Airplane fuel consumption prediction method and device, electronic equipment, medium and product

Abstract

The present application relates to the technical field of aviation big data analysis, and particularly relates to an aircraft fuel consumption prediction method and device, electronic equipment, medium and product, the method comprising: obtaining aircraft flight access records and meteorological reanalysis data, obtaining standardized historical flight data, standardized reference trajectory data and future meteorological field data through data processing; processing the standardized reference trajectory data and the future meteorological field data to obtain future flight segment reconstructed trajectories and four-dimensional meteorological analysis results, respectively, combining the two to extract trajectory point environmental parameters; generating a physical enhancement feature set based on the standardized historical flight data and the trajectory point environmental parameters, obtaining an aircraft fuel consumption prediction result through a preset gradient boosting decision tree, solving the problems of poor physical interpretability, inaccurate complex stage prediction and weak future prediction capability in related technologies, improving the physical interpretability and prediction accuracy of aircraft fuel consumption prediction, and improving the accuracy and robustness of fuel consumption prediction.

Description

Technical Field

[0001] This invention relates to the field of aviation big data analysis technology, and in particular to a method, device, electronic equipment, medium and product for predicting aircraft fuel consumption. Background Technology

[0002] Aviation fuel consumption is one of the largest operating costs for airlines. High-precision, interpretable fuel consumption predictions for all phases of aircraft's history and future flight missions (climb, cruise, descent) are key to saving aviation fuel consumption.

[0003] The relevant technologies mainly rely on physical performance databases such as BADA (Base of Aircraft Data). The model is based on Newtonian mechanics and uses a set of fixed aerodynamic coefficients to represent the average performance of a certain aircraft type, thereby enabling the prediction of aviation fuel consumption.

[0004] However, deep learning models in related technologies are often "black boxes," making it difficult to explain why fuel consumption increases under specific operating conditions. During the aircraft's climb and descent phases, complex thrust changes and aerodynamic configuration alterations are involved, and ordinary regression models struggle to capture the nonlinear coupling relationship between altitude, speed, and fuel consumption. Furthermore, existing systems often focus on historical data analysis, lacking the ability to predict fuel consumption at any future time and for any flight segment under expected weather conditions (i.e., "de-historicized" prediction capability). They also often use only static or simple meteorological data, ignoring the dynamic changes of the atmospheric environment over time (4D weather field). As a result, data-driven methods (such as neural networks), while possessing strong fitting capabilities, are prone to overfitting when lacking physical constraints, and their prediction accuracy for non-steady-state processes such as climb / descent is insufficient, which urgently needs to be addressed. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, medium, and product for predicting aircraft fuel consumption, in order to solve the problems of poor physical interpretability, inaccurate prediction in complex stages, and weak future prediction capabilities in related technologies, thereby improving the physical interpretability and accuracy of aircraft fuel consumption prediction, and enhancing the precision and robustness of fuel consumption prediction.

[0006] To achieve the above objectives, a first aspect of the present invention provides an aircraft fuel consumption prediction method, comprising the following steps: acquiring aircraft flight access records and meteorological reanalysis data; performing feature extraction and spatiotemporal standardization on the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data, and constructing a four-dimensional environmental tensor based on the meteorological reanalysis data to obtain future meteorological field data; reconstructing the kinematic state of the standardized reference trajectory data to obtain a reconstructed trajectory for the future flight segment, and performing spatiotemporal interpolation analysis on the future meteorological field data to obtain a four-dimensional meteorological analysis result; obtaining dynamic environmental parameters of trajectory points based on the reconstructed trajectory for the future flight segment and the four-dimensional meteorological analysis result; obtaining a physically enhanced feature set containing physical prior constraints based on the standardized historical flight data and the dynamic environmental parameters of trajectory points, and inputting the physically enhanced feature set into a preset gradient boosting decision tree to obtain an aircraft fuel consumption prediction result.

[0007] Further, in some embodiments, the step of predicting the standardized reference trajectory data to obtain a future segment reconstructed trajectory includes: acquiring the time axis of the standardized reference trajectory data; mapping the time axis of the standardized reference trajectory data to the time interval of the future target segment to obtain an initial reconstructed trajectory with future time labels; calculating the ground speed of each trajectory point in the future target segment based on the future time labels and latitude and longitude coordinate data corresponding to the initial reconstructed trajectory; generating a future segment speed reconstructed trajectory based on the ground speed of each trajectory point in the future target segment; identifying target trajectory points in the future segment speed reconstructed trajectory where the ground speed of the trajectory points is less than a preset ground speed threshold and the standard pressure altitude of the trajectory points is less than a preset standard pressure threshold; correcting the fuel flow rate of the target trajectory points to a preset flow rate value; and using the adjusted future segment speed reconstructed trajectory as the future segment reconstructed trajectory.

[0008] Furthermore, in some embodiments, the analysis of the future meteorological field data to obtain four-dimensional meteorological analysis results includes: acquiring the barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of the future meteorological field data; aligning the barometric pressure layer data and standard pressure altitude data of the future meteorological field data in the vertical dimension using a preset ISA standard atmospheric conversion mechanism to obtain parameter-aligned barometric pressure layer data and standard pressure altitude data; obtaining a composite wind field vector based on the zonal and meridional wind components of the future meteorological field data according to a preset vector synthesis calculation method; and generating four-dimensional meteorological analysis results based on the parameter-aligned barometric pressure layer data, the parameter-aligned standard pressure altitude data, and the composite wind field vector using a pre-constructed four-dimensional interpolator.

[0009] Furthermore, in some embodiments, the physical enhancement feature set includes approximate Mach number, full energy state, gravitational work power, and aerodynamic drag factor, wherein, The approximate Mach number is:

[0010] The total energy state is:

[0011] The power of work done by gravity is:

[0012] The aerodynamic drag factor is:

[0013] in, For an approximate Mach number, In a state of full energy, Power of work done by gravity It is the aerodynamic drag factor. For ground speed, For the local speed of sound, For the outside temperature, It is gravitational potential energy. As kinetic energy, For aircraft quality, It is the acceleration due to gravity. Standard atmospheric pressure altitude This represents the aircraft's actual speed. For the real-time full weight of the aircraft, Vertical velocity, This is a preset numerical stability constant used to prevent numerical divergence caused by the denominator approaching zero when the aircraft is flying at low altitudes, ensuring the continuity and robustness of the feature across all altitude layers. This is the standard atmospheric pressure altitude.

[0014] Furthermore, in some embodiments, the aircraft flight access record includes at least one of timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow.

[0015] The aircraft fuel consumption prediction method provided by this invention acquires aircraft flight access records and meteorological reanalysis data, and obtains standardized historical flight data, standardized reference trajectory data, and future meteorological field data through data processing. The standardized reference trajectory data and future meteorological field data are processed separately to obtain a reconstructed trajectory for the future flight segment and a four-dimensional meteorological analysis result. Environmental parameters of the trajectory points are extracted by combining the two. A physical enhancement feature set is generated based on the standardized historical flight data and trajectory point environmental parameters, and this set is input into a preset gradient boosting decision tree to output the aircraft fuel consumption prediction result. This solves the problems of poor physical interpretability, inaccurate prediction in complex stages, and weak future prediction capabilities in related technologies, improving the physical interpretability and accuracy of aircraft fuel consumption prediction, and enhancing the precision and robustness of fuel consumption prediction.

[0016] To achieve the above objectives, a second aspect of the present invention provides an aircraft fuel consumption prediction device, comprising: an acquisition module for acquiring aircraft flight access records and meteorological reanalysis data; a data analysis module for processing the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data, and obtaining future meteorological field data based on the meteorological reanalysis data; a four-dimensional meteorological analysis module for predicting the standardized reference trajectory data to obtain a future segment reconstructed trajectory, and analyzing the future meteorological field data to obtain a four-dimensional meteorological analysis result, and obtaining trajectory point environmental parameters based on the future segment reconstructed trajectory and the four-dimensional meteorological analysis result; and a prediction module for obtaining a physical enhancement feature set based on the standardized historical flight data and the trajectory point environmental parameters, and inputting the physical enhancement feature set into a preset gradient boosting decision tree to obtain an aircraft fuel consumption prediction result.

[0017] Furthermore, in some embodiments, the four-dimensional meteorological analysis module is specifically used for: acquiring the time axis of the standardized reference trajectory data and mapping it to the time interval of the future target flight segment, thereby obtaining an initial reconstructed trajectory with future time labels; then, calculating the displacement between adjacent trajectory points using the geographical distance formula based on the latitude and longitude coordinates of the initial reconstructed trajectory, and calculating the ground speed sequence in combination with the sampling period; subsequently, performing outlier truncation on the ground speed sequence based on a physical rationality threshold, that is, considering the subsonic characteristics of civil aircraft under conventional flight profiles, setting the upper limit threshold of ground speed to 300 m / s (approximately 0.9 m / s). Mach (approaching the maximum cruising speed of civil aircraft), and a ground speed lower limit threshold is preset to 0 m / s. Based on this, a sliding window mean filtering algorithm is used for smoothing to construct a future segment speed reconstruction trajectory after eliminating positioning noise. Further, by traversing the trajectory points in the future segment speed reconstruction trajectory, target trajectory points with ground speeds less than the preset ground speed threshold and standard pressure altitudes less than the preset standard pressure threshold are identified. The predicted fuel flow rate of the target trajectory points is corrected to a preset flow rate value to suppress numerical anomalies in the model under zero speed boundary conditions. Finally, the trajectory after fuel flow correction is used as the final future segment reconstruction trajectory output.

[0018] Furthermore, in some embodiments, the four-dimensional meteorological analysis module is also used to: acquire the barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of the future meteorological field data; align the barometric pressure layer data and standard pressure altitude data of the future meteorological field data in the vertical dimension through a preset ISA standard atmospheric conversion mechanism to obtain parameter-aligned barometric pressure layer data and standard pressure altitude data; based on the zonal wind component and meridional wind component of the future meteorological field data, obtain a composite wind field vector according to a preset vector synthesis calculation method; and generate a four-dimensional meteorological analysis result based on the parameter-aligned barometric pressure layer data, the parameter-aligned standard pressure altitude data, and the composite wind field vector, according to a pre-constructed four-dimensional interpolator.

[0019] Furthermore, in some embodiments, the physical enhancement feature set includes approximate Mach number, full energy state, gravitational work power, and aerodynamic drag factor, wherein, The approximate Mach number is:

[0020] The total energy state is:

[0021] The power of work done by gravity is:

[0022] The aerodynamic drag factor is:

[0023] in, For an approximate Mach number, In a state of full energy, Power of work done by gravity It is the aerodynamic drag factor. For ground speed, For the local speed of sound, For the outside temperature, It is gravitational potential energy. As kinetic energy, For aircraft quality, It is the acceleration due to gravity. Standard atmospheric pressure altitude This represents the aircraft's actual speed. For the real-time full weight of the aircraft, Vertical velocity, This is a preset numerical stability constant used to prevent numerical divergence caused by the denominator approaching zero when the aircraft is flying at low altitudes, ensuring the continuity and robustness of the feature across all altitude layers. This is the standard atmospheric pressure altitude.

[0024] Furthermore, in some embodiments, the aircraft flight access record includes at least one of timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow.

[0025] The aircraft fuel consumption prediction device provided in this embodiment of the invention acquires aircraft flight access records and meteorological reanalysis data, and obtains standardized historical flight data, standardized reference trajectory data, and future meteorological field data through data processing; it processes the standardized reference trajectory data and future meteorological field data respectively to obtain the future segment reconstructed trajectory and four-dimensional meteorological analysis results, and extracts trajectory point environmental parameters by combining the two; it generates a physical enhancement feature set based on the standardized historical flight data and trajectory point environmental parameters, inputs it into a preset gradient boosting decision tree, and outputs the aircraft fuel consumption prediction result. Thus, it solves the problems of poor physical interpretability, inaccurate prediction in complex stages, and weak future prediction ability in related technologies, improves the physical interpretability and prediction accuracy of aircraft fuel consumption prediction, and enhances the accuracy and robustness of fuel consumption prediction.

[0026] To achieve the above objectives, a third aspect of the present invention provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aircraft fuel consumption prediction method as described in the above embodiments.

[0027] To achieve the above objectives, a fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the aircraft fuel consumption prediction method as described in the above embodiments.

[0028] A fifth aspect of the present invention provides a computer program product, including a computer program that is executed to implement the aircraft fuel consumption prediction method as described in the above embodiments.

[0029] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0030] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart of an aircraft fuel consumption prediction method provided according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of an aircraft fuel consumption prediction method according to a specific embodiment of the present invention; Figure 3 A schematic diagram of the loss convergence curve of a model training process according to a specific embodiment of the present invention; Figure 4 This is a schematic diagram comparing actual fuel consumption with predicted fuel consumption according to a specific embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the fuel consumption forecast for future flight segments in April, provided according to a specific embodiment of the present invention. Figure 6 This is a schematic diagram illustrating the prediction of future flight segments in March according to a specific embodiment of the present invention; Figure 7 A scatter plot illustrating the prediction effect of the climb phase according to a specific embodiment of the present invention; Figure 8 A scatter plot illustrating the predicted cruise phase effect according to a specific embodiment of the present invention; Figure 9 A scatter plot illustrating the prediction effect of the descent phase according to a specific embodiment of the present invention; Figure 10 A schematic diagram illustrating the overall prediction effect according to a specific embodiment of the present invention; Figure 11 This is a block diagram of an aircraft fuel consumption prediction device provided according to an embodiment of the present invention; Figure 12 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present invention. Detailed Implementation

[0031] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0032] The aircraft fuel consumption prediction method, apparatus, electronic equipment, medium, and product provided according to embodiments of the present invention will be described below with reference to the accompanying drawings. First, the aircraft fuel consumption prediction method provided according to embodiments of the present invention will be described with reference to the accompanying drawings.

[0033] Figure 1 This is a flowchart of an aircraft fuel consumption prediction method provided according to an embodiment of the present invention.

[0034] like Figure 1 As shown, the aircraft fuel consumption prediction method includes the following steps: In step S101, aircraft flight access records and meteorological reanalysis data are acquired.

[0035] Among them, aircraft flight access records refer to the set of flight status and system operating parameters collected and stored in real time by the airborne fast access recorder during aircraft flight. Meteorological reanalysis data refers to meteorological data obtained by systematically and griddedly restoring and reconstructing the atmospheric state over a period of time through numerical weather prediction models combined with meteorological observation data from around the world (ground stations, radiosondes, satellites, aircraft, etc.).

[0036] Specifically, the aircraft flight access record (QAR) in this embodiment of the invention comes from the accumulated fleet operation data of airlines, such as the historical flight records of a medium-sized passenger aircraft fleet of an airline. The meteorological reanalysis data adopts ERA5 high-resolution meteorological reanalysis data, which is obtained from the Climate Data Store (CDS) published by the European Centre for Medium-Range Weather Forecasts (ECMWF).

[0037] In some embodiments, the aircraft flight access record includes at least one of the following: timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow.

[0038] In step S102, feature extraction and spatiotemporal standardization processing are performed on the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data. A four-dimensional environmental tensor is constructed based on meteorological reanalysis data to obtain future meteorological field data.

[0039] Standardized historical flight data refers to parameters that include all dimensions related to aircraft flight status and fuel consumption; standardized reference trajectory data refers to parameters related to aircraft flight trajectory; and future meteorological field data refers to a four-dimensional (i.e., time, latitude, longitude, and altitude) meteorological dataset covering any future time and space range.

[0040] Specifically, the aircraft flight access records are first cleaned, including outlier removal and missing value completion, to remove invalid parameters caused by sensor malfunctions and signal interference during flight, and to reasonably complete a small number of missing key flight parameters. Then, the cleaned aircraft flight access records are standardized in terms of parameter dimensions, and all parameters are calibrated in terms of spatiotemporal dimensions and arranged in an orderly manner according to timestamps. Finally, the aircraft flight access records are classified, and standardized historical flight data is generated by extracting flight parameters from all dimensions, such as timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow. At the same time, standardized reference trajectory data is generated by extracting parameters related to the flight trajectory, such as time axis, latitude and longitude, standard barometric altitude, and ground speed.

[0041] Furthermore, the initial field data of ERA5 meteorological reanalysis is input into a preset large meteorological model (e.g., Pangu-Weather, ONNX format). Multi-step iterative inference is performed with inference step sizes of 1h / 3h / 6h / 24h to simulate the spatiotemporal dynamic changes of the atmospheric environment and generate meteorological forecast data for any future time and any spatiotemporal range. The generated future meteorological forecast data is saved in a standard format (e.g., NetCDF format) and stored in a structured manner according to time, pressure layer, latitude, and longitude to build a local meteorological data warehouse. Finally, future meteorological field data containing four-dimensional information of time, latitude, longitude, and altitude is formed.

[0042] In step S103, kinematic state reconstruction is performed on the standardized reference trajectory data to obtain the reconstructed trajectory for the future flight segment, and spatiotemporal interpolation analysis is performed on the future meteorological field data to obtain the four-dimensional meteorological analysis results. Based on the reconstructed trajectory for the future flight segment and the four-dimensional meteorological analysis results, the dynamic environmental parameters of the trajectory points are obtained.

[0043] Among them, the future flight segment reconstruction trajectory refers to the standardized flight trajectory adapted to the future target flight segment after the standardized reference trajectory data is dehistorically reconstructed in time and space and the physical rules are corrected. The four-dimensional meteorological analysis result refers to the meteorological interpolation analysis result of the four-dimensional input of time, pressure layer, latitude and longitude. The trajectory point environmental parameters refer to the dynamic meteorological environmental parameters that are extracted through four-dimensional meteorological time and space interpolation and match the time and space of the trajectory point.

[0044] In some embodiments, predicting the standardized reference trajectory data to obtain the future segment reconstruction trajectory includes: acquiring the time axis of the standardized reference trajectory data; mapping the time axis of the standardized reference trajectory data to the time interval of the future target segment to obtain the initial reconstruction trajectory with future time labels; calculating the ground speed of each trajectory point in the future target segment based on the future time labels and latitude and longitude coordinate data corresponding to the initial reconstruction trajectory; generating the future segment speed reconstruction trajectory based on the ground speed of each trajectory point in the future target segment; identifying target trajectory points in the future segment speed reconstruction trajectory where the ground speed of the trajectory points is less than a preset ground speed threshold and the standard pressure altitude of the trajectory points is less than a preset standard pressure threshold; correcting the fuel flow rate of the target trajectory points to a preset flow rate value; and using the adjusted future segment speed reconstruction trajectory as the future segment reconstruction trajectory.

[0045] Among them, the time interval of the future target flight segment refers to the specific time period during which a future flight mission is planned to be carried out; ground speed refers to the speed of the aircraft relative to the ground; the future flight segment speed reconstruction trajectory refers to the future flight segment speed curve after eliminating the interference of historical speed parameters; and the future flight segment speed curve refers to the trajectory points in the future flight segment speed reconstruction trajectory that simultaneously meet the conditions of ground speed being less than a preset ground speed threshold and standard air pressure altitude being less than a preset standard air pressure threshold.

[0046] As one possible implementation, this embodiment of the invention first extracts the original timeline inherent in the standardized reference trajectory data. Based on the planned start and end times of the future target flight segment and the segment timing plan, the original timeline is shifted and mapped to the time interval of the future target flight segment. This matches the corresponding future flight time for each trajectory point of the reference trajectory, such as latitude, longitude, and altitude, forming an initial reconstructed trajectory with future time labels. Then, the historical ground speed data in the standardized reference trajectory data is discarded, and the ground speed of each trajectory point in the future target flight segment is recalculated using the Haversine formula based only on the future time labels and the changes in the latitude and longitude coordinates of each trajectory point in the initial reconstructed trajectory. Subsequently, the recalculated ground speed data is processed by sliding window mean filtering. At the same time, the ground speed and standard air pressure values ​​are judged and filtered. Target trajectory points with ground speeds greater than or equal to a preset ground speed threshold or standard air pressure altitudes greater than or equal to a preset standard air pressure threshold are selected, and the fuel flow rate of the point is set to 0. Finally, a smoothed and physically-based future flight segment speed reconstruction trajectory is obtained.

[0047] For example, in this embodiment of the invention, the ground speed threshold is set to 1.0 knots, the standard pressure altitude threshold is set to 2000 feet, and the preset flow rate is set to 0 kg / h. By traversing all trajectory points in the future flight segment speed reconstruction trajectory, the system identifies the ground speed and standard pressure altitude parameters of each point, and then filters out target trajectory points with a ground speed of less than 1.0 knots and a standard pressure altitude of less than 2000 feet. These are determined to be aircraft in a stationary or low-speed taxiing state, and their predicted fuel flow rate is forcibly corrected to 0 kg / h. For trajectory points that do not meet the filtering conditions, their original ground speed, altitude, and time stamp parameters are retained without adjustment. Finally, the trajectory after fuel flow rate correction is output as the reconstructed trajectory of the target predicted flight segment.

[0048] Furthermore, in some embodiments, the future meteorological field data is analyzed to obtain four-dimensional meteorological analysis results, including: acquiring the barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of the future meteorological field data; aligning the barometric pressure layer data and standard pressure altitude data of the future meteorological field data in the vertical dimension through a preset ISA standard atmospheric conversion mechanism to obtain parameter-aligned barometric pressure layer data and standard pressure altitude data; obtaining a composite wind field vector based on the zonal and meridional wind components of the future meteorological field data according to a preset vector synthesis calculation method; and generating four-dimensional meteorological analysis results based on the parameter-aligned barometric pressure layer data, the parameter-aligned standard pressure altitude data, and the composite wind field vector using a pre-constructed four-dimensional interpolator.

[0049] Among them, barosphere data refers to the stratified atmospheric pressure data of different upper-level layers, zonal wind component refers to the east-west wind field component in the vector decomposition of the natural wind field, meridional wind component refers to the north-south wind field component in the vector decomposition of the natural wind field, and ISA standard atmospheric conversion mechanism refers to the rules for mutual conversion between meteorological barosphere data and aviation standard baroaltitude data based on the International Standard Atmosphere (ISA) model.

[0050] As one possible implementation, this embodiment of the invention extracts barosphere data, standard pressure altitude data, zonal wind components, and meridional wind components from future meteorological data generated by the Pangu-Weather meteorological model and stored in NetCDF format. Redundant meteorological elements unrelated to wind field and vertical height are removed. Based on the conversion formula of the international standard atmospheric model, the barosphere data and standard pressure altitude data in the future meteorological data are matched vertically to obtain parameter-aligned barosphere data and standard pressure altitude data. Vector operations are performed on the meridional wind component (U) and zonal wind component (V). Through vector synthesis, the total wind speed and direction are calculated to obtain a synthetic wind field vector that fully characterizes the physical state of the wind field. Finally, using time, barosphere, latitude, and longitude as four-dimensional input dimensions, the parameter-aligned barosphere data, standard pressure altitude data, and synthetic wind field vector are input into a four-dimensional interpolator. The interpolator learns the spatiotemporal distribution patterns of the future meteorological data, and through the modeling and calculation of the interpolator, the final four-dimensional meteorological analysis result is obtained.

[0051] For example, the four-dimensional meteorological interpolator in this embodiment of the invention is constructed based on the scipy.interpolate library and meteorological data. First, a data sequence is constructed, and the system traverses the local meteorological data warehouse to identify the target prediction time window. All NetCDF format meteorological files within the dataset are then concatenated along the "time dimension" using multidimensional array processing libraries such as xarray to construct a continuous four-dimensional tensor. (T is the time step, L is the pressure level, H is latitude, and W is longitude); secondly, coordinate space standardization is performed. Because the RegularGridInterpolator interpolation algorithm requires the grid coordinate axes to be strictly monotonically increasing, and the system has a built-in automatic detection and correction module, if latitude (e.g., 90°) is detected... → 90 If the dimension of the air pressure layer (e.g., 1000 hPa → 10 hPa) is monotonically decreasing, a tensor flip operation will be automatically performed to synchronously adjust the coordinate array and data matrix to meet the requirements of mathematical operations. Then, the interpolator will be instantiated, and independent four-dimensional interpolation objects will be constructed for key meteorological variables such as temperature t, zonal wind u, meridional wind v, and specific humidity q, defining the mapping function as follows. (Where t is the Unix timestamp and p is the standard barometric altitude,) (where λ is the geographical latitude and λ is the geographical longitude). The core employs a linear interpolation strategy, simultaneously interpolating in both time and space dimensions to ensure the spatiotemporal continuity and smoothness of the meteorological field. Finally, dynamic querying and inference are performed for any predicted point in the flight trajectory. First, determine the geometric height. Convert to standard barometric altitude pi, then construct a four-dimensional query vector. ,Will The environment state vector at that spatiotemporal point is calculated in real time after being input into the interpolator. This process ensures that the environmental parameters used in the fuel prediction model are the actual weather conditions that the aircraft will encounter at a specific time and location in the future, rather than static average values. As a result, the construction of the four-dimensional weather interpolator is completed, and the successful implementation of the four-dimensional spatiotemporal interpolation scheme is guaranteed.

[0052] In step S104, a physical enhancement feature set containing physical prior constraints is obtained based on standardized historical flight data and dynamic environmental parameters of trajectory points, and the physical enhancement feature set is input into a preset gradient boosting decision tree to obtain the aircraft fuel consumption prediction result.

[0053] In some embodiments, the physical enhancement feature set includes approximate Mach number, total energy state, gravitational work power, and aerodynamic drag factor, wherein... The approximate Mach number is:

[0054] The full energy state is:

[0055] The power of gravity is:

[0056] The aerodynamic drag factor is:

[0057] in, For an approximate Mach number, In a state of full energy, Power of work done by gravity It is the aerodynamic drag factor. For ground speed, For the local speed of sound, For the outside temperature, It is gravitational potential energy. As kinetic energy, For aircraft quality, It is the acceleration due to gravity. Standard atmospheric pressure altitude This represents the aircraft's actual speed. For the real-time full weight of the aircraft, Vertical velocity, A preset numerical stability constant (300 in this embodiment) is used to prevent numerical divergence caused by the denominator approaching zero when the aircraft is flying at low altitudes, ensuring the continuity and robustness of the feature across all altitude layers. This is the standard atmospheric pressure altitude.

[0058] It should be noted that the approximate Mach number Utilizing ground speed ( ) and local speed of sound (from static temperature) The ratio of the calculated speed of sound to the temperature value represents the air compressibility effect during high-altitude flight, correcting the prediction bias caused by the traditional model neglecting the change of sound speed with temperature. Furthermore, the calculation of the approximate Mach number includes a temperature cutoff protection step: if the obtained ambient temperature is lower than a preset temperature threshold (200 K in this embodiment), the temperature value input to the formula is forcibly corrected to the preset temperature threshold to prevent characteristic value drift caused by abnormal sound speed calculations in extreme high-altitude environments; full energy state. It integrates gravitational potential energy ( ) and kinetic energy This technology can capture the energy conversion patterns of "potential propulsion" or "dive acceleration" during flight, and can be used to solve the problem of inaccurate fuel consumption prediction during non-steady-state flight phases (such as climb and descent); gravity-driven work power. This reflects the work done by the aircraft against gravity per unit time, combined with real-time total weight ( ) and vertical velocity ( This can effectively improve fuel prediction accuracy during climb and descent phases; aerodynamic drag factor It is a semi-empirical feature, utilizing the square of velocity and height ( The inverse relationship between the two implicitly characterizes the impact of atmospheric density variations at different altitudes on aerodynamic drag.

[0059] Furthermore, in this embodiment of the invention, the physical augmentation feature set and the original features (such as latitude, longitude, altitude, real-time total weight, etc.) in the standardized historical flight data are fused to form a high-dimensional feature vector. Fuel flow is used as the model prediction target, and root mean square error (RMSE) is used as the loss function. The XGBoost gradient boosting decision tree is trained iteratively in multiple rounds using a 5-fold cross-validation method. The trained and optimized XGBoost model is used as the preset gradient boosting decision tree. The physical augmentation feature set of each stage of the historical flight is input into the preset model, and the predicted value of fuel flow at each moment during the historical flight is output. The physical augmentation feature set is input into the preset gradient boosting decision tree, and the model outputs the predicted value of fuel flow at each trajectory point of the future flight segment point by point. The predicted value of fuel flow at each trajectory point of the future flight segment is calculated by time integration or accumulation to obtain the total fuel consumption of the future flight segment. Thus, the aircraft fuel consumption prediction result can be obtained.

[0060] For example, iterative training of XGBoost gradient boosting decision trees first requires data preprocessing and cleaning. After reading the original QAR records, deduplication is performed to eliminate duplicate records caused by data transmission or storage errors, ensuring sample uniqueness. To address the imbalance problem in flight data, characterized by long cruise phases and short climb / descent phases, a hierarchical partitioning strategy is adopted. After grouping according to flight phase labels, the training and test sets are independently divided within each phase in a 7:3 ratio, ensuring consistent sample distribution across the training and test sets. Simultaneously, to balance training efficiency and memory usage, random downsampling is performed on the massive QAR training set, selecting... A representative subset of samples is used for model iteration. Secondly, in the physical feature construction stage, based on aerodynamic formulas, using original sensor data such as ground speed, altitude, total weight, and static temperature, four major physical enhancement features are calculated and directly used as model inputs: first, the approximate Mach number, used to characterize air compressibility; second, the total energy state, used to characterize the sum of potential energy and kinetic energy; third, the power of work done by gravity, used to characterize the power of work done against gravity; and fourth, the aerodynamic drag factor, used to characterize the level of aerodynamic drag. It is worth noting that in this embodiment, the feature engineering module directly outputs the combination of physical features and original features without performing additional normalization or principal component analysis (PCA) transformations. This design fully leverages the decision tree structure characteristics of the XGBoost algorithm, which is insensitive to feature dimensions and has strong robustness to the original numerical distribution, thus preserving the original dimensional information of physical features. Finally, in the model iterative training and tuning phase, a 5-fold cross-validation architecture is adopted, dividing the training samples into 5 equal parts, and alternately selecting 4 parts for training and 1 part for validation. The final model performance indicators (R², RMSE, MAE, MAPE) are the statistical values ​​of the 5 validation results. The model training is based on the XGBoost (Extreme Gradient Boosting) algorithm framework, setting the mean squared error as the loss function, and iterating by continuously adding new decision trees to fit the prediction residuals of the previous round. An early stopping mechanism is introduced during the training process, which monitors the validation set RMSE index in real time. The early stopping rounds are set to 50 (early_stopping_rounds=50). If the validation set RMSE does not decrease for 50 consecutive iterations, the training is terminated and rolled back to the optimal number of iteration rounds, effectively avoiding model overfitting.

[0061] It should be understood that, in the model training of this embodiment of the invention, the ratio and number of rounds can be adjusted according to the actual amount of data and the model convergence, and no specific limitations are made here.

[0062] To enable those skilled in the art to better understand the aircraft fuel consumption prediction method of the present invention, the following explanation will be provided in conjunction with specific embodiments.

[0063] Figure 2This is a schematic flowchart of an aircraft fuel consumption prediction method according to a specific embodiment of the present invention, as shown below. Figure 2 As shown, this embodiment of the invention adopts a dual-path parallel architecture. The left path is the historical training link, where the cleaned QAR data is directly trained into an XGBoost model after physical feature engineering (calculating Mach number, energy state, etc.). The right path is the future prediction link, where the reference trajectory is time-shifted, ground speed recalculated, and zero-fuel-consumption corrected. Combined with the future weather field generated by Pangu-Weather driven by ERA5 data, dynamic environmental parameters are obtained through 4D interpolation. These parameters are then input into the trained model for extrapolation through the same physical feature engineering. In this embodiment, aerodynamic physics formulas are transformed into computable feature engineering, injecting domain knowledge into the data model. Secondly, a large meteorological AI model (Pangu-Weather) is introduced to extrapolate the future four-dimensional weather field, replacing static meteorological data. Finally, through the "de-historicization" trajectory reconstruction logic (time shift + Haversine ground speed recalculation), the model can be decoupled from data dependencies at specific historical moments. The two links converge at the physical feature engineering and XGBoost model points, forming a unified technical closed loop of "training-extrapolation," enabling simulation prediction of any future flight segment.

[0064] As one possible approach, in practical applications such as future flight path planning or fuel consumption forecasting, the model operation process first requires obtaining the latitude, longitude, and altitude profiles of the planned flight path, shifting its timeline to the future target prediction date, and then utilizing existing large-scale meteorological models (such as Pangaea). Global meteorological field data (NetCDF format) for future moments is generated from weather data or reanalysis data (such as ERA5). A four-dimensional interpolator based on (Time, Level, Lat, Lon) is constructed. For each predicted point on the trajectory, the external static temperature (SAT) and wind speed vector (Win_Spd, Win_Dir) at the future moment are dynamically obtained through the interpolator. Then, based on the predicted environmental parameters (SAT, Win), physical enhancement features such as approximate Mach number, total energy state, gravitational work power, and aerodynamic drag factor are recalculated. Finally, the feature vector containing future environmental information is input into the pre-trained XGBoost model, and the predicted fuel flow rate (Pred_FF) at that moment is output.

[0065] To clearly define the scope of application and interface specifications of this invention, detailed input and output parameter definition tables are provided below. Table 1 is the input parameter table for the fuel consumption prediction model, and Table 2 is the system output parameter table. Table 1

[0066] Table 2

[0067] As another possible implementation, this embodiment of the invention takes the Airbus A320 as an example to detail the aircraft fuel consumption prediction method. In the data preparation and meteorological acquisition stages, for QAR data, historical flight QAR data of an airline's A320 fleet needs to be collected. This data includes key parameters such as time, latitude and longitude, altitude, ground speed, and fuel flow. Meteorological data acquisition is divided into three steps: initial field acquisition, future field generation, and data storage. The initial field uses ECMWFAPI (cdsapi) to download ERA5 high-resolution reanalysis data (including the ground and upper atmosphere) as the initial input to the large-scale meteorological model. The future field utilizes the Pangu-Weather large-scale meteorological model (ONNX format), using ERA5 data as the initial state, to perform multi-step iterative inference (1h / 3h / 6h / 24h). (h step size) to generate global meteorological field prediction data for future moments. Finally, the generated meteorological data is saved in standard NetCDF format (.nc), including variables such as geopotential height, specific humidity, temperature, and U / V wind components. A local meteorological data warehouse is built. Specifically, QAR data is first preprocessed. In the historical data branch, real meteorological data at historical moments are used to calculate physical features such as approximate Mach number (Machapprox), total energy state (Etotal), gravitational work power (Pgravity), and aerodynamic drag factor (Dfactor). The physical features are then merged with the original features (height, velocity, weight, etc.) to form a high-dimensional feature vector. Subsequently, an XGBoost model is built, using RMSE as the loss function and 5-fold cross-validation for training. Figure 3 This is a schematic diagram of the loss convergence curve of the model training process according to a specific embodiment of the present invention, as shown below. Figure 3 As shown, with the increase of the number of boosting rounds, the RMSE errors of both the training and validation sets converge rapidly and tend to stabilize, proving the rationality of the model parameter configuration and its good fitting ability. Figure 4 This is a schematic diagram comparing actual fuel consumption with predicted fuel consumption according to a specific embodiment of the present invention, such as... Figure 4 As shown, the actual fuel consumption curve and the predicted fuel consumption curve overlapped at high altitudes throughout the entire flight, proving that the model can accurately capture fuel consumption fluctuations during flight.

[0068] Furthermore, in the future prediction branch, spatiotemporal reconstruction is performed first. Given the planned flight path, the timeline is shifted to the future target date, and the ground speed is recalculated using the Haversine formula to eliminate historical GPS noise. Next, 4D meteorological interpolation is conducted. NetCDF meteorological data sequences within the target time period are loaded, and a RegularGridInterpolator4D interpolator based on (Time, Level, Lat, Lon) is constructed. For each point on the flight path, its predicted time, pressure altitude, and latitude and longitude are input, and dynamic interpolation is used to obtain the temperature (t) and wind speed vector (u, v) at that point. Finally, feature engineering and prediction are performed. Based on the interpolated meteorological environment, the physical augmentation feature set is recalculated, input into the pre-trained XGBoost model, and the predicted fuel flow rate is output. Figure 5 This is a schematic diagram illustrating the fuel consumption forecast for future flight segments in April, provided according to a specific embodiment of the present invention. Figure 5 As shown, this figure illustrates the fuel consumption evolution curve calculated by extrapolating the test flight trajectory to a future date (e.g., April 1, 2026) and combining it with real-time predicted meteorological conditions (including temperature and wind fields) for that period. This figure no longer relies on historical true values ​​but is entirely based on the future energy consumption estimate generated by the physical enhancement prediction logic of this invention, and provides the predicted total fuel consumption. Figure 6 A schematic diagram illustrating future flight segment predictions for March according to a specific embodiment of the present invention (2026-03-31). Combined with... Figure 5 and Figure 6 It can be seen that due to the changes in meteorological conditions (especially upper-level temperature, wind direction, and wind speed) in different months, even for the same or similar projected trajectories, their fuel consumption curves exhibit significant differences. This demonstrates the model's ability to keenly capture dynamic changes in meteorology; furthermore, Figures 7 to 9 The predictive performance of the model of this invention at different flight stages is demonstrated. Specifically: Figure 7 This is a scatter plot illustrating the prediction effect of the climb phase according to a specific embodiment of the present invention. Due to the introduction of the gravity-driven work power feature, the model can accurately compensate for the energy conversion requirements caused by drastic changes in altitude. The point set is closely distributed near the diagonal, and its coefficient of determination R... 2 The value reached 0.9868, effectively demonstrating the robustness of prediction under high power output conditions; Figure 8 This is a scatter plot illustrating the prediction effect of the cruise phase according to a specific embodiment of the present invention. By characterizing the effect of approximate Mach number on air compressibility, the model accurately captures the fuel consumption characteristics during high-altitude, high-speed cruise, maintaining an extremely low absolute error (MAE of only 99.22 kg / h). Figure 9This is a scatter plot illustrating the prediction effect of the descent phase according to a specific embodiment of the present invention. By introducing full energy state features, the model can accurately identify the engine idle state and the energy consumption differences under different descent gradients, and still exhibits good linear correlation in the low flow range; Figure 10 This is a schematic diagram illustrating the overall prediction comparison according to a specific embodiment of the present invention. Based on the performance at each stage, the proposed Physical Enhancement XGBoost architecture demonstrates excellent generalization ability across all operating conditions. Table 3 is a schematic table of evaluation metrics provided according to an embodiment of the present invention. As shown in Table 3, the specific evaluation metrics are as follows: Table 3

[0069] Furthermore, regarding key parameters and alternatives, the constants in the physical characteristic calculation formula can be fine-tuned according to different aircraft types or data characteristics; the meteorological interpolation method is not limited to linear interpolation, and more accurate cubic spline interpolation can be used when computational resources permit; the core prediction model XGBoost can be replaced with other interpretable tree ensemble models, such as LightGBM or CatBoost; the threshold for zero fuel consumption logic (ground speed 1 knot, altitude 2000 feet) can be adjusted according to the airport environment and aircraft type. Thus, this invention achieves high-precision historical fitting and future prediction of aircraft fuel consumption, providing strong data support for airlines' refined operations and energy conservation and emission reduction.

[0070] The aircraft fuel consumption prediction method provided by this invention acquires aircraft flight access records and meteorological reanalysis data, and obtains standardized historical flight data, standardized reference trajectory data, and future meteorological field data through data processing. The standardized reference trajectory data and future meteorological field data are processed separately to obtain a reconstructed trajectory for the future flight segment and a four-dimensional meteorological analysis result. Environmental parameters of the trajectory points are extracted by combining the two. A physical enhancement feature set is generated based on the standardized historical flight data and trajectory point environmental parameters, and this set is input into a preset gradient boosting decision tree to output the aircraft fuel consumption prediction result. This solves the problems of poor physical interpretability, inaccurate prediction in complex stages, and weak future prediction capabilities in related technologies, improving the physical interpretability and accuracy of aircraft fuel consumption prediction, and enhancing the precision and robustness of fuel consumption prediction.

[0071] Next, the aircraft fuel consumption prediction device according to an embodiment of the present invention is described with reference to the accompanying drawings.

[0072] Figure 11 This is a block diagram of an aircraft fuel consumption prediction device according to an embodiment of the present invention.

[0073] like Figure 11 As shown, the aircraft fuel consumption prediction device 10 includes: an acquisition module 100, a data analysis module 200, a four-dimensional meteorological analysis module 300, and a prediction module 400.

[0074] The acquisition module 100 is used to acquire aircraft flight access records and meteorological reanalysis data; the data analysis module 200 is used to perform feature extraction and spatiotemporal standardization processing on the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data, and to construct a four-dimensional environmental tensor based on the meteorological reanalysis data to obtain future meteorological field data; the four-dimensional meteorological analysis module 300 is used to reconstruct the kinematic state of the standardized reference trajectory data to obtain the reconstructed trajectory of the future flight segment, and to perform spatiotemporal interpolation analysis on the future meteorological field data to obtain the four-dimensional meteorological analysis results, and to obtain the dynamic environmental parameters of the trajectory points based on the reconstructed trajectory of the future flight segment and the four-dimensional meteorological analysis results; the prediction module 400 is used to obtain a physical enhancement feature set containing physical prior constraints based on the standardized historical flight data and the dynamic environmental parameters of the trajectory points, and to input the physical enhancement feature set into a preset gradient boosting decision tree to obtain the aircraft fuel consumption prediction results.

[0075] Furthermore, in some embodiments, the four-dimensional meteorological analysis module 300 is specifically used for: acquiring the time axis of standardized reference trajectory data, mapping the time axis of standardized reference trajectory data to the time interval of the future target flight segment, and obtaining the initial reconstructed trajectory of the future time label; calculating the ground speed of each trajectory point in the future target flight segment based on the future time label and latitude and longitude coordinate data corresponding to the initial reconstructed trajectory, and generating the future flight segment speed reconstruction trajectory based on the ground speed of each trajectory point in the future target flight segment; identifying target trajectory points in the future flight segment speed reconstruction trajectory where the ground speed of the trajectory point is less than a preset ground speed threshold and the standard pressure altitude of the trajectory point is less than a preset standard pressure threshold, correcting the fuel flow rate of the target trajectory point to a preset flow rate value, and using the adjusted future flight segment speed reconstruction trajectory as the future flight segment reconstruction trajectory.

[0076] Furthermore, in some embodiments, the four-dimensional meteorological analysis module 300 is also used to: acquire barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of future meteorological field data; align the barometric pressure layer data and standard pressure altitude data of future meteorological field data in the vertical dimension through a preset ISA standard atmospheric conversion mechanism to obtain parameter-aligned barometric pressure layer data and standard pressure altitude data; obtain a composite wind field vector based on the zonal wind component and meridional wind component of future meteorological field data according to a preset vector synthesis calculation method; and generate four-dimensional meteorological analysis results based on the parameter-aligned barometric pressure layer data, the parameter-aligned standard pressure altitude data, and the composite wind field vector using a pre-constructed four-dimensional interpolator.

[0077] Furthermore, in some embodiments, the physical enhancement feature set includes approximate Mach number, total energy state, gravitational work power, and aerodynamic drag factor, wherein, The approximate Mach number is:

[0078] The full energy state is:

[0079] The power of gravity is:

[0080] The aerodynamic drag factor is:

[0081] in, For an approximate Mach number, In a state of full energy, Power of work done by gravity It is the aerodynamic drag factor. For ground speed, For the local speed of sound, For the outside temperature, It is gravitational potential energy. As kinetic energy, For aircraft quality, It is the acceleration due to gravity. Standard atmospheric pressure altitude This represents the aircraft's actual speed. For the real-time full weight of the aircraft, Vertical velocity, This is a preset numerical stability constant used to prevent numerical divergence caused by the denominator approaching zero when the aircraft is flying at low altitudes, ensuring the continuity and robustness of the feature across all altitude layers. This is the standard atmospheric pressure altitude.

[0082] Furthermore, in some embodiments, the aircraft flight access record includes at least one of the following: timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow.

[0083] It should be noted that the foregoing explanation of the aircraft fuel consumption prediction method embodiment also applies to the aircraft fuel consumption prediction device of this embodiment, and will not be repeated here.

[0084] The aircraft fuel consumption prediction device provided in this embodiment of the invention acquires aircraft flight access records and meteorological reanalysis data, and obtains standardized historical flight data, standardized reference trajectory data, and future meteorological field data through data processing; it processes the standardized reference trajectory data and future meteorological field data respectively to obtain the future segment reconstructed trajectory and four-dimensional meteorological analysis results, and extracts trajectory point environmental parameters by combining the two; it generates a physical enhancement feature set based on the standardized historical flight data and trajectory point environmental parameters, inputs it into a preset gradient boosting decision tree, and outputs the aircraft fuel consumption prediction result. Thus, it solves the problems of poor physical interpretability, inaccurate prediction in complex stages, and weak future prediction ability in related technologies, improves the physical interpretability and prediction accuracy of aircraft fuel consumption prediction, and enhances the accuracy and robustness of fuel consumption prediction.

[0085] Figure 12 This is a schematic diagram of an electronic device provided according to an embodiment of the present invention. The electronic device may include: The memory 1201, the processor 1202, and the computer program stored on the memory 1201 and executable on the processor 1202.

[0086] When the processor 1202 executes the program, it implements the aircraft fuel consumption prediction method provided in the above embodiments.

[0087] Furthermore, electronic devices also include: Communication interface 1203 is used for communication between memory 1201 and processor 1202.

[0088] The memory 1201 is used to store computer programs that can run on the processor 1202.

[0089] The memory 1201 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0090] If the memory 1201, processor 1202, and communication interface 1203 are implemented independently, then the communication interface 1203, memory 1201, and processor 1202 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 12 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0091] Optionally, in a specific implementation, if the memory 1201, processor 1202, and communication interface 1203 are integrated on a single chip, then the memory 1201, processor 1202, and communication interface 1203 can communicate with each other through an internal interface.

[0092] Processor 1202 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention.

[0093] In addition, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described aircraft fuel consumption prediction method.

[0094] In addition, embodiments of the present invention also provide a computer program product, including a computer program that is executed to implement the above-described aircraft fuel consumption prediction method.

[0095] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0096] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0097] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting aircraft fuel consumption, characterized in that, Includes the following steps: Acquire aircraft flight access records and meteorological reanalysis data; Feature extraction and spatiotemporal standardization processing are performed on the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data. A four-dimensional environmental tensor is constructed based on the meteorological reanalysis data to obtain future meteorological field data. The kinematic state is reconstructed from the standardized reference trajectory data to obtain the reconstructed trajectory for the future flight segment. Spatiotemporal interpolation analysis is performed on the future meteorological field data to obtain four-dimensional meteorological analysis results. Based on the reconstructed trajectory for the future flight segment and the four-dimensional meteorological analysis results, the dynamic environmental parameters of the trajectory points are obtained. Based on the standardized historical flight data and the dynamic environmental parameters of the trajectory points, a physical enhancement feature set containing physical prior constraints is obtained. This physical enhancement feature set is then input into a preset gradient boosting decision tree to obtain the aircraft fuel consumption prediction result. The step of performing spatiotemporal interpolation analysis on the future meteorological field data to obtain four-dimensional meteorological analysis results includes: acquiring the barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of the future meteorological field data; aligning the barometric pressure layer data and standard pressure altitude data in the vertical dimension using a preset ISA standard atmospheric conversion mechanism to obtain parameter-aligned barometric pressure layer data and standard pressure altitude data; obtaining a composite wind field vector based on the zonal and meridional wind components of the future meteorological field data according to a preset vector synthesis calculation method; and generating four-dimensional meteorological analysis results based on the parameter-aligned barometric pressure layer data, the parameter-aligned standard pressure altitude data, and the composite wind field vector using a pre-constructed four-dimensional interpolator.

2. The aircraft fuel consumption prediction method according to claim 1, characterized in that, The process of reconstructing the kinematic state of the standardized reference trajectory data to obtain the reconstructed trajectory for the future flight segment includes: Obtain the time axis of the standardized reference trajectory data, map the time axis of the standardized reference trajectory data to the time interval of the future target flight segment, and obtain the initial reconstructed trajectory of the future time label; The ground speed of each trajectory point in the future target segment is calculated based on the future time tag and latitude / longitude coordinate data corresponding to the initial reconstructed trajectory, and a future segment speed reconstruction trajectory is generated based on the ground speed of each trajectory point in the future target segment. Target trajectory points in the future segment speed reconstruction trajectory are identified whose ground speed is less than a preset ground speed threshold and whose standard pressure altitude is less than a preset standard pressure threshold. The fuel flow rate of the target trajectory point is set to a preset flow rate value, and the adjusted future segment speed reconstruction trajectory is used as the future segment reconstruction trajectory.

3. The method according to claim 1, characterized in that, The physical enhancement feature set includes approximate Mach number, total energy state, gravitational work power, and aerodynamic drag factor, wherein... The approximate Mach number is: The total energy state is: The power of work done by gravity is: The aerodynamic drag factor is: in, For an approximate Mach number, In a state of full energy, Power of work done by gravity It is the aerodynamic drag factor. For ground speed, For the local speed of sound, For the outside temperature, It is gravitational potential energy. As kinetic energy, For aircraft quality, It is the acceleration due to gravity. Standard atmospheric pressure altitude This represents the aircraft's actual speed. For the real-time full weight of the aircraft, Vertical velocity, This is a preset numerical stability constant used to prevent numerical divergence caused by the denominator approaching zero when the aircraft is flying at low altitudes, ensuring the continuity and robustness of the feature across all altitude layers. This is the standard atmospheric pressure altitude.

4. The method according to claim 1, characterized in that, The aircraft flight access records include at least one of the following: timestamp, latitude and longitude, standard barometric altitude, ground speed, vertical speed, ambient temperature, real-time aircraft weight, true flight speed, and fuel flow.

5. An aircraft fuel consumption prediction device, characterized in that, include: The acquisition module is used to acquire aircraft flight access records and meteorological reanalysis data; The data analysis module is used to perform feature extraction and spatiotemporal standardization processing on the aircraft flight access records to obtain standardized historical flight data and standardized reference trajectory data, and to construct a four-dimensional environmental tensor based on the meteorological reanalysis data to obtain future meteorological field data. The four-dimensional meteorological analysis module is used to reconstruct the kinematic state of the standardized reference trajectory data to obtain the reconstructed trajectory of the future flight segment, and to perform spatiotemporal interpolation analysis on the future meteorological field data to obtain the four-dimensional meteorological analysis results. Based on the reconstructed trajectory of the future flight segment and the four-dimensional meteorological analysis results, the dynamic environmental parameters of the trajectory points are obtained. The prediction module is used to obtain a physical enhancement feature set containing physical prior constraints based on the standardized historical flight data and the dynamic environmental parameters of the trajectory points, and input the physical enhancement feature set into a preset gradient boosting decision tree to obtain the aircraft fuel consumption prediction result. The four-dimensional meteorological analysis module is specifically used to: acquire the barometric pressure layer data, standard pressure altitude data, zonal wind component, and meridional wind component of the future meteorological field data; and align the barometric pressure layer data and standard pressure altitude data of the future meteorological field data in the vertical dimension through a preset ISA standard atmospheric conversion mechanism to obtain the parameter-aligned barometric pressure layer data and standard pressure altitude data. Based on the zonal and meridional wind components of the future meteorological field data, a composite wind field vector is obtained according to a preset vector synthesis calculation method; based on the parameter-aligned pressure layer data, the parameter-aligned standard pressure height data, and the composite wind field vector, a four-dimensional meteorological analysis result is generated according to a pre-constructed four-dimensional interpolator.

6. The apparatus according to claim 5, characterized in that, The four-dimensional meteorological analysis module is specifically used for: Obtain the time axis of the standardized reference trajectory data, map the time axis of the standardized reference trajectory data to the time interval of the future target flight segment, and obtain the initial reconstructed trajectory of the future time label; The ground speed of each trajectory point in the future target segment is calculated based on the future time tag and latitude / longitude coordinate data corresponding to the initial reconstructed trajectory, and a future segment speed reconstruction trajectory is generated based on the ground speed of each trajectory point in the future target segment. Target trajectory points in the future segment speed reconstruction trajectory are identified whose ground speed is less than a preset ground speed threshold and whose standard pressure altitude is less than a preset standard pressure threshold. The fuel flow rate of the target trajectory point is set to a preset flow rate value, and the adjusted future segment speed reconstruction trajectory is used as the future segment reconstruction trajectory.

7. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, the processor executing the program to implement the aircraft fuel consumption prediction method as described in any one of claims 1-4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the aircraft fuel consumption prediction method as described in any one of claims 1-4.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the aircraft fuel consumption prediction method as described in any one of claims 1-4.

Images