Airplane fuel consumption calculation method and system based on ADS-B track data

Abstract

The present application relates to the technical field of aircraft operation state monitoring, in particular to an aircraft fuel consumption calculation method and system based on ADS-B track data; the method comprises the following steps: acquiring ADS-B track data and performing multi-level quality detection and hierarchical repair; acquiring aircraft model information and performance level identifier; fusing meteorological reanalysis data and performing space-time interpolation, and calculating derived flight parameters; constructing a multi-dimensional flight profile and identifying at least eight flight stages; extracting pilot operation habit characteristic parameters and controller command style characteristic parameters from the flight profile, and generating corresponding correction factors; injecting human factor correction factors into a multi-level fuel consumption model; calculating instantaneous fuel consumption rate and performing integration; standardizing output fuel consumption and emission data according to requirements; the present application incorporates two types of core human factors, i.e., pilot operation habit and controller command style, into the fuel consumption calculation framework based on ADS-B data, eliminates inherent calculation bias caused by human factors, and has higher calculation accuracy.

Description

Technical Field

[0001] This invention relates to the field of aircraft operational status monitoring technology, specifically to an aircraft fuel consumption calculation method and system based on ADS-B flight track data. Background Technology

[0002] With the continued growth of the global air transport industry, aviation fuel consumption and carbon emissions have received increasing attention. Accurately calculating aircraft fuel consumption is of great significance for airlines to optimize their operating strategies, reduce fuel costs, and meet environmental regulatory requirements.

[0003] In recent years, Automatic Dependent Surveillance-Broadcast (ADS-B) technology has been widely used globally. ADS-B acquires flight parameters such as position, speed, and altitude through GNSS receivers on aircraft and transmits them periodically via broadcast. It has advantages such as high data update rate, wide coverage, and low cost. In terms of instantaneous 95% accuracy, the ground speed accuracy of ADS-B data can reach 2.3 knots, which is better than the 7.3 knots of radar data. However, ADS-B data also has problems such as uneven distribution of data points, inconsistent interval width, and large variations in flight duration, making it difficult to align it with fuel consumption data.

[0004] Existing technologies have attempted to use ADS-B data for fuel consumption calculation. For example, Chinese patent applications CN109785462A (Aircraft Fuel Consumption Calculation System) and CN109738035A (Aircraft Fuel Consumption Calculation Method Based on ADS-B Flight Track Data) mainly employ the following methods: acquiring ADS-B data and storing it in a database; dividing flight operations into LTO and CCD phases using flight schedule information and the altitude information of ADS-B flight track recording points; determining the fuel flow rate for each phase based on the data; calculating the fuel consumption for each phase and summing them to obtain the total fuel consumption; these methods rely on empirical fuel consumption rate models in internationally recognized aircraft performance databases. Based on the aforementioned background technology, a key issue that has long been overlooked can be identified: human factors have a significant and systematic impact on aircraft fuel consumption, but existing technical solutions do not incorporate human factors into the fuel consumption calculation framework. Specifically, human factors mainly include two dimensions: on the one hand, the impact of differences in pilot operating habits on fuel consumption; on the other hand, the impact of differences in air traffic controller command styles on fuel consumption.

[0005] The main issues are: (1) The flight is simply divided into two stages, LTO and CCD, which is too coarse and cannot accurately reflect the characteristics of fuel consumption changes in different flight states; (2) The standard fuel flow parameters in the aircraft performance database are directly used, and the impact of meteorological conditions (especially high-altitude winds), real-time changes in aircraft weight, and changes in atmospheric temperature on fuel consumption during actual flight is not fully considered; (3) There is a lack of effective quality control measures for ADS-B raw data. Problems such as noise, outliers, and missing data in ADS-B data will affect the accuracy of fuel consumption calculation; (4) (5) There is a lack of graded optimization models for specific aircraft types. Different aircraft types have different fuel consumption characteristics under the same flight conditions. A single model is difficult to meet the calculation needs of multiple aircraft types. (6) The existing technology does not effectively connect the fuel consumption calculation results with the ICAOCORSIA emission accounting framework. It lacks a standardized output format, which limits the regulatory application value of the calculation results. (7) The existing technology completely ignores the systematic impact of two core human factors on fuel consumption: differences in pilot operating habits and differences in controller command style. This results in inherent biases in the fuel consumption calculation model, which cannot reflect the main role of humans in the human-machine-environment closed-loop system. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies, this invention provides an aircraft fuel consumption calculation method and system based on ADS-B flight track data. This method can utilize publicly available ADS-B flight track data to achieve high-precision calculation of instantaneous and phased fuel consumption throughout the entire operation of an aircraft. This is achieved through multi-layer quality inspection and repair of the data, fusion and spatiotemporal interpolation of meteorological data, construction of multi-dimensional flight profiles, extraction and quantification of pilot operating habit characteristics, identification and modeling of controller command style characteristics, and selection and optimization of multi-level fuel consumption calculation models. The calculation results are then standardized into an emission data format that meets the requirements of ICAO CORSIA accounting.

[0007] The aircraft fuel consumption calculation method based on ADS-B track data described in this invention includes: S1. Obtain aircraft ADS-B track data; S2. Perform integrity, rationality, and time-series consistency checks on ADS-B data, mark outliers, and perform graded repairs. S3. Obtain aircraft type information and performance level identification based on aircraft identification; S4. Acquire matching meteorological data and interpolate it spatiotemporally to each track point; S5. Determine the aircraft's speed parameters, energy parameters, and environmental correction parameters based on ADS-B data and meteorological data in the current flight state; S6. Construct a multi-dimensional flight profile that includes altitude, speed, acceleration, vertical motion, and energy state profiles to divide the aircraft flight process into stages and determine the operational stages of the aircraft throughout the entire flight process. S7. Select a fuel consumption calculation model from a multi-level fuel consumption model library based on aircraft type information and performance level identifier. The multi-level fuel consumption model library includes a basic layer model, a standard layer model, and an enhanced layer model. The basic layer model is a lookup model established based on an aircraft performance parameter library. The standard layer model is a regression model established based on historical flight data recorder data. The enhanced layer model is a neural network model trained based on historical flight data recorder data. S8. Using flight profile characteristics, meteorological data and derived parameters as input, calculate the instantaneous fuel consumption rate using a modified model; S9. Perform second-order smooth monotonic interpolation on the instantaneous fuel consumption rate to obtain a continuous curve, and integrate to obtain the fuel consumption of each stage and the total fuel consumption. S10. Calculate aircraft emissions based on fuel consumption, fuel type, and preset emission factors, and perform field mapping, unit unification, integrity verification, and structured output on fuel consumption and emission data.

[0008] Specifically, S6 includes S6A: extracting characteristic parameters of pilot operating habits and controller command style from flight profiles, and generating correction factors through a human factor quantification model. The extraction of characteristic parameters of pilot operating habits includes: The climb rate fluctuation coefficient, the descent profile curvature change, and the height maintenance deviation are extracted from the vertical motion profile as vertical profile manipulation characteristic parameters. Acceleration fluctuation coefficient, deceleration gradient fluctuation coefficient, cruise speed fluctuation coefficient, and approach speed deviation are extracted from the velocity profile and acceleration profile as speed management characteristic parameters; Energy conversion deviation, energy profile deviation, and energy distribution deviation are extracted from the energy state profile as energy management characteristic parameters. The above feature parameters are standardized to construct a comprehensive feature vector of pilot operating habits; The pilot's operating habit feature vector is matched with the feature vector of the same type of flight in the historical database to identify the type of operating habit of the pilot currently operating the flight. The type of operating habit includes one of the following: economic, standard and aggressive. Based on the identified type of pilot maneuvering, the baseline correction coefficient corresponding to that type is queried from the pre-built quantitative model library of human factors, and the vector distance between the comprehensive feature vector of the pilot maneuvering habits of the flight and the center vector of the corresponding maneuvering habit type is calculated. A pilot control habit correction factor is generated based on the reference correction factor, the vector distance, and the pre-calibrated deviation correction factor.

[0009] Specifically, the climb rate fluctuation coefficient is the ratio of the standard deviation of the vertical rate sequence during the climb phase to the average absolute value of the vertical rate sequence. The change in curvature of the descent profile is the average of the absolute values ​​of the second difference of the height sequence with respect to time during the descent phase; The altitude maintenance deviation is the average of the absolute values ​​of the differences between the barometric altitude of each track point and the reference altitude of that level flight phase, where the reference altitude is the median of the barometric altitude during that level flight phase. Wherein, the acceleration fluctuation coefficient is the ratio of the standard deviation of the acceleration sequence within the corresponding flight phase to the average absolute value of the acceleration sequence; The deceleration gradient fluctuation coefficient is the ratio of the standard deviation of the velocity change rate sequence during the deceleration phase to the average absolute value of the velocity change rate sequence. The cruise speed fluctuation coefficient is the ratio of the standard deviation of the vacuum speed sequence to the average value of the vacuum speed sequence during the cruise phase. The approach speed deviation is the average of the absolute values ​​of the differences between the vacuum speed at each track point and the approach speed of the corresponding aircraft type target during the approach phase; The energy conversion deviation is the average of the absolute values ​​of the differences between the current flight's energy altitude change per unit time and the historical benchmark energy altitude change per unit time for the same aircraft type during the same period. The energy profile deviation is the average of the absolute values ​​of the differences between the current flight energy altitude sequence and the historical benchmark energy altitude sequence of the same aircraft type at the same stage. The energy allocation deviation is the average of the absolute values ​​of the differences between the current flight energy allocation factor sequence and the historical benchmark energy allocation factor sequence of the same aircraft type and period.

[0010] Specifically, the extraction of controller command style characteristic parameters as described in S6A includes: The heading change frequency, altitude change frequency, average length of level flight segment, and route offset are extracted from ADS-B track data as track morphology characteristic parameters. Based on the time series data of ADS-B tracks, implicit control instructions are identified through rule matching methods. The types of control instructions include at least heading instructions, altitude instructions, and speed instructions. The type, time, location, and parameter changes before and after execution of each instruction are recorded. Based on the identified control command sequence, the command density, command complexity, command lead time, speed intervention tendency, and high intervention tendency are statistically analyzed as command style characteristic indicators, and after standardization, a command style characteristic vector of controllers is constructed. Based on the geographical location and timestamp of the flight path, obtain the control sector information corresponding to the current flight. The controller's command style feature vector is compared with the average command style feature vector of the same sector, time period, and type of flight in the historical database to calculate the command style deviation degree, and a controller command style correction factor is generated based on the pre-constructed controller command style and fuel consumption impact mapping model.

[0011] Specifically, S7 includes S7A: injecting human factor correction factors into the selected model to form a corrected model, specifically: The pilot's operating habits correction factor and the controller's command style correction factor are weighted and fused together, and the weighting is dynamically adjusted according to the flight phase. During the takeoff roll, initial climb, en route climb, and landing roll phases, the weight of the pilot's handling habits correction factor is greater than the weight of the controller's command style correction factor. During cruise and step climb phases, the weight of the controller's command style correction factor is greater than the weight of the pilot's handling habits correction factor. During descent and approach, the pilot's handling habits correction factor and the controller's command style correction factor have equal weights. The weighted fusion yields a comprehensive human factor correction factor; For the base layer model and the standard layer model, a multiplicative correction method is adopted, which multiplies the comprehensive human factor correction factor with the instantaneous fuel consumption rate output by the model to obtain the corrected instantaneous fuel consumption rate. For the enhancement layer model, the comprehensive human factor correction factor is used as an additional input feature of the neural network model to participate in the forward propagation calculation.

[0012] Specifically, the method for constructing the multi-level fuel consumption model library in S7 includes: Collect a large amount of real flight data from QAR or FDR of historical flights, extract multi-dimensional flight profile features and real fuel consumption for each flight, and extract pilot control input parameters, control sector information and corresponding control instruction sequences experienced by each flight. Using pilot control input parameters extracted from QAR data as features, an unsupervised clustering algorithm was used to perform cluster analysis on the pilots' historical flight samples, identifying three types of pilot control habits: economic, standard, and aggressive. Construct a database of control sectors, time periods, and fuel consumption benchmarks; analyze the deviation between the controller command style feature vector of each specific flight and the benchmark feature vector of the same sector and time period; establish the mapping relationship between the deviation and the actual fuel consumption deviation; and construct a mapping model of controller command style and fuel consumption impact. As new flight data accumulates, the clustering model of pilot operating habits and the mapping model of controller command style and fuel consumption impact are incrementally updated and optimized.

[0013] Specifically, the waypoints in S1 include at least some of the following fields: timestamp, ICAO 24-bit address code, flight call sign, longitude, latitude, barometric altitude, geometric altitude, ground speed, heading, and vertical rate. The integrity check described in S2 is used to determine whether any of the required fields in each track point are missing. The required fields include timestamp, longitude, latitude, barometric altitude, and ground speed. The rationality detection is used to determine whether the values ​​of each field are within a preset physical reasonable range; The temporal consistency detection is used to determine whether the rate of change of parameters between adjacent waypoints exceeds a preset allowable change threshold. The graded repair includes: for a single isolated abnormal track point, a sliding window mean filter or median filter is used to replace the abnormal point data with the statistical value of the normal data within the window; for a data missing segment consisting of multiple consecutive abnormal track points, when the length of the missing segment is less than a preset threshold, linear interpolation or cubic spline interpolation is used to complete the data; when the length of the missing segment is greater than or equal to the preset threshold, a data quality warning mark is added to the track segment.

[0014] Specifically, in S6: The altitude profile is composed of a sequence of changes in air pressure altitude over time; The velocity profile includes the ground velocity profile and the vacuum velocity profile, which are composed of time-varying sequences of ground velocity and vacuum velocity, respectively. The acceleration profile is composed of a sequence of the rate of change of ground velocity with respect to time; The vertical motion profile consists of a sequence of vertical velocity changes over time. The energy state profile consists of a sequence of energy height changes over time; The operational phases of an aircraft during its flight include takeoff roll, initial climb, en route climb, cruise, step climb, descent, approach, and landing roll. The identification methods for operational phases include: distinguishing the climb phase, level flight phase, and descent phase based on the sign and magnitude of the vertical rate; within the climb phase, distinguishing the initial climb phase, en route climb phase, and step climb phase based on the pressure-altitude range; within the descent phase, distinguishing the descent phase and approach phase based on the pressure-altitude range and the magnitude of the vertical rate; and distinguishing the takeoff roll phase and landing roll phase based on the trends in altitude and ground speed changes.

[0015] Specifically, in S7: The basic layer model is based on the performance parameters in the aircraft performance parameter library, and the standard fuel consumption rate parameters are obtained by looking up tables according to the aircraft type, flight phase, altitude and speed. The standard layer model is constructed by collecting historical flight data from the flight data recorder of the target aircraft and grouping it according to flight phases to establish a multiple regression model. The dependent variable of the multiple regression model is the fuel consumption rate, and the independent variables include altitude, vacuum speed, acceleration, vertical speed, energy distribution factor and weather correction factor. The enhancement layer model is constructed by training historical data from the flight data recorder using a neural network. The input layer of the neural network receives multidimensional flight profile feature parameters, meteorological parameters, and derived flight parameters, while the output layer outputs the instantaneous fuel consumption rate.

[0016] Specifically, in S10: The human factor analysis report includes: the pilot's operating habits type and confidence level identified for this flight, the value and contribution ratio of the pilot's operating habits correction factor, the deviation degree of the controller's command style characteristic vector, the value and contribution ratio of the controller's command style correction factor, and the value and phased decomposition value of the comprehensive human factor correction factor.

[0017] On the other hand, the present invention also provides an aircraft fuel consumption calculation system based on ADS-B track data, the system comprising: The data acquisition module is used to obtain ADS-B track data of the target aircraft from the ADS-B ground station network or data service provider; The data quality control module, connected to the data acquisition module, is used to perform integrity detection, rationality detection, and temporal consistency detection on the ADS-B track data, and to perform graded repair processing on abnormal data. The aircraft type identification module, connected to the data quality control module, is used to obtain the aircraft type information and performance level identifier of the target aircraft from the aircraft basic information database based on the unique aircraft identifier in the ADS-B track data. The meteorological data fusion module is used to acquire meteorological data that matches the spatiotemporal range of ADS-B track data, and to interpolate the meteorological data to each track point location using a spatiotemporal interpolation method. The flight parameter derivation module is connected to the data quality control module and the meteorological data fusion module. It is used to calculate derived flight parameters based on the ADS-B track data and the interpolated meteorological data. The derived flight parameters include at least vacuum speed, Mach number, energy altitude, energy distribution factor and meteorological correction factor. The flight profile construction module, connected to the flight parameter derivation module, is used to construct a multi-dimensional flight profile including altitude profile, velocity profile, acceleration profile, vertical motion profile and energy state profile, and to identify flight phases based on the multi-dimensional flight profile. The flight phases include at least the takeoff roll phase, initial climb phase, en-route climb phase, cruise phase, step climb phase, descent phase, approach phase and landing roll phase. The human factor feature extraction and quantification module is connected to the flight profile construction module. It is used to extract pilot operation habit feature parameters and controller command style feature parameters from the multi-dimensional flight profile and the identified flight phases. It also generates pilot operation habit correction factors and controller command style correction factors through a pre-built human factor influence quantification model library. The model selection module, connected to the model identification module, is used to select the corresponding fuel consumption calculation model level from a pre-built multi-level fuel consumption model library based on the model information and performance level identifier. The human factor correction and fusion module is connected to the human factor feature extraction and quantification module and the model selection module. It is used to inject the generated pilot operation habit correction factor and controller command style correction factor as human factor correction items into the selected fuel consumption calculation model to form a fuel consumption calculation model with human factor correction. The fuel consumption calculation module is connected to the flight profile construction module and the human factor correction fusion module, and is used to calculate the instantaneous fuel consumption rate of each flight path point using the fuel consumption calculation model with human factor correction. The fuel consumption integration and output module, connected to the fuel consumption calculation module, is used to perform second-order smooth monotonic interpolation and integration calculation on the instantaneous fuel consumption rate to obtain the fuel consumption of each flight phase and the total fuel consumption of the flight, and to perform field mapping, unit unification, integrity verification and structured output on the fuel consumption data and emission data.

[0018] The beneficial effects of this invention are as follows: This invention systematically incorporates two core human factors—differences in pilot operating habits and differences in controller command styles—into a fuel consumption calculation method based on ADS-B flight track data, filling the gap in the quantification of the influence of human factors in existing technologies. By extracting characteristic parameters of pilot operating habits and controller command styles from ADS-B flight tracks and generating corresponding human factor correction factors, the fuel consumption calculation deviation caused by individual differences in pilots and differences in controller command is effectively eliminated, further improving the calculation accuracy.

[0019] Experimental verification shows that after incorporating corrections for human factors, the average absolute percentage error in total fuel consumption calculation is further reduced by about 15% to 20%, which has important supporting value for application scenarios such as refined management of airlines, fuel-saving performance evaluation of pilots, and green command capability evaluation of air traffic controllers. Attached Figure Description

[0020] Figure 1 This is a flowchart of the aircraft fuel consumption calculation method based on ADS-B track data provided in this embodiment of the invention; Figure 2 This is a schematic diagram of flight phase division provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the multi-level fuel consumption model library provided in the embodiments of the present invention; Figure 4 This is a structural block diagram of the aircraft fuel consumption calculation system based on ADS-B track data provided in this embodiment of the invention; Figure 5 This is a schematic diagram of the internal structure of the human factor feature extraction and quantification module provided in an embodiment of the present invention; Figure 6 This is a flowchart of the pilot control habit correction factor generation process provided in an embodiment of the present invention; Figure 7 This is a flowchart of the controller command style correction factor generation process provided in an embodiment of the present invention.

[0021] 101. Data Acquisition Module; 102. Data Quality Control Module; 103. Aircraft Type Identification Module; 104. Meteorological Data Fusion Module; 105. Flight Parameter Derivation Module; 106. Flight Profile Construction Module; 107. Model Selection Module; 108. Multi-level Fuel Consumption Model Library; 109. Fuel Consumption Calculation Module; 110. Fuel Consumption Integration and Output Module; 111. Human Factor Feature Extraction and Quantification Module; 112. Human Factor Correction and Fusion Module; 113. Human Factor Influence Quantification Model Library. Detailed Implementation

[0022] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0023] Example 1: Please see Figures 1-7 As shown, this embodiment provides a method for calculating aircraft fuel consumption based on ADS-B track data, including the following steps: S1: Acquire ADS-B track data; The methods for obtaining ADS-B track data include, but are not limited to: receiving ADS-B messages broadcast by aircraft in real time through ADS-B ground station receivers located in various locations; obtaining data in batches through historical data interfaces of commercial ADS-B data aggregation service platforms (such as Flightradar24, FlightAware, OpenSkyNetwork, etc.); obtaining globally covered ADS-B data through satellite-based ADS-B systems (such as the Aireon system); ADS-B data is transmitted in 1090MHz Extended Message (1090ES) or 978MHz Universal Access Transceiver (UAT) data link format, and after being decoded by ground stations, it forms track point records containing multiple fields; Each waypoint in ADS-B track data typically contains the following fields: (1) Timestamp: Records the UTC time corresponding to the track point, with an accuracy of usually in the second range; (2) ICAO 24-bit address code: a unique identifier for an aircraft, uniformly assigned by ICAO to identify a specific individual aircraft; (3) Flight call sign: The flight identification code assigned by the airline, such as CES5101; (4) Longitude and latitude: Geographical coordinates expressed in the WGS-84 coordinate system. The accuracy of longitude and latitude is usually 5 decimal places (about 1 meter level accuracy). (5) Barometric Altitude: The flight altitude measured by the aircraft's air data system and broadcast via ADS-B messages, in feet; (6) Geometric height: The height of the ground as measured by a GNSS receiver, in feet; (7) Ground speed: The horizontal speed of an aircraft relative to the ground, measured in knots; (8) Heading: The direction of flight of an aircraft (true heading or magnetic heading), in degrees; (9) Vertical rate: the rate of change of aircraft altitude. A positive value indicates climbing and a negative value indicates descent. The unit is feet per minute.

[0024] S2: Data quality inspection and repair; The data quality inspection in this embodiment includes the following three levels of inspection: (1) Integrity check: Check each of the required fields in each track point one by one; the required fields include timestamp, longitude, latitude, barometric altitude and ground speed; if any of the above fields is missing, the track point is marked as incomplete; (2) Reasonableness check: The physical reasonableness of the values ​​of each field is judged; the preset physical reasonable range of each field is: air pressure altitude: -1000 feet to 60000 feet; ground speed: 0 knots to 650 knots; vertical speed: -8000 feet / minute to 8000 feet / minute; longitude and latitude: conform to the valid range of global longitude and latitude (longitude -180° to 180°, latitude -90° to 90°); data that exceeds the above range is marked as unreasonable; (3) Timing consistency detection: Calculate the rate of change of parameters between adjacent waypoints and determine whether it exceeds the preset allowable change threshold; the ground speed change between two adjacent points is divided by the time difference to obtain the acceleration value, which should not exceed the maximum acceleration capability of the aircraft (typical value is 0.2g to 0.3g, i.e., about 6.4 knots / second to 9.6 knots / second); the altitude change rate between two adjacent points should not exceed the maximum vertical speed of the aircraft (typical value is 4000 feet / minute to 6000 feet / minute); changes exceeding the threshold are marked as inconsistent; For detected abnormal waypoints, this embodiment adopts a hierarchical repair strategy: for a single isolated abnormal waypoint, a sliding window filtering method is used for repair, and the size of the sliding window is usually 5 waypoints; for data missing segments composed of multiple consecutive abnormal waypoints, an interpolation method is used for data completion; for cases where the vertical velocity field is missing, it is calculated by dividing the height difference between adjacent points by the time difference.

[0025] S3: Obtaining model information and performance level identification; Based on the ICAO 24-bit address code in the ADS-B data, a query is performed in the Aircraft Basic Information Database to obtain the registration information of the target aircraft; through related queries, aircraft type information, engine type and performance level identifier can be obtained; the performance level identifier is divided into three levels: basic, standard and enhanced.

[0026] S4: Meteorological data acquisition and spatiotemporal interpolation; In this embodiment, the China Atmospheric Reanalysis Product (CRA-40) developed by the China National Meteorological Information Center is used as the main source of meteorological data; the data is gridded meteorological data with a temporal resolution of 1 hour and a spatial resolution of 0.25°×0.25° (approximately 28 km×28 km); In the time dimension, linear time interpolation is used; in the spatial dimension, bilinear spatial interpolation is used; and in the vertical dimension, logarithmic pressure linear interpolation is used to interpolate between different pressure layers of meteorological data based on the pressure altitude of the track points.

[0027] S5: Calculation of derived flight parameters; Based on raw ADS-B data and interpolated meteorological data, vacuum speed, Mach number, energy height, energy distribution factor, and meteorological correction factor are calculated. Vacuum speed is calculated by subtracting the ground speed vector from the upper-level wind vector. Energy height equals the pressure height plus the kinetic energy equivalent height. Energy distribution factor equals the ratio of the rate of change of potential energy to the rate of change of kinetic energy. Meteorological correction factor is used to quantify the comprehensive impact of meteorological conditions on fuel consumption.

[0028] S6: Multidimensional flight profile construction and flight phase identification; This embodiment constructs a multi-dimensional flight profile that includes altitude profile, velocity profile, acceleration profile, vertical motion profile, and energy state profile, and finely divides the flight operation process into the takeoff roll phase, initial climb phase, en-route climb phase, cruise phase, step climb phase, descent phase, approach phase, and landing roll phase.

[0029] S6A: Feature extraction and quantification of human factors; Figure 5 The internal structure of the human factor feature extraction and quantification module is shown. Figure 6 The process for generating pilot handling habit correction factors is shown. Figure 7 This demonstrates the process for generating controller command style correction factors; The aim is to extract characteristic parameters reflecting pilot operating habits and controller command styles from ADS-B track data and multi-dimensional flight profiles, and quantify them into correction factors that can be directly used to correct fuel consumption models.

[0030] S6A-1: Pilot control habit feature extraction; (a) Extraction of vertical profile manipulation features: The following feature parameters are extracted from the vertical motion profile and height profile constructed from S6: Climb rate fluctuation coefficient: During the climb phase (between 10,000 feet and cruising altitude), the standard deviation of the vertical rate sequence is calculated. A smaller standard deviation indicates smoother and more economical climb control by the pilot; a larger standard deviation indicates more frequent changes in vertical rate during the climb, potentially indicating aggressive control. Typical values ​​range from 50 to 300 feet per minute. Descent profile smoothness: Calculates the standard deviation of the rate of change of vertical velocity (i.e., vertical acceleration) during the descent and approach phases; this indicator reflects the pilot's control smoothness during descent. Typical values ​​range from 30 to 200 feet per minute². Altitude holding accuracy: During the cruise phase, the root mean square value of the difference between the actual barometric altitude and the target cruise altitude (usually the median barometric altitude of the cruise segment) is calculated; this indicator reflects the pilot's ability to maintain altitude during the cruise phase and is closely related to fuel efficiency.

[0031] (b) Speed ​​management feature extraction: The following feature parameters are extracted from the velocity and acceleration profiles: Acceleration consistency: During takeoff and initial climb, calculate the coefficient of variation (the ratio of standard deviation to mean) of the acceleration sequence; good acceleration consistency indicates that the pilot has adopted a stable and smooth thrust application strategy. Deceleration gradient uniformity: During approach and landing phases, the variance of the deceleration sequence is calculated; this indicator reflects the pilot's control smoothness during deceleration. Cruise speed stability: During the cruise phase, the standard deviation of the ground speed sequence is calculated; the smaller this index is, the more stable the cruise speed management is, and the closer the fuel consumption is to the optimal value. Approach speed management precision: During the approach phase, calculate the integral of the absolute value of the difference between the actual ground speed and the target approach speed (determined according to the aircraft type); the smaller the integral value, the more precise the pilot's speed management.

[0032] (c) Energy management feature extraction: The following characteristic parameters are extracted from the energy state profile: Energy conversion efficiency: During the climb phase, the average ratio of the rate of change of energy at altitude to the vertical rate is calculated; the closer this ratio is to the theoretical optimum (approximately 1.0 to 1.1), the more effective the pilot's management of potential and kinetic energy conversion. Energy profile deviation: During the descent phase, the integral of the deviation between the actual energy altitude change rate and the ideal continuous descent energy management curve is calculated; the ideal curve is a reference curve in which energy altitude decreases linearly with distance under conditions of continuous descent without level flight; the smaller this integral value, the better the pilot's descent energy management. Energy allocation deviation: This analyzes the changing patterns of the energy allocation factor during the climb, cruise, and descent phases. Economical pilots allocate more energy to potential energy during the climb phase (energy allocation factor greater than 1 and remaining stable), maintain energy balance during the cruise phase, and utilize potential energy to convert it into kinetic energy during the descent phase to reduce fuel consumption.

[0033] (d) Construction of comprehensive feature vectors for manipulation habits and type recognition: All the extracted feature parameters are standardized to construct a comprehensive feature vector of pilot operating habits with dimensions of 10 to 15. The Euclidean distance between the feature vector of the current flight and the center vectors of the same aircraft type and various handling habit types pre-stored in the human factor influence quantification model library is calculated. The type with the smallest distance is identified as the handling habit type of the pilot currently operating the flight. The handling habit types are divided into the following three categories: Economy type: A handling style that prioritizes fuel economy, characterized by gentle acceleration, continuous climb (reducing the level flight phase), preference for high cruise altitude, delayed flap deployment, and continuous descent approach; the baseline correction factor is approximately 0.94 (meaning a fuel saving of about 6% compared to the standard type). Standard type: The operating style follows standard operating procedures, and all operating parameters are at the median level of the statistical distribution; the baseline correction factor is 1.00. Aggressive type: A handling style primarily aimed at shortening flight time or simplifying operations, characterized by rapid acceleration, high climb rate, early attainment of target altitude, early flap deployment, and a preference for step descent; the baseline correction factor is approximately 1.06 (meaning it consumes about 6% more fuel than the standard type).

[0034] (e) Generation of pilot handling habit correction factors: Under the same aircraft type, weather conditions, and route, this is the ratio of fuel consumption to a baseline level due to differences in pilot operating habits. In the above implementation, the current flight's waypoint data is first acquired. This waypoint data includes timestamps, barometric altitude, vacuum velocity, vertical speed, acceleration, aircraft type identifier, and flight phase identifier. If the original data does not directly contain the flight phase identifier, the flight can be divided into climb, cruise, level flight, descent, deceleration, and approach phases based on barometric altitude change trends, vertical speed thresholds, vacuum velocity change trends, and flight operation time sequence. The waypoints within each phase are arranged chronologically, and data with missing values, anomalous jump values, or inconsistent sampling intervals are interpolated, removed, or resampled to ensure that subsequent feature extraction is based on a unified time scale.

[0035] In the extraction of vertical profile control feature parameters, for the climb phase, a vertical rate sequence is acquired, its standard deviation is calculated, and its mean absolute value is also calculated. The ratio of these two values ​​is used as the climb rate fluctuation coefficient. This parameter characterizes the stability of the rate of climb maintained by the pilot or autopilot control strategy during the climb; a larger value indicates more significant climb rate fluctuation. To avoid instability caused by the mean absolute value approaching zero, a preset minimum positive number ε can be added to the denominator. For the descent phase, the second-order difference of the altitude sequence relative to time is acquired. The second-order difference represents the magnitude of change between adjacent altitude change rates. The average of these absolute values ​​yields the change in descent profile curvature. This parameter reflects the curvature of the descent trajectory and the continuity of the profile during the descent. For the level flight phase, the median pressure altitude within this phase is used as the reference altitude. The average of the absolute values ​​of the differences between the pressure altitude at each trackpoint and the reference altitude is then calculated to obtain the altitude holding deviation. Using the median as the reference altitude can reduce the impact of a few abnormal altitude points on the reference value, making the altitude maintenance deviation more stably reflect the altitude control level during level flight.

[0036] During the extraction of speed management characteristic parameters, for the acceleration sequence within the corresponding flight phase, the ratio of the standard deviation of acceleration to the average absolute value of acceleration is calculated to obtain the acceleration fluctuation coefficient. This parameter characterizes the smoothness of acceleration or deceleration operations during speed adjustments. For the deceleration phase, the speed change rate sequence is calculated based on the vacuum speed difference and time difference between adjacent track points, and then the ratio of the standard deviation to the average absolute value of this speed change rate sequence is calculated to obtain the deceleration gradient fluctuation coefficient. This parameter reflects whether the deceleration process is uniform and whether there are control characteristics such as abrupt throttle reduction, drag release devices, or discontinuous deceleration. For the cruise phase, the ratio of the standard deviation to the average value of the vacuum speed sequence is calculated to obtain the cruise speed fluctuation coefficient, which characterizes the degree of speed stability during cruise. For the approach phase, the target approach speed corresponding to the current aircraft type is read, and the average value of the absolute values ​​of the differences between the vacuum speed and the target approach speed at each track point within the approach phase is calculated to obtain the approach speed deviation. The target approach speed can be determined from the aircraft performance database, operations manual data, or historical stable approach data of the same aircraft type.

[0037] In the process of extracting energy management characteristic parameters, the energy altitude of an aircraft at any track point can be represented as the sum of pressure altitude and velocity-energy altitude, i.e., energy altitude equals pressure altitude plus the square of vacuum velocity divided by twice the gravitational acceleration. The energy altitude sequence of the current flight is time-diffused to obtain the change in energy altitude per unit time, and this is compared with the benchmark change in energy altitude per unit time for the same aircraft type and stage in the past. The average of the absolute values ​​of the differences is taken to obtain the energy conversion deviation. This parameter characterizes the degree of deviation of the current flight in the process of altitude-energy and velocity-energy conversion. Further, the energy altitude sequence of the current flight is interpolated and aligned with the benchmark energy altitude sequence of the same aircraft type and stage in the past according to a unified stage process, and the average of the absolute values ​​of the differences at corresponding positions is calculated to obtain the energy profile deviation. The energy allocation factor can represent the proportion of velocity-energy altitude in the total energy altitude, or the proportional relationship between the change in velocity-energy and the change in total energy. The energy allocation factor sequence of the current flight is aligned and compared with the benchmark energy allocation factor sequence of the same aircraft type and stage in the past, and the average of the absolute values ​​of the differences is calculated to obtain the energy allocation deviation. Therefore, the pilot's energy management habits can be characterized from three aspects: energy conversion rate, overall energy profile, and the relationship between energy distribution at altitude and at speed.

[0038] After extracting the aforementioned features, the climb rate fluctuation coefficient, descent profile curvature change, altitude hold deviation, acceleration fluctuation coefficient, deceleration gradient fluctuation coefficient, cruise speed fluctuation coefficient, approach speed deviation, energy conversion deviation, energy profile deviation, and energy allocation deviation are combined to form the original feature set. Each feature parameter is then standardized, for example, using the mean and standard deviation of historical samples from the same aircraft model for Z-score standardization, resulting in a comprehensive feature vector of pilot operating habits. The standardized features are on similar numerical scales, avoiding unreasonable weighting of some features in similarity calculations due to differences in dimensions.

[0039] When identifying pilot handling habit types, the comprehensive feature vector of the current flight's pilot handling habits is matched with the feature vectors of similar flights of the same aircraft type in the historical database. The historical database can pre-store standardized feature vectors, flight phase information, handling habit type labels, and center vectors for each type of flight of the same aircraft type. During matching, the Euclidean distance between the current feature vector and the center vectors of the economy, standard, and aggressive types can be calculated, or the cosine similarity between the current feature vector and the feature vectors of historical samples can be calculated. When using Euclidean distance, the type with the smallest distance is identified as the current pilot's handling habit type; when using cosine similarity, the type with the highest similarity is identified as the current pilot's handling habit type. The economy type typically corresponds to relatively stable speed adjustments, small energy deviations, and low profile fluctuations; the standard type typically corresponds to handling characteristics close to the historical baseline profile; and the aggressive type typically corresponds to large acceleration fluctuations, significant deceleration gradient changes, or high energy profile deviations.

[0040] When generating the human factor influence correction factor, the baseline correction coefficient corresponding to the identified pilot operating habit type is retrieved from the human factor influence quantification model library. Then, the vector distance between the comprehensive feature vector of the pilot's operating habits for the current flight and the center vector of the corresponding operating habit type is calculated. This vector distance characterizes the degree of deviation of the current flight within that type. Based on the baseline correction coefficient, the vector distance, and the pre-calibrated deviation correction coefficient, the pilot operating habit correction factor can be generated as follows: Pilot operating habit correction factor = Baseline correction coefficient × (1 + Deviation correction coefficient × Vector distance). The deviation correction coefficient can be determined through historical flight operation data, expert calibration data, or model training results. Using this calculation method, when the current flight is closer to the center of its type, the correction factor is mainly determined by the baseline correction coefficient of that type; when the current flight deviates more significantly from the center of its type, the correction factor is adjusted accordingly with the vector distance, thereby achieving a quantitative correction of the degree of influence of human operating habits.

[0041] S6A-2: Extraction of controller command style characteristics; (a) Extraction of track morphology features: Extract the following track morphology parameters from ADS-B track data to reflect controller decision-making: Heading change frequency: Count the number of times the heading changes significantly (more than 10 degrees) during the entire flight, divide by the total flight time to get the heading change frequency per unit time; the higher the frequency, the more frequently the controller intervenes in the lateral guidance of the flight. Altitude change frequency: Count the number of times altitude instructions change (vertical rate sign change or level flight segment occurs) during the entire flight, divide by the total flight time to obtain the altitude change frequency per unit time; the higher the frequency, the more frequently the controller intervenes in the vertical profile of the flight. Average length of level flight segment: During climb or descent, the duration of level flight segments (vertical velocity absolute value less than 200 feet / minute and lasting more than 30 seconds) is counted, and the average length of all level flight segments is taken. A longer average length of level flight segment indicates that the controller tends to use a step-by-step guidance strategy. A shorter level flight segment or no level flight segment indicates that the controller tends to use a continuous climb / descent guidance strategy. Route deviation: A statistical value calculated as the lateral distance between the actual flight path and the planned (or optimal) route. A large route deviation may reflect heading guidance by air traffic controllers for flow management or spacing adjustments.

[0042] (b) Reverse identification of control orders: This embodiment reverse-engineers implicit control instructions from ADS-B tracks; the method does not rely on the acquisition of control voice data (because control voice data is usually difficult to obtain publicly), but infers the type of instructions issued by the controller by analyzing the change patterns of the track. The specific identification rules are as follows: Heading instruction recognition: When the rate of change of heading is detected to exceed a preset threshold (e.g., 2 degrees / second) and the direction of change is stable and continuous, a heading instruction point is marked. The instruction value is the stable heading value after the change; when a significant change in heading is detected (more than 20 degrees) and points to a waypoint, it is marked as a direct flight instruction; Altitude command recognition: When a change in the vertical rate sign is detected (from positive to negative or from negative to positive) or the vertical rate changes from near zero to a significantly positive / negative value, an altitude command point is marked; the command value is the changed target altitude or climb / descent rate; when a level flight segment lasting more than 30 seconds is detected during climb or descent, it is marked as an altitude hold command. Speed ​​command recognition: When a significant change in ground speed is detected during a non-acceleration / deceleration transition phase, and the direction of the change is unrelated to wind changes, a speed command point is marked. For each identified command point, record the following information: command type, command occurrence time (timestamp), command occurrence location (longitude, latitude, barometric altitude), and changes in flight parameters before and after command execution (change in heading, change in altitude, change in speed).

[0043] (c) Construction of controller command style feature vector: Based on the identified sequence of control instructions, the following command style characteristic indicators were statistically analyzed: Command density: Total number of command points divided by total flight time, expressed as commands per hour; this indicator reflects the overall frequency of controller interventions in flights. Command complexity: The percentage of combined commands (i.e., two or more commands of different types issued consecutively at the same time or within a short period of time) in the total number of commands; this indicator reflects the complexity of the controller's command strategy. Lead time for instructions: For heading change instructions and altitude change instructions, calculate the distance or time difference between the time of instruction occurrence and the waypoint or airspace boundary point; a larger lead time indicates that the controller tends to provide forward guidance, while a smaller lead time indicates that the controller tends to provide reactive guidance. Speed ​​intervention tendency: the proportion of speed-related instructions in total instructions; this indicator reflects the controller's preference for flight speed adjustments; Altitude-related intervention tendency: The proportion of altitude-related instructions in total instructions; this indicator reflects the controller's preference for vertical profile interventions of flights; After standardizing the above indicators, a controller command style feature vector with dimensions of 8 to 12 is constructed.

[0044] (d) Association of controlled sectors with time period information: Based on the geographical location and timestamp of the flight path, query the air traffic control sector database to obtain the control sector information corresponding to the current flight at that location and time. This database contains the spatial boundary definition of each control sector (terminal sector is a polygon centered on the airport, and regional sector is a jurisdictional block with reference to the air route), sector number and type (terminal sector / regional sector). In addition, obtain the on-duty team information for this sector during this time period.

[0045] (e) Generation of controller command style correction factors: First, the historical average command style feature vector of the same sector, time period and type of flight (same aircraft type and same flight direction) is obtained from the quantitative model library of human factors influence, and used as the benchmark command style in this scenario. Secondly, calculate the deviation between the current air traffic controller's command style feature vector and the baseline feature vector. The deviation is calculated by taking the absolute value of the difference between the current feature vector and the baseline feature vector in each feature dimension, multiplying it by the weight coefficient of each feature dimension, summing the results, and then dividing by the modulus of the baseline vector. A deviation greater than 0 indicates that the current air traffic controller's command style deviates from the typical level of the sector. Finally, the deviation is input into a pre-built controller command style and fuel consumption impact mapping model. This model is a nonlinear mapping model trained with historical data. The input is the deviation, and the output is the controller command style correction factor. The physical meaning of this correction factor is: the fuel consumption correction coefficient caused by the controller command style deviating from the baseline level. A correction factor greater than 1 indicates that the command style is biased towards the direction of unfavorable fuel consumption (such as frequent intervention, too many level flight segments), and less than 1 indicates that the command style is biased towards the direction of fuel saving (such as continuous climb / descent, forward guidance).

[0046] S6A-3: Comprehensive calculation of human factor correction factors; The pilot control habit correction factor generated by S6A-1 (denoted as...) ) and the controller command style correction factor generated by S6A-2 (denoted as The weighted fusion is performed to obtain the comprehensive human factor correction factor (denoted as ). ); The weighted fusion method is as follows: ; Among them, the weighting coefficient and The adjustments are made dynamically based on the different phases of flight; the adjustment rules are as follows: Takeoff roll, initial climb, en route climb, and landing roll: These phases are primarily controlled by the pilot, with minimal air traffic control intervention. Take a value of 0.7 to 0.8. Take a value of 0.2 to 0.3; Cruise and stepped climb phases: During these phases, the aircraft is typically in autopilot mode, and its flight path is primarily guided by air traffic controllers; at this time... Take a value of 0.3 to 0.5. Take a value of 0.5 to 0.7.

[0047] Descent and approach phases: During these two phases, pilot maneuvers and air traffic control commands jointly influence the flight profile, exhibiting close interaction. Take a value of 0.5 to 0.6. Take a value of 0.4 to 0.5.

[0048] S7: Multi-level fuel consumption model selection; This embodiment constructs a multi-level fuel consumption model library, which includes three levels of fuel consumption calculation models: basic layer model (based on aircraft performance parameters), standard layer model (based on multivariate regression modeling of QAR historical flight data), and enhancement layer model (based on deep neural network training on large-scale QAR / FDR historical data).

[0049] The model selection module automatically selects the corresponding model level from the multi-level fuel consumption model library based on the model information and performance level identifier obtained by S3.

[0050] S7A: Human factor correction injection; The comprehensive human factor correction factor generated by S6A-3 As a correction term for human factors, it is injected into the fuel consumption calculation model level selected by S7 to form a fuel consumption calculation model with correction for human factors.

[0051] The injection method varies depending on the model level: For the base layer model: a multiplicative correction method is used. The original instantaneous fuel consumption rate output by the base layer model is... (Values ​​obtained from aircraft performance parameter databases and adjusted for weather correction factors), the corrected instantaneous fuel consumption rate is .

[0052] For the standard layer model: the multiplicative correction method is also used. The instantaneous fuel consumption rate originally output by the standard layer model is FF0. reg (Based on values ​​calculated using a multiple regression equation), the corrected instantaneous fuel consumption rate is For the enhancement layer model: feature augmentation is employed; the comprehensive human factor correction factor is incorporated. As an additional input feature to the deep neural network, it is fed into the network's input layer along with the original 20 to 30 input features (altitude, ground speed, vacuum speed, acceleration, vertical velocity, energy altitude, energy distribution factor, meteorological parameters, etc.); the network has already learned these features during the training phase. The mapping relationship between features and fuel consumption rate allows the inference stage to naturally utilize these features for fuel consumption prediction; the advantage of this approach is that it can capture... Non-linear interaction relationships with other features.

[0053] S8: Instantaneous fuel consumption rate calculation; Using the fuel consumption calculation model with human factor correction formed by S7A, and taking the flight profile characteristic parameters, meteorological data and derived flight parameters constructed by S5 and S6 as inputs, the instantaneous fuel consumption rate corresponding to each ADS-B track point is calculated.

[0054] S9: Fuel Consumption Credits and Smoothing Processing; S8 calculates the instantaneous fuel consumption rate at each discrete waypoint. To obtain the total fuel consumption for the entire flight, time integration is required. This embodiment uses a second-order smoothed monotonic interpolation method: first, first-order spline interpolation is performed on the instantaneous fuel consumption rate at each discrete waypoint; then, monotonicity constraints are applied; and finally, second-order smoothing is performed. Based on the processed continuous fuel consumption rate curve, time integration is performed over the time intervals of each flight phase to obtain the fuel consumption for each flight phase; the fuel consumption for each flight phase is summed to obtain the total fuel consumption for the flight.

[0055] S10: Standardized output and emissions accounting; The fuel consumption for each flight phase and the total fuel consumption of the flight calculated by S9 are output in a standardized format according to the requirements of the ICAO CORSIA emissions accounting framework. The output includes flight identification information, time information, phased fuel consumption data, summary data (total fuel consumption, total CO2 emissions), human factors analysis report, and ICAO CORSIA emissions report format data package.

[0056] The human factor analysis report includes the following data items: the pilot's operating habits type and confidence level identified for this flight, the comparison of each dimension of operating characteristic parameters with the benchmark value, the value of the pilot's operating habits correction factor and its contribution percentage in the total fuel consumption correction; the deviation of the controller's command style characteristic vector identified for this flight from the benchmark vector, the comparison of each dimension of command style indicators with the benchmark value, the value of the controller's command style correction factor and its contribution percentage; and the value and phased decomposition value of the comprehensive human factor correction factor.

[0057] Example 2: This embodiment provides an aircraft fuel consumption calculation system based on ADS-B track data. Figure 5 The system's structural block diagram is shown, including the following functional modules: Data acquisition module 101: used to acquire ADS-B track data of the target aircraft from the ADS-B ground station network or commercial data service provider.

[0058] Data quality control module 102: connected to data acquisition module 101, used to perform integrity detection, rationality detection and timing consistency detection on ADS-B track data, and to perform graded repair processing on the detected abnormal data.

[0059] Aircraft type identification module 103: connected to data quality control module 102, used to query the aircraft type information, engine model and performance level identifier of the target aircraft based on the ICAO 24-bit address code or flight call sign in ADS-B data.

[0060] Meteorological data fusion module 104: used to acquire meteorological data that matches the spatiotemporal range of ADS-B track data, and interpolate the meteorological data to each track point location through a spatiotemporal interpolation method.

[0061] Flight parameter derivation module 105: Connected to data quality control module 102 and meteorological data fusion module 104, it is used to calculate derived flight parameters such as vacuum speed, Mach number, energy altitude, energy distribution factor and meteorological correction factor.

[0062] The Mach number is obtained by dividing the vacuum velocity by the local speed of sound calculated based on atmospheric temperature; the specific calculation formula is as follows: , ; in, Mach number, For vacuum scalar quantity, For local sound speed, This is the specific heat ratio of air (usually taken as 1.4 for air). The gas constant is... It is at atmospheric static temperature.

[0063] Flight profile construction module 106: Connected to flight parameter derivation module 105, used to construct multi-dimensional flight profiles and automatically identify flight stages.

[0064] Human Factor Feature Extraction and Quantification Module 111: Connected to the Flight Profile Construction Module 106, it is used to extract pilot operation habit feature parameters and controller command style feature parameters from the multi-dimensional flight profile and the identified flight phases, and generate pilot operation habit correction factors and controller command style correction factors through a pre-built human factor influence quantification model library. Figure 6 The internal structure of the module is shown.

[0065] Model selection module 107: connected to the model identification module 103, used to select the corresponding fuel consumption calculation model level from the multi-level fuel consumption model library 108 according to the model information and performance level identifier.

[0066] Human Factor Correction Fusion Module 112: Connected to the Human Factor Feature Extraction and Quantification Module 111 and the Model Selection Module 107, it is used to inject the generated pilot operation habit correction factor and controller command style correction factor as human factor correction items into the selected fuel consumption calculation model to form a fuel consumption calculation model with human factor correction.

[0067] Multi-level fuel consumption model library 108: Used to store and manage multi-level fuel consumption calculation models.

[0068] Human Factors Impact Quantification Model Library 113: Connected to the Human Factors Feature Extraction and Quantification Module 111, it is used to store and manage pilot operation habit type clustering models, controller command style, fuel consumption impact mapping models, and provides an interface for model updates and maintenance.

[0069] Fuel consumption calculation module 109: connected to flight profile construction module 106 and human factor correction fusion module 112, used to calculate the instantaneous fuel consumption rate corresponding to each track point using a fuel consumption calculation model with human factor correction.

[0070] Fuel consumption integral and output module 110: connected to fuel consumption calculation module 109, used to perform second-order smooth monotonic interpolation processing and integral calculation on instantaneous fuel consumption rate, and output it in a standardized format.

[0071] Example 3: This embodiment provides an aircraft fuel consumption calculation device based on ADS-B flight path data, including a processor and a memory. The processor can be a central processing unit (CPU), graphics processing unit (GPU), tensor processor (TPU), field-programmable gate array (FPGA), or other devices with computing capabilities. The memory includes random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive, or other types of storage media.

[0072] The memory stores a computer program containing program instructions for implementing the aircraft fuel consumption calculation method and system based on ADS-B track data as described in Embodiment 1. When the processor executes the computer program, it performs all the steps of the above method.

[0073] Experimental verification and performance evaluation: In order to verify the technical effect of the present invention, especially the effect of the introduction of the human factor correction mechanism on improving the accuracy of fuel consumption calculation, this embodiment carried out systematic experimental verification work. This experiment aims to verify the technical effects in the following four aspects: (1) The effect of the pilot's operating habit correction factor on the accuracy of fuel consumption calculation; (2) The effect of controller command style correction factor on improving fuel consumption calculation accuracy; (3) The effect of comprehensive human factor correction (considering both pilots and controllers) on improving fuel consumption calculation accuracy; (4) The differentiated performance of human factor correction mechanisms in different flight phases.

[0074] The following indicators are used to evaluate the performance of the fuel consumption calculation method: mean absolute percentage error (MAPE), root mean square error (RMSE), R² coefficient of determination, fuel consumption deviation at each stage, and stability of the human factor correction factor (characterized by the consistency of the human factor correction factor for multiple flights on the same route and with the same aircraft type). Design three sets of control experiments: Experimental Group A (Baseline Scheme): The method described in Example 1 is used, but S6A and S7A are not performed, that is, no human factor correction is added; this scheme represents the fuel consumption calculation level of the prior art; Experimental Group B (Pilot Correction Only): Based on the baseline scheme, only the pilot handling habit correction factor is added, without the controller command style correction factor.

[0075] Experimental Group C (with controller correction only): Based on the baseline scheme, only controller command style correction factor is added, without pilot handling habit correction factor; Experimental Group D (Complete Scheme): Based on the baseline scheme, pilot operation habit correction factors and controller command style correction factors are added simultaneously (i.e., the complete technical scheme of this invention).

[0076] Experimental model: The experiment selected four aircraft types that are widely representative of commercial air transport, and the number of experimental flights for each aircraft type is as follows: Table 1. Number of experimental flights corresponding to the four aircraft types in commercial air transport.

[0077] Human factor annotation data: To train and validate the human factor correction mechanism, additional labeled data was obtained from partner airlines: (1) Pilot control habit type labeling: Based on the pilot historical QAR data analysis report provided by the airline's flight quality monitoring department, the control habit type (economy / standard / aggressive) of the pilots operating the above 3,210 flights was labeled for 1,842 flights. The labeling method is as follows: The airline's flight quality monitoring department comprehensively evaluates the control habit type of each pilot based on the control input parameters such as the throttle position change rate, control stick input amplitude, flap deployment altitude threshold, and landing gear deployment distance threshold in the QAR data, combined with the pilot's fuel-saving performance history. The pilot type of the remaining 1,368 flights was determined by clustering identification using the method of this invention.

[0078] (2) Control sector and command style labeling: Through cooperation with a regional control center and a terminal control center in China, information on the control sectors and duty teams experienced by 1,523 of the above 3,210 flights was obtained; for each sector and duty team combination, the cooperating unit provided historical evaluation data (four levels: excellent, good, average, and needing improvement) on the command fuel efficiency of the team. This evaluation data was derived from the QAR fuel consumption analysis report; accordingly, the control command style type experienced by each flight was labeled (high efficiency fuel saving type / standard type / high fuel consumption type).

[0079] Real-world fuel consumption benchmark data The actual fuel consumption of all experimental flights was obtained from QAR fuel flow record data provided by the partner airlines.

[0080] Experimental results: Pilot control habit correction factor accuracy verification First, the accuracy of the pilot control habit feature parameters extracted from ADS-B flight tracks in this invention for identifying actual pilot control habit types was verified. Using 1,842 manually labeled flights as a test set, the control habit types identified by the method of this invention were compared with the labeled types. The results are as follows: Table 2. Manipulation habit types and annotation types identified by the method of the present invention

[0081] The results above indicate that flight profile features extracted solely from ADS-B track data (without relying on QAR control input parameters) can identify pilot control habits with approximately 90% accuracy. The accuracy of identifying the economic type is slightly higher than that of the aggressive type because the characteristics of the economic type (such as smooth speed changes, continuous climb profiles, and delayed descent deceleration) are more pronounced in the ADS-B track.

[0082] Controller command style correction factor accuracy verification: Using 1,523 flights with controlled sector labels as the test set, the effectiveness of the controller command style correction factor was verified. A correlation analysis was performed between the calculated controller command style correction factor for each flight and the labeled command style type. The results are as follows: Table 3. Analysis Results of Air Traffic Controller Command Style Correction Factors and Labeled Command Style Types

[0083] The results show that the controller command style correction factor generated by the method of this invention has good consistency with the actual evaluation results of the control unit. The average correction factor for the high-efficiency fuel-saving team is about 0.965 (i.e., saving about 3.5% of fuel compared to the standard level), and the average correction factor for the high-fuel-consumption team is about 1.048 (i.e. consuming about 4.8% more fuel compared to the standard level), which is consistent with the historical fuel consumption deviation data.

[0084] Verification of the effect of comprehensive human factor correction: The following is a comparison of the total fuel consumption (MAPE) of the B738 engine under four experimental schemes: A, B, C, and D: Table 4. Comparison of MAPE total fuel consumption under the four experimental schemes A, B, C, and D.

[0085] The results show that adding the pilot correction factor alone can reduce MAPE from 2.80% to 2.45% (accuracy improvement of 12.5%); adding the controller correction factor alone can reduce MAPE to 2.38% (accuracy improvement of 15.0%); and adding both types of human factor correction factors simultaneously can further reduce MAPE to 1.98% (accuracy improvement of 29.3%). This indicates that both pilot operating habits and controller command styles have significant and independent effects on fuel consumption, and their combined effect can explain about 30% of the original calculation error.

[0086] Verification of the effectiveness of phased human factor correction: Further analysis of the accuracy improvement effect of human factor correction at different flight stages (taking the B738 model as an example): Table 5. Accuracy improvement effect of human factor correction at different flight stages

[0087] The results show that human factor correction has the most significant effect on accuracy improvement during the descent / approach phase (MAPE reduction of 35.7%). This is because pilots have greater decision-making freedom in speed management, descent profile control, and configuration management during the descent and approach phases, with significant individual differences. At the same time, controller guidance strategies (such as radar guidance and speed control commands) also show significant differences during this phase. The MAPE reduction during the cruise phase is 30.9%, and this improvement is mainly attributed to controller command style correction (controller commands have a more significant impact on track morphology during the cruise phase).

[0088] Stability verification of the human factor correction factor: To verify the stability of the human factor correction factor, 120 flights operated by B738 aircraft on the same route (location A-B, approximately 1,200 km), during the same time period (weekday morning rush hour), and under similar weather conditions were selected. These flights were operated by different pilots and experienced different air traffic control crews. The distribution of the human factor correction factor for these 120 flights was statistically analyzed. Table 6. Distribution of Human Factor Correction Factors for 120 Flights

[0089] Under the same objective conditions, the standard deviation of the human factor correction factor is approximately 0.03-0.04, corresponding to a fuel consumption variation of approximately 3%-4%. This is basically consistent with the range of fuel consumption variation caused by human factors reported in previous studies (4%-6%). This result shows that the human factor correction factor of the present invention has good stability and interpretability, and can effectively capture the fuel consumption differences caused by different people and different management teams under the same conditions.

[0090] Cross-model human factor correction effect verification: The complete protocol (Experimental Group D) was applied to four experimental models and compared with the baseline protocol (Experimental Group A). ​​The results are as follows: Table 7 Comparison of accuracy improvement between experimental group D and experimental group A

[0091] The results show that the human factor correction mechanism achieved significant and consistent results on all four aircraft types, with MAPE reduction ranging from 28% to 30%. The baseline error of the wide-body aircraft (B77W-A333) was greater than that of the narrow-body aircraft, but the relative accuracy improvement brought about by the human factor correction was basically the same, indicating that the human factor correction method of the present invention has good versatility for different aircraft types.

[0092] Through the above systematic experimental verification, the following conclusions can be drawn: (1) The method for extracting pilot control habit characteristics based on ADS-B flight track data proposed in this invention can identify the pilot's control habit type (economic / standard / aggressive) with an accuracy of about 90%, and the generated pilot control habit correction factor has a good correlation with the actual fuel consumption deviation. (2) The method proposed in this invention for the first time to identify control instructions based on ADS-B track and quantify controller command style can effectively distinguish the influence of command style on fuel consumption of different control teams (high-efficiency fuel-saving type / standard type / high fuel consumption type), and the generated controller command style correction factor is consistent with the actual evaluation results of the control unit. (3) The introduction of the comprehensive human factor correction mechanism reduced the MAPE of total fuel consumption calculation from 2.80% to 1.98%, improving the accuracy by about 29%; among them, the accuracy improvement was most significant in the descent / approach phase (35.7%), which is consistent with the objective law that human factors have a greater impact in this phase; (4) The human factor correction factor has good stability (standard deviation of about 0.03-0.04) under the same objective conditions, and can effectively explain about 3%-4% of the fuel consumption variation, which is consistent with the conclusions of existing studies; (5) The human factor correction mechanism has achieved significant results in all four common models, and has good versatility and scalability.

[0093] The above experimental data fully verify the technical effect of the present invention, especially the innovative value and practical effect of the human factor correction mechanism, indicating that the present invention can effectively solve the problem of ignoring the influence of human factors in the prior art, and further improve the accuracy of aircraft fuel consumption calculation based on publicly available ADS-B data.

[0094] The aircraft fuel consumption calculation method and system based on ADS-B flight track data provided by this invention have achieved significant technological innovations in multi-level quality detection and repair of ADS-B flight track data, spatiotemporal fusion interpolation of meteorological data, construction of multi-dimensional flight profiles, fine division of flight phases, extraction and quantification of pilot operating habits, identification and modeling of controller command style characteristics, selection and optimization of multi-level fuel consumption models, second-order smooth monotonic interpolation of instantaneous fuel consumption rate, and standardized emission data output. It has outstanding advantages such as high calculation accuracy, strong adaptability, open data sources, good regulatory compliance, and the first systematic inclusion of human factors into the fuel consumption calculation framework, and has good industrial applicability.

[0095] It should be noted that, in the implementation of this invention, all types of data acquired and used, including but not limited to ADS-B track data, meteorological reanalysis data, aircraft basic information data, and historical flight data used for model training, are acquired and used in strict accordance with relevant data security regulations. Among them, ADS-B data is publicly broadcast aircraft data; meteorological data comes from legally publicly released scientific data platforms; the technical solution of this invention aims to improve aviation operation efficiency and environmental protection, and does not involve the unauthorized collection and use of sensitive data such as personal privacy and trade secrets.

[0096] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calculating aircraft fuel consumption based on ADS-B track data, characterized in that, include: S1. Obtain aircraft ADS-B track data; S2. Perform integrity, rationality, and time-series consistency checks on ADS-B data, mark outliers, and perform graded repairs. S3. Obtain aircraft type information and performance level identification based on aircraft identification; S4. Acquire matching meteorological data and interpolate it spatiotemporally to each track point; S5. Determine the aircraft's speed parameters, energy parameters, and environmental correction parameters based on ADS-B data and meteorological data in the current flight state; S6. Construct a multi-dimensional flight profile that includes altitude, speed, acceleration, vertical motion, and energy state profiles to divide the aircraft flight process into stages and determine the operational stages of the aircraft throughout the entire flight process. S7. Select a fuel consumption calculation model from a multi-level fuel consumption model library based on aircraft type information and performance level identifier. The multi-level fuel consumption model library includes a basic layer model, a standard layer model, and an enhanced layer model. The basic layer model is a lookup model established based on an aircraft performance parameter library. The standard layer model is a regression model established based on historical flight data recorder data. The enhanced layer model is a neural network model trained based on historical flight data recorder data. S8. Using flight profile characteristics, meteorological data and derived parameters as input, calculate the instantaneous fuel consumption rate using a modified model; S9. Perform second-order smooth monotonic interpolation on the instantaneous fuel consumption rate to obtain a continuous curve, and integrate to obtain the fuel consumption of each stage and the total fuel consumption. S10. Calculate aircraft emissions based on fuel consumption, fuel type and preset emission factors, and perform field mapping, unit unification, integrity verification and structured output on fuel consumption data and emission data. S6 includes S6A: extracting characteristic parameters of pilot operating habits and controller command style from flight profiles, and generating correction factors through a human factor quantification model; The feature parameters extracted from the pilot's operating habits include: The climb rate fluctuation coefficient, the descent profile curvature change, and the height maintenance deviation are extracted from the vertical motion profile as vertical profile manipulation characteristic parameters. Acceleration fluctuation coefficient, deceleration gradient fluctuation coefficient, cruise speed fluctuation coefficient, and approach speed deviation are extracted from the velocity profile and acceleration profile as speed management characteristic parameters; Energy conversion deviation, energy profile deviation, and energy distribution deviation are extracted from the energy state profile as energy management characteristic parameters. The above feature parameters are standardized to construct a comprehensive feature vector of pilot operating habits; The pilot's operating habit feature vector is matched with the feature vector of the same type of flight in the historical database to identify the type of operating habit of the pilot currently operating the flight. The type of operating habit includes one of the following: economic, standard and aggressive. Based on the identified type of pilot maneuvering, the baseline correction coefficient corresponding to that type is queried from the pre-built quantitative model library of human factors, and the vector distance between the comprehensive feature vector of the pilot maneuvering habits of the flight and the center vector of the corresponding maneuvering habit type is calculated. A pilot control habit correction factor is generated based on the reference correction coefficient, the vector distance, and the pre-calibrated deviation correction coefficient; S7 includes S7A: injecting human factor correction factors into the selected model to form a corrected model, specifically: The pilot's operating habits correction factor and the controller's command style correction factor are weighted and fused together, and the weighting is dynamically adjusted according to the flight phase. During the takeoff roll, initial climb, en route climb, and landing roll phases, the weight of the pilot's handling habits correction factor is greater than the weight of the controller's command style correction factor. During cruise and step climb phases, the weight of the controller's command style correction factor is greater than the weight of the pilot's handling habits correction factor. During descent and approach, the pilot's handling habits correction factor and the controller's command style correction factor have equal weights. The weighted fusion yields a comprehensive human factor correction factor; For the base layer model and the standard layer model, a multiplicative correction method is adopted, which multiplies the comprehensive human factor correction factor with the instantaneous fuel consumption rate output by the model to obtain the corrected instantaneous fuel consumption rate. For the enhancement layer model, the comprehensive human factor correction factor is used as an additional input feature of the neural network model to participate in the forward propagation calculation.

2. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 1, characterized in that, The climb rate fluctuation coefficient is the ratio of the standard deviation of the vertical rate sequence during the climb phase to the average absolute value of the vertical rate sequence. The change in curvature of the descent profile is the average of the absolute values ​​of the second difference of the height sequence with respect to time during the descent phase; The altitude maintenance deviation is the average of the absolute values ​​of the differences between the barometric altitude of each track point and the reference altitude of that level flight phase, where the reference altitude is the median of the barometric altitude during that level flight phase. Wherein, the acceleration fluctuation coefficient is the ratio of the standard deviation of the acceleration sequence within the corresponding flight phase to the average absolute value of the acceleration sequence; The deceleration gradient fluctuation coefficient is the ratio of the standard deviation of the velocity change rate sequence during the deceleration phase to the average absolute value of the velocity change rate sequence. The cruise speed fluctuation coefficient is the ratio of the standard deviation of the vacuum speed sequence to the average value of the vacuum speed sequence during the cruise phase. The approach speed deviation is the average of the absolute values ​​of the differences between the vacuum speed at each track point and the approach speed of the corresponding aircraft type target during the approach phase; The energy conversion deviation is the average of the absolute values ​​of the differences between the current flight's energy altitude change per unit time and the historical benchmark energy altitude change per unit time for the same aircraft type during the same period. The energy profile deviation is the average of the absolute values ​​of the differences between the current flight energy altitude sequence and the historical benchmark energy altitude sequence of the same aircraft type at the same stage. The energy allocation deviation is the average of the absolute values ​​of the differences between the current flight energy allocation factor sequence and the historical benchmark energy allocation factor sequence of the same aircraft type and period.

3. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 1, characterized in that, The extraction of controller command style characteristic parameters as described in S6A includes: The heading change frequency, altitude change frequency, average length of level flight segment, and route offset are extracted from ADS-B track data as track morphology characteristic parameters. Based on the time series data of ADS-B tracks, implicit control instructions are identified through rule matching methods. The types of control instructions include at least heading instructions, altitude instructions, and speed instructions. The type, time, location, and parameter changes before and after execution of each instruction are recorded. Based on the identified control command sequence, the command density, command complexity, command lead time, speed intervention tendency, and high intervention tendency are statistically analyzed as command style characteristic indicators, and after standardization, a command style characteristic vector of controllers is constructed. Based on the geographical location and timestamp of the flight path, obtain the control sector information corresponding to the current flight. The controller's command style feature vector is compared with the average command style feature vector of the same sector, time period, and type of flight in the historical database to calculate the command style deviation degree, and a controller command style correction factor is generated based on the pre-constructed controller command style and fuel consumption impact mapping model.

4. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 3, characterized in that, The method for constructing the multi-level fuel consumption model library in S7 includes: Collect a large amount of real flight data from QAR or FDR of historical flights, extract multi-dimensional flight profile features and real fuel consumption for each flight, and extract pilot control input parameters, control sector information and corresponding control instruction sequences experienced by each flight. Using pilot control input parameters extracted from QAR data as features, an unsupervised clustering algorithm was used to perform cluster analysis on the pilots' historical flight samples, identifying three types of pilot control habits: economic, standard, and aggressive. Construct a database of control sectors, time periods, and fuel consumption benchmarks; analyze the deviation between the controller command style feature vector of each specific flight and the benchmark feature vector of the same sector and time period; establish the mapping relationship between the deviation and the actual fuel consumption deviation; and construct a mapping model of controller command style and fuel consumption impact. As new flight data accumulates, the clustering model of pilot operating habits and the mapping model of controller command style and fuel consumption impact are incrementally updated and optimized.

5. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 4, characterized in that, The waypoints in S1 include at least some of the following fields: timestamp, ICAO 24-bit address code, flight call sign, longitude, latitude, barometric altitude, geometric altitude, ground speed, heading, and vertical rate. The integrity check described in S2 is used to determine whether any of the required fields in each track point are missing. The required fields include timestamp, longitude, latitude, barometric altitude, and ground speed. The rationality detection is used to determine whether the values ​​of each field are within a preset physical reasonable range; The temporal consistency detection is used to determine whether the rate of change of parameters between adjacent waypoints exceeds a preset allowable change threshold. The hierarchical repair includes: for a single isolated abnormal track point, a sliding window mean filter or median filter is used to replace the abnormal point data with the statistical value of the normal data within the window; For data missing segments consisting of multiple consecutive abnormal track points, linear interpolation or cubic spline interpolation is used to complete the data when the length of the missing segment is less than a preset threshold. When the length of the missing segment is greater than or equal to the preset threshold, a data quality warning mark is added to the track segment.

6. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 5, characterized in that, In S6: The altitude profile is composed of a sequence of changes in air pressure altitude over time; The velocity profile includes the ground velocity profile and the vacuum velocity profile, which are composed of time-varying sequences of ground velocity and vacuum velocity, respectively. The acceleration profile is composed of a sequence of the rate of change of ground velocity with respect to time; The vertical motion profile consists of a sequence of vertical velocity changes over time. The energy state profile consists of a sequence of energy height changes over time; The operational phases of an aircraft during its flight include takeoff roll, initial climb, en route climb, cruise, step climb, descent, approach, and landing roll. The identification methods for operational phases include: distinguishing the climb phase, level flight phase, and descent phase based on the sign and magnitude of the vertical rate; within the climb phase, distinguishing the initial climb phase, en route climb phase, and step climb phase based on the pressure-altitude range; within the descent phase, distinguishing the descent phase and approach phase based on the pressure-altitude range and the magnitude of the vertical rate; and distinguishing the takeoff roll phase and landing roll phase based on the trends in altitude and ground speed changes.

7. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 6, characterized in that, In S7: The basic layer model is based on the performance parameters in the aircraft performance parameter library, and the standard fuel consumption rate parameters are obtained by looking up tables according to the aircraft type, flight phase, altitude and speed. The standard layer model is constructed by collecting historical flight data from the flight data recorder of the target aircraft and grouping it according to flight phases to establish a multiple regression model. The dependent variable of the multiple regression model is the fuel consumption rate, and the independent variables include altitude, vacuum speed, acceleration, vertical speed, energy distribution factor and weather correction factor. The enhancement layer model is constructed by training historical data from the flight data recorder using a neural network. The input layer of the neural network receives multidimensional flight profile feature parameters, meteorological parameters, and derived flight parameters, while the output layer outputs the instantaneous fuel consumption rate.

8. The aircraft fuel consumption calculation method based on ADS-B track data according to claim 7, characterized in that, The output of S10 includes flight identification information, time information, phased fuel consumption data, summary data, human factor analysis report, and ICAO CORSIA emissions report format data package; The human factor analysis report includes: the pilot's operating habits type and confidence level identified for this flight, the value and contribution ratio of the pilot's operating habits correction factor, the deviation degree of the controller's command style characteristic vector, the value and contribution ratio of the controller's command style correction factor, and the value and phased decomposition value of the comprehensive human factor correction factor.

9. An aircraft fuel consumption calculation system based on ADS-B track data, which is applied to the aircraft fuel consumption calculation method based on ADS-B track data as described in any one of claims 4-8, characterized in that the system... include: The data acquisition module is used to obtain ADS-B track data of the target aircraft from the ADS-B ground station network or data service provider; The data quality control module, connected to the data acquisition module, is used to perform integrity detection, rationality detection, and temporal consistency detection on the ADS-B track data, and to perform graded repair processing on abnormal data. The aircraft type identification module, connected to the data quality control module, is used to obtain the aircraft type information and performance level identifier of the target aircraft from the aircraft basic information database based on the unique aircraft identifier in the ADS-B track data. The meteorological data fusion module is used to acquire meteorological data that matches the spatiotemporal range of ADS-B track data, and to interpolate the meteorological data to each track point location using a spatiotemporal interpolation method. The flight parameter derivation module is connected to the data quality control module and the meteorological data fusion module. It is used to calculate derived flight parameters based on the ADS-B track data and the interpolated meteorological data. The derived flight parameters include at least vacuum speed, Mach number, energy altitude, energy distribution factor and meteorological correction factor. The flight profile construction module, connected to the flight parameter derivation module, is used to construct a multi-dimensional flight profile including altitude profile, velocity profile, acceleration profile, vertical motion profile and energy state profile, and to identify flight phases based on the multi-dimensional flight profile. The flight phases include at least the takeoff roll phase, initial climb phase, en-route climb phase, cruise phase, step climb phase, descent phase, approach phase and landing roll phase. The human factor feature extraction and quantification module is connected to the flight profile construction module. It is used to extract pilot operation habit feature parameters and controller command style feature parameters from the multi-dimensional flight profile and the identified flight phases. It also generates pilot operation habit correction factors and controller command style correction factors through a pre-built human factor influence quantification model library. The model selection module, connected to the model identification module, is used to select the corresponding fuel consumption calculation model level from a pre-built multi-level fuel consumption model library based on the model information and performance level identifier. The human factor correction and fusion module is connected to the human factor feature extraction and quantification module and the model selection module. It is used to inject the generated pilot operation habit correction factor and controller command style correction factor as human factor correction items into the selected fuel consumption calculation model to form a fuel consumption calculation model with human factor correction. The fuel consumption calculation module is connected to the flight profile construction module and the human factor correction fusion module, and is used to calculate the instantaneous fuel consumption rate of each flight path point using the fuel consumption calculation model with human factor correction. The fuel consumption integration and output module, connected to the fuel consumption calculation module, is used to perform second-order smooth monotonic interpolation and integration calculation on the instantaneous fuel consumption rate to obtain the fuel consumption of each flight phase and the total fuel consumption of the flight, and to perform field mapping, unit unification, integrity verification and structured output on the fuel consumption data and emission data.

Images