trees adjustment, and depth of regression trees adjustment. The smaller the weight, the slower the convergence speed of the model. The more regression trees there are, the higher the model complexity. The deeper the depth of the regression tree, the stronger the fitting ability of the model.

[0084] Step 7: Input flight data into the fuel consumption evaluation model through K-fold cross validation, the flight data including aircraft engine model, aircraft weight, temperature, humidity, wind speed, wind direction, operating altitude, and operating speed, and output fuel consumption through model calculation.

[0085] In order to make the process of the present invention clearer, the calculation process of the present invention is described as follows:

[0086] The training set data is shown in the following table:

[0087] x x x ... x y y y ... y

[0088] When executing step S51, F0(x) is calculated in sequence, as shown in the following table:

[0089] F0(x1)=y1;

[0090] F0(x2)=(y1+y2) / 2;

[0091] F0(x3)=(y1+y2+y3) / 3; ...

[0093] F0(x N )=(y1+y2+y3+...+y N ) / N;

[0094] Execute step S52 to calculate the residual r between the predicted value and the actual value in the training set in the first iteration i (m),

[0095] r1(1)=y1-F0(x1);

[0096] r2(1)=y2-F0(x2);

[0097] r3(1)=y3-F0(x3); ...

[0099] r N (1) = y N -F0(x N );

[0100] Update the model and obtain: F1(x)=F0(x)+βh1(x);

[0101] Execute step S53, calculate the current loss function, if the loss function change is greater than or equal to the set threshold, execute step S52;

[0102] Assuming that the current loss function change is greater than or equal to the set threshold, when executing step S52, a second iteration is performed:

[0103] r1(2)=y1-F1(x1);

[0104] r2(2)=y2-F1(x2);

[0105] r3(2)=y3-F1(x3); ...

[0107] r N (2) = y N -F1(x N );

[0108] Update the model and obtain: F2(x)=F1(x)+βh2(x);

[0109] Execute step S53 and calculate the current loss function. If the change in the loss function is less than the set threshold, the optimized model is F2(x)=F1(x)+βh2(x).

[0110] This embodiment is only a further explanation of the present invention, not a limitation of the present invention. After reading this specification, those skilled in the art may make non-creative modifications to this embodiment as needed, but as long as it is within the scope of the claims of the present invention, it will be protected by the patent law.

Claims

1. A method for calculating aircraft fuel consumption driven by meteorological data, characterized in that: The method includes: Step 1: Collect basic data that affect aircraft fuel consumption, including aircraft parameter data, navigation data and meteorological data. Aircraft parameter data includes aircraft engine model and aircraft weight. Meteorological data includes temperature, humidity, wind speed and wind direction. Navigation data includes operating altitude, operating speed and actual fuel consumption. Step 2: preprocess the basic data, eliminate noise and abnormal data in the basic data, and standardize the basic data to form a data set; Step 3: Divide the preprocessed data set into a training set and a test set, wherein the training set is used to construct a fuel consumption evaluation model, and the test set is used to evaluate the test of the fuel consumption evaluation model; Step 4: Construct a GBDT model for fuel consumption evaluation; Where: F(x) is the predicted value of fuel consumption; F0(x) is the initial value of the model; β is the weight coefficient; h m (x) is the regression tree generated by the mth round of iterative training, where m is the number of iterations; Step 5: Take the average fuel consumption of the training set data as the initial value of the model, calculate the residual between the predicted value of the current model and the actual value in the training set, and calculate the current loss function; in each round of iteration, generate a new regression tree based on the residual to minimize the current residual, and stop the iteration when the change of the loss function is less than the set threshold, and obtain the optimized fuel consumption evaluation model; Step 6: Perform K-fold cross validation on the optimized fuel consumption evaluation model based on the data of the test set. If the validation fails, adjust the hyperparameters and execute step 5 until the K-fold cross validation passes. Step 7: Input the aircraft engine model, aircraft weight, temperature, humidity, wind speed, wind direction, operating altitude, and operating speed into the fuel consumption evaluation model through K-fold cross validation, and output the fuel consumption through model calculation.

2. The method for calculating aircraft fuel consumption based on meteorological data according to claim 1, characterized in that: When the data is standardized in step 2, it includes: S21, performing time matching and space matching of navigation data and meteorological data; S22, integrating and fusing the aircraft parameter data, navigation data and meteorological data to form a two-dimensional matrix including time, elements, latitude and longitude, altitude, value and unit; S23. Standardize the data according to its mean and standard deviation and convert it into normally distributed data with a mean of 0 and a variance of 1.

3. The method for calculating aircraft fuel consumption based on meteorological data according to claim 2, characterized in that: In step S21, when time matching is performed, the time matching of the navigation data and the meteorological data is performed based on the standard that the time of the navigation data and the meteorological data is closest and the time difference does not exceed 30 minutes, and the occurrence event of the matched meteorological data is converted into the navigation data time; When performing spatial matching, the distance between the navigation data and the meteorological data is the closest, and the horizontal distance is ≤0.04° and the height distance is ≤500m. The matched meteorological data is converted into the latitude, longitude and altitude of the navigation data according to the latitude, longitude and altitude.

4. The method for calculating aircraft fuel consumption based on meteorological data according to claim 1, characterized in that: In step 5, the following steps are included: S51, model initialization, using the average fuel consumption of the training set data as the model initial value: Where: y i is the fuel consumption value of each group of data in the training set data; S52, calculate the residual r between the predicted value of the current model in the mth iteration and the actual value in the training set i (m), train the new regression tree h through the residual m (x); r i (m)=y i -F m-1 (x i ) S53, adding the prediction result of the new regression tree to the existing model according to the weight to update the model, and obtaining the updated model as follows: F m (x)=F m-1 (x)+βh m (x) S54. Use mean square error to construct loss function: Where: y i is the actual fuel consumption, To predict fuel consumption; When the loss function changes less than the set threshold, the iteration is stopped to obtain the optimized fuel consumption evaluation model. If the loss function changes greater than or equal to the set threshold, step S52 is executed.

5. The method for calculating aircraft fuel consumption based on meteorological data drive according to claim 1, characterized in that: In step 6, when performing K-fold cross validation, it includes: S61, dividing the entire data set into K subsets of equal size; S62, using one of the subsets as a validation set and the remaining K-1 subsets as training sets, and iteratively training and validating the optimized fuel consumption evaluation model; S63. Record the mean square error of the model on each validation set; S64. Calculate the average of the mean square errors of K validations to obtain the final performance verification result of the model.

6. The method for calculating aircraft fuel consumption based on meteorological data drive according to claim 1, characterized in that: In step 6, the hyperparameter adjustment includes weight adjustment, adjustment of the number of regression trees, and adjustment of the depth of the regression trees; the smaller the weight, the slower the convergence speed of the model; the more the number of regression trees, the higher the model complexity; the deeper the depth of the regression tree, the stronger the fitting ability of the model.

7. The method for calculating aircraft fuel consumption based on meteorological data drive according to claim 1, characterized in that: In step 3, the data volume ratio of the training set and the test set is 7:

3.

8. The method for calculating aircraft fuel consumption based on meteorological data drive according to claim 1, characterized in that: The weight coefficient β is between 0.01 and 0.1.

Images