Civil aviation aircraft fuel consumption and pollutant emission estimation method

By extracting navigation time and route attribute information from QAR data, and combining with the XGBoost model, the fuel consumption of civil aviation aircraft is accurately estimated, which solves the problem of large calculation errors in the existing technology, and achieves more accurate fuel consumption and pollutant emissions Estimation of .

CN119089192APending Publication Date: 2024-12-06CHINA ACAD OF CIVIL AVIATION SCI & TECH

Patent Information

Application Number
CN202310665599.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-06-06
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

When estimating the fuel consumption of civil aviation aircraft, the prior art failed to fully consider the impact of different flights and routes on the duration of the flight stage, resulting in large calculation errors, and there are differences in fuel flow and emission coefficients in different flight stages, resulting in carbon dioxide. Emission estimates are not accurate enough.

Method used

By extracting navigation time information and route attribute information from QAR data, a training neural network model is constructed, and the XGBoost model is used to combine flight data and QAR data to accurately estimate the fuel consumption in each flight stage, and calculate it based on the fuel consumption multiplied by the emission coefficient. Pollutant emissions.

Benefits of technology

A more accurate estimate of fuel consumption is achieved, the calculation error is reduced, and the fuel consumption can be estimated more accurately for different routes. The estimation accuracy is improved through nonlinear models, making up for the insufficient time spent in various stages in the prior art. flaws.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119089192A_ABST
    Figure CN119089192A_ABST
Patent Text Reader

Abstract

The invention belongs to the field of civil aviation information estimation, and provides a civil aviation aircraft fuel consumption estimation method and a pollutant emission estimation method, and the method comprises the steps: extracting navigation time information, route attribute information and total fuel consumption from QAR data; constructing a sample set training neural network model, wherein each sample in the training sample set comprises an input vector and a truth value label; the input vector comprises the navigation time information and the route attribute information; the true value label is the total fuel consumption; the method comprises the following steps: acquiring flight data and corresponding QAR data of a to-be-estimated aviation aircraft, extracting flight time information from the flight data, extracting flight route attribute information from the corresponding QAR data, and forming a to-be-estimated feature vector; and inputting the to-be-estimated feature vector into the trained neural network model, outputting an estimated value of the total fuel consumption, and estimating the pollutant emission based on the total fuel consumption. According to the invention, the fuel consumption can be estimated more accurately for different routes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of civil aviation information estimation, and in particular relates to a method for estimating fuel consumption of a civil aviation aircraft, and a method for estimating pollutant emissions based on the fuel consumption. Background Art

[0002] Since the beginning of the 21st century, the increasingly severe climate environment and extreme weather phenomena have threatened people's property and personal safety, and the greenhouse effect and other impacts caused by excessive carbon dioxide emissions will directly or indirectly affect the environment. To achieve emission reduction, it is necessary to accurately estimate and monitor the total amount of carbon emissions in various fields. As the civil aviation transportation industry with the highest emission intensity among all modes of transportation, a more accurate and rapid method for estimating total emissions is needed.

[0003] The carbon dioxide produced by flight activities is mainly generated by fuel combustion, and carbon dioxide is its most important product. Therefore, it is generally believed that carbon dioxide emissions are linearly correlated with the amount of fuel consumed. Most current studies indirectly estimate carbon dioxide emissions by estimating fuel consumption. That is, the carbon dioxide emissions of a single flight are mainly calculated through the following steps:

[0004] 1) A single flight is divided into push-out, take-off, vertical climb, climb, cruise, descent, approach, landing, and taxi-in phases.

[0005] 2) According to the aircraft model, the total fuel consumed in each stage is obtained by directly multiplying the consumption time of each stage with the fuel flow rate of each stage obtained from the ICAO database or calculated based on previous data.

[0006] 3) Add up the fuel consumption in each stage and multiply it by the emission coefficient of aviation fuel and carbon dioxide to obtain the total amount of carbon dioxide emitted by a single flight.

[0007] The fuel flow and emission coefficient of an aircraft during flight vary to varying degrees in different flight phases. Currently, most studies use the average time of each LTO phase recommended by ICAO, combine this time with the corresponding fuel flow in the ICAO database to obtain fuel consumption, and calculate the corresponding emissions based on this. This rough calculation does not take into account the impact of different flights / routes on the duration of the flight phase, which will introduce large errors in the bottom-up calculation. Summary of the invention

[0008] The purpose of the present invention is to solve the above-mentioned problems existing in the prior art and to provide a method for estimating fuel consumption of civil aviation aircraft.

[0009] The present invention is implemented by the following technical scheme: A method for estimating fuel consumption of a civil aviation aircraft comprises the following steps:

[0010] Extracting flight time information, route attribute information and total fuel consumption from the QAR data, wherein the flight time information includes the duration of each stage in the entire flight process, and the route attribute information includes the flight altitude information and the total weight of the aircraft before takeoff;

[0011] Constructing a sample set training neural network model, wherein each sample in the training sample set includes an input vector and a true value label; the input vector includes the flight time information and the route attribute information; the true value label is the total fuel consumption;

[0012] Obtain flight data and corresponding QAR data of the aircraft to be estimated, extract flight time information from the flight data, extract route attribute information from the corresponding QAR data, and form a feature vector to be estimated;

[0013] The feature vector to be estimated is input into the trained neural network model, and an estimated value of the total fuel consumption is output.

[0014] Furthermore, the entire navigation process includes a taxi-out phase, a flight phase and a taxi-in phase; the flight phase includes a take-off phase, a climb phase, a cruise phase, a descent phase, an approach phase and a landing phase.

[0015] Furthermore, the steps of extracting the flight time information from the flight data are as follows: according to the four time points recorded in the flight data set: departure time, take-off time, landing time and arrival time, the duration of the taxi-out phase, the flight phase and the taxi-in phase are calculated; and the duration of the take-off phase, the climb phase, the cruise phase, the descent phase, the approach phase and the landing phase are calculated according to the fixed duration ratio of each phase in the flight phase multiplied by the flight phase duration.

[0016] Furthermore, the flight altitude information includes take-off altitude, landing altitude and maximum cruising altitude.

[0017] Furthermore, the total fuel consumption is extracted from the QAR data in any of the following ways:

[0018] 1) The total amount of fuel remaining at departure minus the total amount of fuel remaining at destination;

[0019] 2) Accumulate the fuel flow from the departure time to the arrival time of the destination.

[0020] Furthermore, the neural network model adopts the XGBoost model.

[0021] Furthermore, a grid-search strategy is used during training to optimize the hyperparameters of the neural network model.

[0022] Furthermore, a corresponding neural network model is trained for each aircraft type to estimate the total fuel consumption of a single voyage.

[0023] The present invention also provides a method for estimating pollutant emissions of civil aviation aircraft. The method for estimating fuel consumption of civil aviation aircraft described in the present invention is used to obtain an estimated value of total fuel consumption, and the estimated value of total fuel consumption is multiplied by a corresponding pollutant emission coefficient to obtain pollutant emissions.

[0024] Furthermore, the estimated value of total fuel consumption is multiplied by the carbon dioxide emission factor to obtain carbon dioxide emissions.

[0025] Compared with the prior art, the beneficial effects of the present invention include:

[0026] 1. From the training data, the present invention applies the specific information extracted from the QAR data, and can obtain the time consumed in each stage more accurately than other methods. In addition, the present invention extracts the route attribute information from the QAR data, taking into account the impact of different flights / routes on the duration of the flight phase. The present invention can estimate the fuel consumption more accurately for different routes.

[0027] 2. From the perspective of fuel estimation, the present invention utilizes smaller flight data. Relevant information is extracted from the flight information, such as the time consumed in each stage. And the QAR data is used to estimate the stages not reflected in the flight information data, that is, the duration of the take-off stage, climb stage, cruise stage, landing stage, approach stage and landing stage is calculated based on the fixed duration ratio of each stage in the flight stage multiplied by the flight stage duration, which makes up for the defect that the time consumed in each stage of the flight data is not detailed enough.

[0028] 3. From the estimation model point of view, the present invention uses the XGBoost model to estimate fuel consumption. The model can establish a nonlinear model of fuel consumption of many independent variables to achieve more accurate estimation. The model provides parallel tree boosting and is used to solve many data science problems. In addition, the present method uses grid-search to optimize the hyperparameters of XGBoost to achieve the best effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Flow chart of this specific implementation method. DETAILED DESCRIPTION

[0030] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0031] A method for estimating fuel consumption of a civil aviation aircraft comprises the following steps:

[0032] Extracting flight time information, route attribute information and total fuel consumption from the QAR data, wherein the flight time information includes the duration of each stage in the entire flight process, and the route attribute information includes the flight altitude information and the total weight of the aircraft before takeoff;

[0033] Constructing a sample set training neural network model, wherein each sample in the training sample set includes an input vector and a true value label; the input vector includes the flight time information and the route attribute information; the true value label is the total fuel consumption;

[0034] Obtain flight data and corresponding QAR data of the aircraft to be estimated, extract flight time information from the flight data, extract route attribute information from the corresponding QAR data, and form a feature vector to be estimated;

[0035] The feature vector to be estimated is input into the trained neural network model, and an estimated value of the total fuel consumption is output.

[0036] The neural network model used in this specific implementation is the XGBoost model, which can be replaced by other neural network models. However, considering the following advantages of the XGBoost model, the XGBoost model is preferred:

[0037] Efficient: XGBoost uses parallel processing to quickly train large-scale data sets;

[0038] Stability: XGBoost can steadily improve the accuracy of the model;

[0039] Strong applicability: XGBoost can handle both numerical and categorical features, and can handle missing values.

[0040] The present invention utilizes the QAR data and flight data of the Chinese civil aviation. The following specifically describes the technical solution of the present invention by calculating the corresponding information based on the QAR data and flight data.

[0041] The following QAR data are provided by the China Civil Aviation Flight Quality Monitoring Base Station for all flights operated at airports in China (excluding Hong Kong, Macao and Taiwan) from July 14, 2021 to July 21, 2021. The flight data are flight information data from January 2018 to November 2021. The XGBoost model is implemented with the help of the "xgboost" third-party function library developed in Python.

[0042] refer to Figure 1 As shown, the process of this specific implementation includes the following steps:

[0043] Step 1: Preprocess the QAR data so that it can be used for training the estimation model, and extract the flight time information, route attribute information and total fuel consumption.

[0044] This specific implementation utilizes the QAR data field shown in Table 1.

[0045] Table 1 QAR data attribute information related to fuel consumption estimation

[0046]

[0047] First, the actual time of each stage is extracted according to the attribute value in FLIGHT_PHASE, that is, the time consumed in the taxi-out stage, take-off stage, climb stage, cruise stage, descent stage, approach stage, landing stage, and taxi-in stage respectively.

[0048] Secondly, extract the attribute information related to altitude from the ALT-STDC field, such as take-off altitude, landing altitude and maximum cruising altitude. Then extract the total weight of the aircraft before take-off from the GW field. The route attribute information includes the navigation altitude information and the total weight of the aircraft before take-off.

[0049] Finally, the estimated amount is extracted based on the FF1, FF2 and FQTY fields, that is, the total fuel consumption of the aircraft during the entire process. The total fuel consumption can be extracted from the QAR data in any of the following ways:

[0050] 1) The total amount of fuel remaining at departure minus the total amount of fuel remaining at destination;

[0051] 2) Accumulate the fuel flow from the departure time to the arrival time of the destination, the fuel flow includes FF1 and FF2.

[0052] Step 2: Train a corresponding neural network model for each aircraft type to estimate the total fuel consumption of a single voyage.

[0053] Constructing a sample set training neural network model, wherein each sample in the training sample set includes an input vector and a true value label; the input vector includes the flight time information and the route attribute information; the true value label is the total fuel consumption;

[0054] The model input is the time taken for each flight phase, take-off altitude, landing altitude, maximum cruising altitude and the total weight of the aircraft before take-off obtained in step 1, and the model output is the total fuel consumption.

[0055] During training, the model was divided into a training set and a test set at a ratio of 7:3, and the grid-search strategy was used to optimize the model's hyperparameters. The R value (recall rate) obtained by the training XGBoost model on some representative models and the corresponding number of samples are shown in Table 2.

[0056] Table 2 R2 values ​​and sample sizes of some models using XGBoost model

[0057]

[0058] Step 3: Extract flight information and construct the feature vector to be estimated.

[0059] Obtain flight data and corresponding QAR data of the aircraft to be estimated, extract flight time information from the flight data, extract route attribute information from the corresponding QAR data, and form a feature vector to be estimated;

[0060] Three time quantities are obtained from the four time points recorded in the flight information, namely the departure time, take-off time, landing time and arrival time, namely the taxi-out phase time (from the departure time to the take-off time), the flight phase time (from the take-off time to the landing time) and the taxi-in phase time (from the landing time to the arrival time).

[0061] The steps to extract flight time information from flight data are as follows: according to the four time points recorded in the flight data set: departure time, take-off time, landing time and arrival time, calculate the duration of the taxi-out phase, flight phase and taxi-in phase; and calculate the duration of the take-off phase, climb phase, cruise phase, descent phase, approach phase and landing phase by multiplying the fixed duration proportion of each phase in the flight phase by the flight phase duration.

[0062] The fixed duration ratio of each stage in the flight phase is obtained as follows: the average value of the ratio of the time of each stage in the flight phase of each sample in the training sample set to the total duration of the flight phase is calculated.

[0063] Step 4: Estimation of total fuel consumption.

[0064] The feature vector to be estimated is input into the trained neural network model, and an estimated value of the total fuel consumption is output.

[0065] The flight time information and route attribute information extracted in step 3 are used as feature vectors to be estimated, and are input into the XGBoost model trained in step 2 to estimate the total fuel consumption.

[0066] Multiply the estimated total fuel consumption by the corresponding pollutant emission factor to obtain pollutant emissions, such as nitrogen oxides.

[0067] In this specific implementation, the total fuel consumption is multiplied by the carbon dioxide emission coefficient of aviation fuel to obtain the final carbon dioxide emissions. The carbon emission coefficient used here is 3.15, that is, for every unit weight of aviation fuel consumed, 3.15 units weight of carbon dioxide will be emitted.

[0068] The above technical solution is only one implementation mode of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the principles disclosed in the present invention, and it is not limited to the technical solution described in the above specific embodiments of the present invention. Therefore, the above description is only preferred and does not have a restrictive meaning.

Claims

1. A method for estimating fuel consumption of a civil aviation aircraft, characterized in that: The following steps are involved: Extracting flight time information, route attribute information and total fuel consumption from the QAR data, wherein the flight time information includes the duration of each stage in the entire flight process, and the route attribute information includes the flight altitude information and the total weight of the aircraft before takeoff; Constructing a sample set training neural network model, wherein each sample in the training sample set includes an input vector and a true value label; the input vector includes the flight time information and the route attribute information; the true value label is the total fuel consumption; Obtain flight data and corresponding QAR data of the aircraft to be estimated, extract flight time information from the flight data, extract route attribute information from the corresponding QAR data, and form a feature vector to be estimated; The feature vector to be estimated is input into the trained neural network model, and an estimated value of the total fuel consumption is output.

2. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: The whole navigation process includes the taxi-out phase, the flight phase and the taxi-in phase; the flight phase includes the take-off phase, the climb phase, the cruise phase, the descent phase, the approach phase and the landing phase.

3. The method for estimating fuel consumption of a civil aviation aircraft according to claim 2, characterized in that: The steps to extract flight time information from flight data are as follows: according to the four time points recorded in the flight data set: departure time, take-off time, landing time and arrival time, calculate the duration of the taxi-out phase, flight phase and taxi-in phase; and calculate the duration of the take-off phase, climb phase, cruise phase, descent phase, approach phase and landing phase by multiplying the fixed duration proportion of each phase in the flight phase by the flight phase duration.

4. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: The navigation altitude information includes take-off altitude, landing altitude and maximum cruising altitude.

5. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: Extract the total fuel consumption from the QAR data in any of the following ways: 1) The total amount of fuel remaining at departure minus the total amount of fuel remaining at destination; 2) Accumulate the fuel flow from the departure time to the arrival time of the destination.

6. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: The neural network model adopts the XGBoost model.

7. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: During training, a grid-search strategy is used to optimize the hyperparameters of the neural network model.

8. The method for estimating fuel consumption of a civil aviation aircraft according to claim 1, characterized in that: A corresponding neural network model is trained for each aircraft type to estimate the total fuel consumption of a single voyage.

9. A method for estimating pollutant emissions from civil aviation aircraft, characterized by: The method for estimating fuel consumption of a civil aviation aircraft as claimed in any one of claims 1 to 8 is used to obtain an estimated value of total fuel consumption, and the estimated value of total fuel consumption is multiplied by a corresponding pollutant emission coefficient to obtain pollutant emissions.

10. The method for estimating pollutant emissions from civil aviation aircraft according to claim 9, characterized in that: Multiply the estimate of total fuel consumption by the CO2 emission factor to obtain CO2 emissions.

Citation Information

Patent Citations

  • Estimation method for fuel emissions of civil aviation aircraft

    CN116126888A

Cited By

  • Method and device for estimating fuel consumption of aircraft in vertex drop

    CN120337490A

  • Method and device for estimating fuel consumption during peak descent of an aircraft

    CN120337490B