An aircraft overload prediction method, system, electronic device and storage medium

Abstract

This invention relates to the field of aviation safety monitoring technology, and discloses an aircraft overload prediction method, system, electronic equipment, and storage medium. The method includes: calculating air density based on flight environment parameters using the wet air state equation; calculating the true airspeed based on flight environment parameters and positioning parameters using vector synthesis; calculating the standard engine thrust based on EDR parameters; calculating the actual engine thrust using an environmental correction model; calculating lift and drag using an aerodynamic model; calculating the instantaneous overload value based on the balance relationship between the synthesized total normal force and gravity; predicting future overload sequences using a deep learning model based on an input sequence composed of historical EDR parameters, flight environment parameters, and instantaneous overload values; counting the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculating the overload frequency per unit time, and outputting a turbulence intensity index. This invention can improve the conversion accuracy, prediction efficiency, and reliability of aircraft overload prediction.

Description

Technical Field

[0001] This invention belongs to the field of aviation safety monitoring technology, and in particular relates to an aircraft overload prediction method, system, electronic device and storage medium. Background Technology

[0002] Currently, in the field of aviation safety monitoring, aircraft overload is a core indicator for quantifying the intensity of flight turbulence, and its accurate conversion and prediction are crucial for preventing structural damage and ensuring passenger comfort. Existing technologies mainly have the following limitations:

[0003] Physical model defects:

[0004] Overload calculations rely on standard aerodynamic formulas, but ignore the coupled effect of humidity on air density, and the wind speed correction is simplified, which leads to amplified errors in airspeed and lift calculations in complex environments such as high altitudes and high temperatures.

[0005] Limitations of predictive models:

[0006] Temporal prediction often uses LSTM / RNN models, which are computationally inefficient and difficult to capture long-range dependencies, and lack real-time performance for high-frequency data. At the same time, existing attention mechanisms lack the integration of physical laws and prior knowledge.

[0007] Insufficient security assessment:

[0008] Instantaneous overload values ​​cannot directly reflect the intensity of continuous turbulence, and a quantitative mapping between overload frequency and turbulence level has not been established.

[0009] Therefore, this application provides an aircraft overload prediction method to solve the above-mentioned technical problems. Summary of the Invention

[0010] The purpose of this invention is to provide an aircraft overload prediction method, system, electronic device, and storage medium to solve the technical problem that the existing aircraft overload transition prediction has limitations, resulting in low prediction accuracy and reliability.

[0011] To address the aforementioned technical problems, this invention provides an aircraft overload prediction method, comprising:

[0012] In response to the acquired flight environment parameters, positioning parameters, and EDR parameters, the air density is calculated based on the flight environment parameters using the wet air state equation, and the true airspeed is calculated based on the flight environment parameters and the positioning parameters using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle; the positioning parameters include ground speed and heading angle; and the EDR parameters include engine speed and fuel flow rate.

[0013] The atmospheric temperature is converted into thermodynamic temperature. The standard engine thrust is calculated based on the EDR parameters. The wing geometry parameters are called to obtain the angle of attack in the real-time flight status parameters. The actual engine thrust is calculated through an environmental correction model by combining the air density, thermodynamic temperature and standard engine thrust. The lift and drag are calculated through an aerodynamic model by combining the air density, real airspeed, wing geometry parameters and angle of attack.

[0014] The pitch attitude angle in the real-time flight status parameters is obtained, and the actual engine thrust, lift and drag are decomposed into the normal direction according to the pitch attitude angle. The instantaneous overload value is calculated based on the balance relationship between the total normal resultant force and gravity after synthesis.

[0015] The input sequence is composed of historical EDR parameters, flight environment parameters, and instantaneous overload values. A deep learning model is used to predict future overload sequences.

[0016] Within a set time window, count the number of times the instantaneous overload value exceeds the overload threshold, calculate the overload frequency per unit time, and output the turbulence intensity index.

[0017] In some specific embodiments, in response to the acquired flight environment parameters, positioning parameters, and EDR parameters, the air density is calculated based on the flight environment parameters using the wet air state equation, and the true airspeed is calculated based on the flight environment parameters and the positioning parameters using vector synthesis, further including:

[0018] Atmospheric temperature is converted to Kelvin temperature using a thermodynamic temperature scale to generate thermodynamic temperature.

[0019] Based on atmospheric pressure, relative humidity, and thermodynamic temperature, the partial pressure of dry air and the partial pressure of water vapor are separated by the law of partial pressure of moist air.

[0020] The density components of dry air and water vapor are calculated separately based on the gas constant of dry air and the gas constant of water vapor, and then superimposed to generate the density of moist air.

[0021] By using the trigonometric relationship between heading angle and wind direction angle, the wind speed vector is decomposed into a headwind component parallel to the heading and a crosswind component perpendicular to the heading. The true airspeed is then calculated based on the vector difference between the ground speed and the wind speed vector components.

[0022] In some specific embodiments, the standard engine thrust is calculated based on the EDR parameters; the wing geometry parameters are called to obtain the angle of attack from the real-time flight status parameters; the actual engine thrust is calculated using an environmental correction model, combining the air density, thermodynamic temperature, and standard engine thrust; and lift and drag are calculated using an aerodynamic model, combining the air density, true airspeed, wing geometry parameters, and angle of attack. Further, the calculation includes:

[0023] Based on engine speed and fuel flow, the standard state thrust value is calculated through a pre-calibrated engine thrust mapping relationship;

[0024] Standard environmental parameters are obtained as a benchmark. The ratio of actual air density to standard air density and the ratio of actual thermodynamic temperature to standard thermodynamic temperature are calculated. Combined with the engine temperature influence coefficient, the standard state thrust value is corrected by a nonlinear scaling function to generate the actual engine thrust value.

[0025] The pre-stored wing reference area is called, and the current lift coefficient is calculated using the lift coefficient function based on the real-time angle of attack. The product of air density and the square of the true airspeed is defined as dynamic pressure, which is multiplied by the lift coefficient and the wing reference area to obtain lift. Based on the lift coefficient, the total drag coefficient is calculated using the induced drag model, which is multiplied by the dynamic pressure and the wing reference area to obtain drag.

[0026] It outputs the actual thrust, lift, and drag of the engine.

[0027] In some specific embodiments, the pitch attitude angle in the real-time flight status parameters is obtained, and the actual engine thrust, lift, and drag are decomposed into the normal direction according to the pitch attitude angle. The instantaneous overload value is calculated based on the balance relationship between the synthesized total normal force and gravity, and further includes:

[0028] Based on the pitch angle, thrust and drag are decomposed into axial components parallel to the flight path and normal components perpendicular to the flight path, respectively.

[0029] The positive normal components of lift and thrust perpendicular to the flight path are added together, and the negative normal component of drag is subtracted to generate the total normal resultant force.

[0030] The vector difference between the total normal resultant force and gravity is calculated as the scalar ratio of gravity to the total normal resultant force. This is the instantaneous overload value.

[0031] In some specific embodiments, the input sequence is composed of historical EDR parameters, flight environment parameters, and instantaneous overload values. A deep learning model is used to predict future overload sequences, further including:

[0032] By aligning historical EDR parameters, flight environment parameters, and instantaneous overload values ​​with a unified timestamp, a three-dimensional feature tensor containing time dimension, feature dimension, and parameter dimension is constructed.

[0033] Using the three-dimensional feature tensor as input, local temporal fluctuation features are extracted through a multi-layer causal convolutional network, and a local feature vector is output.

[0034] The local feature vectors are input into an encoder with causal masking, and global physical correlations ranging from seconds to hours are captured through a multi-head attention mechanism;

[0035] In the attention score calculation, the time decay factor and the feature importance factor are integrated to optimize the distribution of time decay weight and feature importance weight;

[0036] The encoder output is mapped to a future overload prediction sequence for consecutive future time steps through a fully connected layer.

[0037] In some specific embodiments, the time decay factor and feature importance factor are fused in the attention score calculation, and the distribution of time decay weights and feature importance weights is optimized, further including:

[0038] The time decay weight is calculated using a negative exponential function based on the time interval between the current moment and historical moments.

[0039] The importance weights of features are initialized based on the physical correlation strength of overload based on the input feature set, wherein the input feature set includes flight environment parameters, EDR parameters and flight state parameters;

[0040] Multiply the time decay weight by the feature importance weight and use it as a bias term in the attention score calculation;

[0041] A piecewise weighted loss function is used to impose a nonlinear penalty on the prediction error of the overload scenario.

[0042] In some specific embodiments, the method further includes counting the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculating the overload frequency per unit time, and outputting a turbulence intensity index.

[0043] A first threshold is set to quantify the degree of deviation in occupant comfort, and a second threshold is set to quantify structural safety risks;

[0044] Within a sliding window of a set time length, detect discrete events in which the instantaneous overload value exceeds the first threshold or the second threshold.

[0045] The number of discrete events exceeding the first and second thresholds per unit time is counted to generate a standardized overload frequency value.

[0046] Based on the preset range where the overload frequency value is located, it is mapped to the corresponding bump intensity level output.

[0047] Based on the same concept, the present invention also provides an aircraft overload prediction system, comprising:

[0048] The multi-source parameter acquisition and standardization processing module is configured to respond to acquired flight environment parameters, positioning parameters, and EDR parameters. Based on the flight environment parameters, it calculates air density using the wet air state equation. Based on the flight environment parameters and the positioning parameters, it calculates true airspeed using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle. The positioning parameters include ground speed and heading angle. The EDR parameters include engine speed and fuel flow rate.

[0049] The core physical quantity conversion module is configured to convert the atmospheric temperature into thermodynamic temperature, calculate the standard engine thrust based on the EDR parameters, call the wing geometry parameters, obtain the angle of attack in the real-time flight status parameters, combine the air density, thermodynamic temperature and engine standard thrust, calculate the actual engine thrust through the environmental correction model, and combine the air density, real airspeed, wing geometry parameters and angle of attack to calculate lift and drag through the aerodynamic model.

[0050] The attitude-based overload calculation module is configured to obtain the pitch attitude angle in the real-time flight status parameters, decompose the actual engine thrust, lift and drag into the normal direction according to the pitch attitude angle, and calculate the instantaneous overload value based on the balance relationship between the synthesized total normal resultant force and gravity.

[0051] The overload timing prediction module is configured to use historical timing parameters, flight environment parameters, and instantaneous overload values ​​as input sequences to predict future overload sequences through a deep learning model.

[0052] The turbulence intensity quantification assessment module is configured to count the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculate the overload frequency per unit time, and output the turbulence intensity index.

[0053] Based on the same concept, the present invention also provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an aircraft overload prediction method.

[0054] Based on the same concept, the present invention also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of an aircraft overload prediction method.

[0055] Compared with existing technologies, its advantages are as follows:

[0056] This invention discloses an aircraft overload prediction method, system, electronic device, and storage medium, which can improve the conversion accuracy, prediction efficiency, and reliability during aircraft overload prediction, and ensure flight safety.

[0057] This technology aims to address the limitations in existing aircraft overload transition prediction, which result in low prediction accuracy and reliability. Attached Figure Description

[0058] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0059] Figure 1 This is a flowchart illustrating some specific embodiments of the aircraft overload prediction method of the present invention;

[0060] Figure 2 This is a schematic diagram of the structure of an aircraft overload prediction system according to some specific embodiments of the present invention;

[0061] Figure 3 This is a schematic diagram of the structure of an electronic device according to some specific embodiments of the present invention;

[0062] In the diagram, 710 is the processor; 720 is the memory; 730 is the input device; and 740 is the output device. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0064] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “said,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0065] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0066] It should be understood that although the terms first, second, third, etc., may be used in the embodiments of this application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, first may also be referred to as second without departing from the scope of the embodiments of this application, and similarly, second may also be referred to as first.

[0067] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0068] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.

[0069] It should be noted that any symbols and / or numbers present in the specification that are not marked in the accompanying drawings are not reference numerals.

[0070] Reference Figure 1 An aircraft overload prediction method, comprising:

[0071] S101, in response to the acquired flight environment parameters, positioning parameters, and EDR parameters, calculate the air density based on the flight environment parameters using the wet air state equation, and calculate the true airspeed based on the flight environment parameters and the positioning parameters using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle; the positioning parameters include ground speed and heading angle; and the EDR parameters include engine speed and fuel flow rate.

[0072] S102, convert the atmospheric temperature into thermodynamic temperature, calculate the standard engine thrust based on the EDR parameters, call the wing geometry parameters, obtain the angle of attack in the real-time flight status parameters, combine the air density, thermodynamic temperature and standard engine thrust, calculate the actual engine thrust through the environmental correction model, and combine the air density, real airspeed, wing geometry parameters and angle of attack, calculate lift and drag through the aerodynamic model.

[0073] S103, obtain the pitch attitude angle in the real-time flight status parameters, decompose the actual engine thrust, lift and drag into the normal direction according to the pitch attitude angle, and calculate the instantaneous overload value based on the balance relationship between the synthesized total normal resultant force and gravity.

[0074] S104, based on historical time series EDR parameters, flight environment parameters and instantaneous overload values, forms an input sequence, and a deep learning model is used to predict future overload sequences;

[0075] S105 counts the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculates the overload frequency per unit time, and outputs the turbulence intensity index.

[0076] Specifically, in this embodiment of the invention, in response to the acquired flight environment parameters, positioning parameters, and EDR parameters, the air density is first calculated based on atmospheric temperature, air pressure, and relative humidity using the wet air state equation. This equation separates the total atmospheric pressure into dry air partial pressure and water vapor partial pressure according to Dalton's law of partial pressures, and calculates the mixed density by combining the dry air gas constant and the water vapor gas constant. Simultaneously, the true airspeed is calculated based on ground speed, heading angle, wind speed vector, and wind direction angle using the vector synthesis principle. Specifically, the wind speed is decomposed into a heading-parallel component (headwind and tailwind) and a heading-perpendicular component (crosswind), and the true airspeed is generated based on the vector difference between the ground speed vector and the wind speed component. After converting the atmospheric temperature to thermodynamic temperature, the standard thrust is calculated based on the engine speed and fuel flow rate in the EDR parameters using a pre-calibrated engine thrust mapping relationship. The actual thrust is then obtained by nonlinear scaling based on the air density ratio, thermodynamic temperature ratio, and temperature influence coefficient. Finally, the pre-stored wing reference area is called, and the lift coefficient is calculated using the lift coefficient function in combination with the real-time angle of attack. The product of air density and the square of the true airspeed is defined as... The dynamic pressure is multiplied by the lift coefficient and the wing reference area to obtain the lift. Based on the lift coefficient, the total drag coefficient is calculated using the induced drag model, and then multiplied by the dynamic pressure and the wing reference area to obtain the drag. Real-time pitch angles are acquired, and the actual thrust is decomposed into normal and axial components using the sine and cosine functions of the attitude angles. The drag is also decomposed into normal and axial components. The positive normal components of lift and thrust are accumulated, and the negative normal component of drag is subtracted to generate the total normal resultant force. The instantaneous overload value is calculated using the vector difference between the total normal resultant force and gravity as a scalar ratio to gravity. The calculation is performed using historical time sequences. The system constructs a three-dimensional feature tensor from EDR parameters, flight environment parameters, and overload values. Local temporal fluctuation features are extracted through causal convolutional layers and then input into a Transformer encoder with causal masking to capture long-range dependencies. The attention mechanism integrates time decay factors and feature importance factors to optimize the weight distribution and outputs the future overload sequence. Within a sliding time window, the system detects the number of times the instantaneous overload value exceeds the occupant comfort threshold or structural safety threshold. The number of times the threshold is exceeded is divided by the window duration to generate the overload frequency, which is mapped to the turbulence intensity level for output.

[0077] For example, taking flight data of a certain flight as an example: Air density calculation: Obtain atmospheric temperature 25℃, air pressure 101325Pa, relative humidity 60%, convert to thermodynamic temperature 298.15K; calculate saturated vapor pressure at 25℃ 3169Pa, water vapor partial pressure = 60% × 3169Pa = 1901.4Pa; dry air partial pressure = 101325Pa - 1901.4Pa = 99423.6Pa; dry air density component = 99423.6Pa / (287.05J / kg·K × 298.15K) = 1.160kg / m³; water vapor density component = 1901.4Pa / (461.495J / kg·K × 298.15K) = 0.014kg / m³; moist air density = 1.160 + 0.014 = 1.174kg / m³. Actual airspeed calculation: Ground speed 220m / s (heading angle 90°), wind speed 10m / s (wind direction angle 0°), headwind component = 10m / s × cos(90° - 0°) = 0m / s, crosswind component = 10m / s × sin(90° - 0°) = 10m / s; Actual airspeed = √[(220m / s - 0m / s)² + (0m / s - 10m / s)²] = 220.5m / s. Thrust correction: EDR parameters: 85% speed, 1200 kg / h fuel flow → mapped standard thrust 120000 N; Actual environment: air pressure 65000 Pa, temperature 300 K; Standard environment: air pressure 101325 Pa, temperature 288.15 K; Density ratio = (65000 Pa / 101325 Pa) × (288.15 K / 300 K) = 0.62; Temperature influence coefficient γ = 0.85; Actual thrust = 120000 N × 0.62 × 0.85 = 63240 N. Lift calculation: Wing area 124m², angle of attack 5°, lift coefficient = 2π×5° / 57.3≈0.55; dynamic pressure = 0.5×1.174kg / m³×(220.5m / s)²≈28500Pa; lift = 28500Pa×124m²×0.55≈1,940,000N. Overload calculation: Pitch angle 8°, thrust normal component = 63240N × cos8° ≈ 62500N, drag normal component = 220,000N × sin8° ≈ 30,600N; total normal resultant force = 1,940,000N + 62500N - 30,600N = 1,971,900N; aircraft mass 80,000kg, gravity = 784,000N; instantaneous overload value = (1,971,900N - 784,000N) / 784,000N ≈ 1.51g.Overload prediction: Input a 30-second historical sequence (300 time points × 21 dimensions), use a causal convolution kernel of size 5, and output local features; the Transformer encoder fuses a time decay factor e^(-0.1×Δt) (weight 0.82 when Δt=2 seconds) and feature weighting factors (angle of attack weight 0.9, humidity weight 0.2), outputting an overload sequence for the next 3 seconds [1.52g, 1.49g, 1.48g]. Frequency statistics: |n were detected within 1 hour. x The number of times 0.3g was applied was 42, and the overload frequency was 42 times / hour, which is mapped to "moderate turbulence".

[0078] In some applications, in response to acquired flight environment parameters, positioning parameters, and EDR parameters, air density is calculated based on the flight environment parameters using the wet air state equation. Based on the flight environment parameters and the positioning parameters, true airspeed is calculated through vector synthesis, including converting atmospheric temperature to Kelvin temperature using a thermodynamic temperature scale to generate thermodynamic temperature; separating dry air partial pressure and water vapor partial pressure based on atmospheric pressure, relative humidity, and thermodynamic temperature using the wet air partial pressure law; calculating the dry air density component based on the dry air gas constant and the water vapor density component based on the water vapor gas constant, and superimposing them to generate wet air density; decomposing the wind speed vector into a headwind component parallel to the heading and a crosswind component perpendicular to the heading using the trigonometric relationship between the heading angle and the wind direction angle; and calculating the true airspeed based on the vector difference between the ground speed and the wind speed vector components.

[0079] Understandably, the process involves converting atmospheric temperature to Kelvin temperature using a thermodynamic temperature scale. Based on atmospheric pressure, relative humidity, and this thermodynamic temperature, the total atmospheric pressure is separated into dry air partial pressure and water vapor partial pressure using the law of partial pressure of moist air. The water vapor partial pressure is determined by the product of relative humidity and the saturated vapor pressure at the current temperature. The dry air density component is calculated using the ratio of the dry air gas constant to the dry air partial pressure and thermodynamic temperature, and the water vapor density component is calculated using the ratio of the water vapor gas constant to the water vapor partial pressure and thermodynamic temperature. These two density components are then superimposed to generate the moist air density. Simultaneously, based on the ground speed and heading angle in the positioning parameters, and the wind speed vector and wind direction angle in the flight environment parameters, the wind speed vector is decomposed into a headwind component parallel to the heading and a crosswind component perpendicular to the heading using the trigonometric relationship between the heading angle and the wind direction angle. The true airspeed is then calculated based on the vector difference between the ground speed vector and the headwind, crosswind, and water vapor components.

[0080] For example, converting an atmospheric temperature of 25℃ to a thermodynamic temperature of 298.15 K, an atmospheric pressure of 101325 Pa, and a relative humidity of 60%, the partial pressure of water vapor is calculated as follows: multiplying 60% by the saturated vapor pressure at 25℃ (3169 Pa) gives 1901.4 Pa. Subtracting 1901.4 Pa from the partial pressure of dry air (101325 Pa) equals 99423.6 Pa. Dividing the dry air density component (99423.6 Pa) by the product of the dry air gas constant (287.05 J / (kg·K)) and the thermodynamic temperature (298.15 K) equals 1.160 kg / m³. Dividing the water vapor density component (1901.4 Pa) by the product of the water vapor gas constant (461.495 J / (kg·K)) and the thermodynamic temperature (298.15 K) equals 0.014. The density of moist air is 1.160 kg / m³ plus 0.014 kg / m³, which equals 1.174 kg / m³. Taking a ground speed of 220 m / s, a heading angle of 90°, a wind speed of 10 m / s, and a wind direction angle of 0° as examples, the cosine of the wind component (cos(90°)) multiplied by 10 m / s and 90° minus 0° equals 0 m / s. The crosswind component (sin(90°)) multiplied by 10 m / s and 90° minus 0° equals 10 m / s. The true airspeed is the square root of the sum of the square of the difference between the ground speed vector (220 m / s eastward) and the wind component (0 m / s), plus the square of the difference between the crosswind component (10 m / s northward), which is √[(220 m / s - 0 m / s)² + (0 m / s - 10 m / s)²], equal to 220.5 m / s.

[0081] In some applications, the standard engine thrust is calculated based on the EDR parameters. The wing geometry parameters are then used to obtain the angle of attack from real-time flight parameters. Combining the air density, thermodynamic temperature, and standard engine thrust, the actual engine thrust is calculated using an environmental correction model. Finally, combining the air density, true airspeed, wing geometry parameters, and angle of attack, lift and drag are calculated using an aerodynamic model. This includes calculating the standard state thrust value based on engine speed and fuel flow rate using a pre-calibrated engine thrust mapping relationship. Standard environmental parameters are used as a benchmark to calculate the actual air density and standard airspeed. The engine's actual thrust value is generated by combining the ratio of air density and the ratio of actual thermodynamic temperature to standard thermodynamic temperature with the engine temperature influence coefficient and correcting the standard state thrust value using a nonlinear scaling function. The pre-stored wing reference area is called, and the current lift coefficient is calculated using the lift coefficient function based on the real-time angle of attack. The product of air density and the square of true airspeed is defined as dynamic pressure, which is multiplied by the lift coefficient and the wing reference area to obtain lift. Based on the lift coefficient, the total drag coefficient is calculated using the induced drag model, which is multiplied by the dynamic pressure and the wing reference area to obtain drag. The engine's actual thrust value, lift, and drag are then output.

[0082] Understandably, the standard state thrust value is calculated based on the engine speed and fuel flow rate in the EDR parameters through a pre-calibrated engine thrust mapping relationship; the wing reference area in the pre-stored aircraft inherent parameters is called; the angle of attack in the real-time flight state parameters is obtained; and the actual engine thrust value is calculated through an environmental correction model by combining air density, thermodynamic temperature, and the standard engine thrust value. The environmental correction model uses standard environmental parameters as a benchmark to calculate the ratio of actual air density to standard air density and the ratio of actual thermodynamic temperature to standard thermodynamic temperature, combined with the engine temperature influence coefficient. The thrust value under standard conditions is corrected using a nonlinear scaling function. Lift and drag are calculated using an aerodynamic model by combining air density, true airspeed, wing reference area, and angle of attack. Lift calculation is based on the real-time angle of attack, where the current lift coefficient is determined using a lift coefficient function. The product of air density and the square of true airspeed is defined as dynamic pressure, which is multiplied by the lift coefficient and wing reference area to generate lift. Drag calculation is based on the lift coefficient, where the total drag coefficient is determined using an induced drag model, which is multiplied by the dynamic pressure and wing reference area to generate drag. The environmentally corrected actual engine thrust, lift, and drag values ​​are output to subsequent processing stages.

[0083] For example, using an engine speed of 85% and a fuel flow rate of 1200 kg / h, a standard thrust of 120000 N is obtained through a pre-calibrated thrust mapping relationship. A wing reference area of ​​124 m² is used to obtain a real-time angle of attack of 5°. The actual ambient air density is 1.174 kg / m³, and the thermodynamic temperature is 300 K. The standard ambient air density is 1.225 kg / m³, and the thermodynamic temperature is 288.15 K. The density ratio is 1.174 / 1.225 = 0.958, the temperature ratio is 288.15 / 300 = 0.9605, and the temperature influence coefficient is 0.85. The standard thrust is corrected using a nonlinear scaling function to be approximately 120000 N × 0.958 × 0.9605 × 0.85 ≈ 93800 N. The lift coefficient is calculated as 2π. The linear function value of the angle of attack in radians is 0.55 for 0.087 rad. The dynamic pressure is calculated as 0.5 × 1.174 kg / m³ × (220.5 m / s)² = 28500 Pa, and the lift is 28500 Pa × 0.55 × 124 m² = 1,940,000 N. The induced drag model calculates the total drag coefficient as the ratio of the zero-lift drag coefficient of 0.02 plus the square of the lift coefficient of 0.3025 to the elliptic integral factor of 0.8, which is 0.3025 / (π × 0.8) ≈ 0.12. The total drag coefficient is 0.14, and the drag is 28500 Pa × 0.14 × 124 m² ≈ 494,000 N. The actual output thrust is 93800 N, the lift is 1,940,000 N, and the drag is 494,000 N.

[0084] In some applications, the pitch attitude angle in real-time flight status parameters is obtained, and the actual engine thrust, lift, and drag are decomposed into normal directions according to the pitch attitude angle. The instantaneous overload value is calculated based on the balance relationship between the synthesized total normal resultant force and gravity. This includes decomposing the thrust and drag into axial components parallel to the flight path and normal components perpendicular to the flight path based on the pitch attitude angle; accumulating the normal positive components of lift and thrust perpendicular to the flight path, subtracting the normal negative component of drag, and generating the total normal resultant force; and calculating the instantaneous overload value as the scalar ratio of the vector difference between the total normal resultant force and gravity to gravity.

[0085] Understandably, the pitch attitude angle in the real-time flight status parameters is obtained. Based on this pitch attitude angle, the actual thrust vector of the engine is decomposed into an axial component parallel to the flight path and a normal positive component perpendicular to the flight path. At the same time, the drag vector is decomposed into an axial component parallel to the flight path and a normal negative component perpendicular to the flight path. The lift vector perpendicular to the flight path and the normal positive component of the thrust are accumulated, and the normal negative component of the drag is subtracted to generate the total normal resultant force. According to the principle of force balance, the vector difference between the total normal resultant force and gravity is calculated as the scalar ratio of gravity to the instantaneous overload value.

[0086] For example, taking a pitch angle of 8°, actual engine thrust of 93,800 N, lift of 1,940,000 N, and drag of 494,000 N as an example: the normal component of thrust is 93,800 N multiplied by the cosine of 8° (cos8°) ≈ 92,800 N, and the axial component of thrust is 93,800 N multiplied by the sine of 8° (sin8°) ≈ 13,000 N; the normal component of drag is 494,000 N multiplied by the sine of 8° (sin8°) ≈ 69,000 N, and the axial component of drag is 494,000 N. Multiplying by the cosine value of 8° (cos8°) ≈ 489,000 N; the total normal resultant force is lift 1,940,000 N plus the positive normal component of thrust 92,800 N minus the negative normal component of drag 69,000 N, equaling 1,963,800 N; the aircraft mass of 80,000 kg corresponds to gravity 784,000 N, and the instantaneous overload value is the difference between the total normal resultant force 1,963,800 N and gravity 784,000 N, 1,179,800 N divided by gravity 784,000 N, equaling 1.50 g.

[0087] In some applications, the input sequence is composed of historical EDR parameters, flight environment parameters, and instantaneous overload values. A deep learning model predicts future overload sequences. This involves aligning the historical EDR parameters, flight environment parameters, and instantaneous overload values ​​with a unified timestamp to construct a three-dimensional feature tensor containing time, feature, and parameter dimensions. Using this three-dimensional feature tensor as input, a multi-layer causal convolutional network extracts local temporal fluctuation features, outputting a local feature vector. This local feature vector is then input to an encoder with a causal mask, capturing global physical correlations from seconds to hours through a multi-head attention mechanism. The attention score calculation integrates a time decay factor and a feature importance factor, optimizing the distribution of time decay weights and feature importance weights. Finally, a fully connected layer maps the encoder output as a future overload prediction sequence for consecutive future moments.

[0088] Understandably, based on historical EDR parameters, flight environment parameters, and instantaneous overload values ​​aligned with a unified timestamp, a three-dimensional feature tensor containing time, feature, and parameter dimensions is constructed. Using this three-dimensional feature tensor as input, a multi-layer causal convolutional network extracts local temporal fluctuation features under historical time step constraints, outputting a dimensionality-reduced local feature vector. This local feature vector is then input to an encoder with a causal mask, which simultaneously captures the global physical correlation between second-level turbulent fluctuations and hour-level engine performance degradation through a multi-head attention mechanism. During the attention score calculation, a time decay factor based on negative exponential decay of time intervals and a feature importance factor based on the strength of feature physical correlations are fused to jointly optimize the weight decay rate of historical moments and the priority allocation of feature dimensions. Finally, a fully connected layer maps the high-dimensional features output by the encoder to a sequence of overload prediction values ​​for consecutive future moments.

[0089] For example, a 30-second historical sequence is constructed using a 10Hz sampling rate to form a 3D feature tensor with 300 time steps × 21 features × 1 parameter. This tensor is then input into a 5-layer causal convolutional network (kernel size 5) to extract local features and output a 300×64-dimensional feature vector. This vector is then input into a 3-layer Transformer encoder (8-head attention, with causal masking limiting visibility to only historical data). When calculating the attention at time t=300, a time decay factor e^(-0.1*2)=0.82 is applied to the data at time t=298. Angle of attack features are assigned a feature importance factor of 0.9, while humidity features are assigned 0.2. After fusion, the attention weights are adjusted, and the final fully connected layer outputs the overload prediction sequence for the next 3 seconds and 30 time steps as 1.52g, 1.49g, and 1.48g.

[0090] In some applications, the attention score calculation integrates the time decay factor and the feature importance factor, optimizing the distribution of the time decay weight and the feature importance weight. This includes calculating the time decay weight using a negative exponential function based on the time interval between the current moment and historical moments; initializing the feature importance weight based on the physical correlation strength of overload based on the input feature set, wherein the input feature set includes flight environment parameters, EDR parameters, and flight state parameters; multiplying the time decay weight and the feature importance weight as a bias term for the attention score calculation; and applying a piecewise weighted loss function to impose a nonlinear penalty on the prediction error of the overload scenario.

[0091] Understandably, the attention score calculation incorporates a time decay factor and a feature importance factor to optimize the weight distribution. This includes calculating the time decay weight using a negative exponential function based on the time interval between the current and historical moments, where the weight decays exponentially with the duration of the historical moment; initializing the feature importance weight based on the physical correlation strength of the overload based on the input feature set, which includes flight environment parameters, EDR parameters, and flight state parameters, where the physical correlation strength is determined by prior knowledge of aircraft dynamics; multiplying the time decay weight and the feature importance weight to generate a joint correction factor, which serves as an additive bias term in the attention score calculation; and employing a piecewise weighted loss function to apply a non-linearly increasing penalty weight to the prediction error of high overload dangerous scenarios, where the penalty weight increases with the magnitude of the overload value exceeding the safety threshold.

[0092] For example, taking the attention calculation at historical time t-2 (time interval Δt = 2 seconds) as an example: the time decay weight is calculated to be 0.82 using the negative exponential function e^(-0.1×2); the input feature set has 21 dimensions (12-dimensional EDR parameters such as rotational speed, 8-dimensional flight environment parameters such as wind speed, and 1-dimensional flight state parameters such as angle of attack), and the importance weight of the angle of attack feature is initialized to 0.9 and the importance weight of the humidity feature is 0.2 based on prior knowledge; the joint correction factors are 0.82×0.9=0.738 (angle of attack) and 0.82×0.2=0.164 (humidity), which are added to the original attention score as bias terms; when the predicted overload value is 1.52g, the actual value is 1.60g, because |n x The 1.5g triggers a dangerous scenario penalty mechanism. The loss function uses piecewise weighted Huber loss. The base error of 0.08g is within the range of δ=0.2g but belongs to a dangerous scenario. Therefore, a penalty weight of ω2=5 is applied, and the loss value is 5×0.5×0.08²=0.016.

[0093] In some applications, the number of times the instantaneous overload value exceeds the overload threshold within a set time window is counted, the overload frequency per unit time is calculated, and a turbulence intensity index is output. This includes: setting a first threshold to quantify the degree of deviation in occupant comfort, and setting a second threshold to quantify structural safety risks; detecting discrete events where the instantaneous overload value exceeds the first or second threshold within a sliding window of a set time length; counting the number of discrete events that exceed the first and second thresholds per unit time to generate a standardized overload frequency value; and mapping the overload frequency value to a corresponding turbulence intensity level output based on the preset interval in which it falls.

[0094] Understandably, a first overload threshold is set to quantify the degree of deviation in occupant comfort, and a second overload threshold is set to quantify structural safety risks. Within a sliding time window of fixed length, discrete events where the instantaneous overload value exceeds the first or second threshold are detected in real time. The total number of occurrences of all discrete events exceeding the threshold within a unit of time is counted, and the total number is divided by the sliding window duration to generate a standardized overload frequency value. Based on a preset overload frequency interval mapping relationship, the calculated overload frequency value is converted into the corresponding turbulence intensity level output.

[0095] For example, a first threshold of ±0.3 times gravitational acceleration is set for occupant comfort monitoring, and a second threshold of 1.2 times gravitational acceleration is set for structural safety monitoring. The sliding time window length is 1 hour. Within this window, 20 events exceeding +0.3 times gravitational acceleration, 15 events exceeding -0.3 times gravitational acceleration, and 7 events exceeding 1.2 times gravitational acceleration are detected, for a total of 42 discrete events exceeding the threshold. The overload frequency value is calculated as 42 times divided by 1 hour, which equals 42 times per hour. According to the preset mapping range of 0 to 20 times per hour for mild turbulence, 20 to 40 times per hour for moderate turbulence, and above 40 times per hour for severe turbulence, 42 times per hour is mapped to the severe turbulence level for output.

[0096] Another embodiment of the present invention is described below:

[0097] Aircraft overload (unit: g) is essentially the ratio of the resultant normal force to gravity, and its magnitude is determined by engine thrust, wing aerodynamic forces (lift / drag), and flight attitude. Engine parameters (such as speed and fuel flow) recorded by the EDR (Engine Data Recorder) directly reflect thrust output, while environmental parameters (wind, temperature, and air pressure) indirectly change aerodynamic forces by affecting air density and airspeed. Together, these two constitute the core input for overload calculation.

[0098] This embodiment includes:

[0099] Physical mapping: Using aerodynamic and flight mechanics formulas, EDR parameters and environmental parameters are converted into physical quantities (density, airspeed, thrust, etc.) that can be directly used to calculate overload.

[0100] Data-driven prediction: Utilize deep learning models to capture nonlinear relationships that are difficult to describe by physical formulas (such as sudden turbulence and the gradual effects of engine aging) to achieve dynamic prediction of future overloads.

[0101] Multi-scale assessment: By quantifying the turbulence intensity at different time windows through overload frequency, the instantaneous overload value is transformed into a statistical indicator that can be used for safety decision-making.

[0102] Environmental parameters (temperature, pressure, wind) significantly affect aerodynamic calculations and need to be corrected using physical formulas to represent characteristics under "equivalent standard conditions".

[0103] Air density calculation (wet air state equation):

[0104] ;

[0105] Moist air is a mixture of dry air and water vapor, and it follows Dalton's law of partial pressures (total pressure = partial pressure of dry air + partial pressure of water vapor). Its density is the sum of the densities of the two gases. Air density (kg / m³) is the mass of air per unit volume. For example, the density of air at an altitude of 3000 meters (pressure 70000 Pa, temperature 268 K, humidity 50%) is approximately 0.909 kg / m³, only 74% of that at sea level (1.225 kg / m³), which would result in a significant decrease in lift. Total atmospheric pressure (Pa), standard value at sea level is 101325 Pa, which decreases with increasing altitude. Water vapor partial pressure (Pa) is determined by relative humidity. (e.g., 60%) and saturated vapor pressure (Increases with increasing temperature, approximately 3169 Pa at 25℃) Calculation: . Thermodynamic temperature (K) (t is the temperature in Celsius). As the temperature increases, the air density decreases. The gas constant is given because the molecular weight of water vapor (18.02 g / mol) is less than that of dry air (28.97 g / mol). .

[0106] True airspeed correction (vector composition principle): True airspeed (the speed of the aircraft relative to the air) determines the magnitude of lift, and the influence of wind speed must be subtracted from the ground speed.

[0107] ;

[0108] True airspeed is the vector difference between ground speed and wind speed (aircraft speed relative to air = aircraft speed relative to ground - air speed relative to ground). True airspeed (m / s). For example, if an aircraft has a ground speed of 220 m / s (heading angle 90° eastward) and encounters a 10 m / s northerly wind (wind direction angle 0°), the crosswind component does not affect the airspeed, so the true airspeed remains 220 m / s. Ground speed (m / s): The speed of the aircraft relative to the ground (measured directly by GPS). Heading angle (rad), nose pointing (0 rad due north), (Due East). Headwind / frontwind and crosswind components (m / s), derived from wind speed and wind direction angle calculate: .

[0109] Overload is calculated by combining engine thrust and aerodynamic force with force balance equations.

[0110] Engine thrust correction (similar theory): The thrust recorded by EDR is a standard state value and needs to be corrected to the thrust under actual flight conditions.

[0111] ;

[0112] Engine thrust is directly proportional to intake air density (density = pressure / temperature × constant), and the temperature influence coefficient of a turbofan engine is... (This can be derived from finite element analysis). Actual thrust (N). For example, the standard thrust of an engine is 120,000 N, and after correction at a high-temperature, high-altitude airport (P=65,000 Pa, T=300 K), it is approximately 72,000 N (a decrease of 40%). The standard thrust (N) recorded by EDR is determined by the rotational speed (e.g., the high-pressure rotor speed). ), fuel flow calculation (the higher the speed, the greater the thrust). Standard atmospheric parameters at sea level. Turbofan engine coefficient. Finite element models can be established for aircraft structures and engine systems to simulate the stress and deformation of aircraft structures under different temperature loads, analyze the impact of temperature changes on the expansion and contraction of engine components on the overall mechanical properties of the aircraft, and thus determine the influence coefficient of temperature on overload.

[0113] Lift and drag calculations (thin wing theory + empirical formulas):

[0114] ;

[0115] ;

[0116] Lift is generated by the pressure difference between the upper and lower surfaces of the wing, and drag includes zero-lift drag. ) and induced drag (proportional to the square of lift). In the formulas for lift and drag Derived from the definition of "dynamic pressure" in fluid mechanics, the overall structure of the formula simplifies complex aerodynamics into the product of a "measurable physical quantity" and a "dimensionless aerodynamic characteristic," facilitating engineering applications. The kinetic energy density of the fluid (here, air) is... ( For density, The velocity is defined as "dynamic pressure". It is the kinetic energy formula ( The natural coefficient after mass densityization in (). Lift, drag (N), and air density airspeed squared Wing area (For example, a Boeing 737 is about 124m²) directly proportional. Lift and drag coefficients (dimensionless). With angle of attack Increases linearly ( After reaching a supercritical angle of attack (approximately 15°), the speed drops sharply (stall). Includes basic resistance ( ) and induced resistance ( The larger the aspect ratio, the greater the Oswald factor. The smaller the value, the better. Oswald factor: a key parameter measuring wing aerodynamic efficiency, used to quantify the magnitude of induced drag when the wing generates lift. A higher value (typically between 0.7 and 0.85, closer to 1 for elliptical airfoils) indicates lower induced drag and better aerodynamic performance. The core principle is explained by the formula... ( The total drag coefficient is... Zero-lift drag coefficient, The lift coefficient, The aspect ratio affects the aircraft's drag characteristics. Lift and resistance Essentially, it is the force exerted by air on an object, the magnitude of which is related to dynamic pressure (reflecting the "kinetic energy intensity" of the air) and the area of ​​interaction between the object and the airflow. (e.g., wing area, reflecting the range of forces), and aerodynamic efficiency (expressed as lift coefficient) determined by factors such as the shape of the object and angle of attack. drag coefficient These two dimensionless quantities are directly related. Therefore, the formula is constructed as: dynamic pressure × area × aerodynamic coefficient.

[0117] Total force composition and overload calculation (normal force balance): The total normal force needs to balance gravity and inertial force, and is ultimately converted into overload.

[0118] ;

[0119] ;

[0120] Total normal force formula The structure involves decomposing each force along a normal direction perpendicular to the flight path and then summing them. Lift It is perpendicular to the airflow direction and directly acts as a normal positive force; drag Along the direction of airflow, due to angle of attack Generate normal reverse component Therefore, it needs to be subtracted; thrust. Along the fuselage axis, the normal positive component is generated due to the angle of attack. Therefore, it needs to be added. The sum of the three forces yields the total normal force. Overload formula The structure of the object, the ratio of the acceleration produced by the resultant normal force to the acceleration due to gravity. Total resultant normal force. Gravity needs to be balanced And provide normal inertial force ( (normal acceleration), that is Overload, expressed as the ratio of inertial force to gravity, is substituted into the equation to quantify how many times the aircraft's weight equals the additional load, facilitating the assessment of structural strength and passenger comfort. Normal resultant force. =gravity ( ) + inertial force ( ), overload ( (where is the normal acceleration), therefore .

[0121] To address the temporal dependencies, causality, and feature heterogeneity of flight data, a causal convolutional Transformer model is designed:

[0122] Model structure:

[0123] Causal convolutional layers: capture local temporal features (such as initial fluctuations in turbulence) by restricting the convolutional kernel to operate only on historical data. Guarantee causality:

[0124] ;

[0125] in (Convolution kernel size), after stacking 3 layers, the receptive field reaches 125 time steps (12.5 seconds), and parallel computing improves efficiency.

[0126] Efficiency Improvement of Causal Convolution: Causal convolutional layers can improve efficiency through parallel computation, the core of which lies in the independence of output computation at different time steps. The output at each time step depends only on its preceding output. The input at each time step has no dependency on the computation window at each time step; for example, calculating... use arrive ,calculate use arrive Both can be performed simultaneously. This allows the outputs at all times to be batched through a single convolution operation, quickly capturing local temporal features such as initial turbulent fluctuations.

[0127] Transformer encoder: Models long-range dependencies (such as the effects of engine aging), ensures temporal accuracy through causal masking (future data weights are 0), and uses 8 attention heads to focus on dependencies at different scales (seconds / hours). Dynamic temporal-feature fusion attention: Improves the self-attention mechanism by introducing a two-factor correction.

[0128] ;

[0129] Time decay factor Simulates physical memory decay (such as the weakening of turbulence effects over time). It is worth learning. This refers to the time difference (in seconds) between the current moment and a historical moment. For example, calculating... When focusing on a moment, refer to Data at any given time ;refer to Data at any given time Feature weighting factor Combining prior knowledge (such as angle of attack being more important than humidity), learnable vectors can be used. Optimize weights. Meaning: Prior feature importance vector (dimension consistent with the number of input features). The more significant the "physical correlation strength" and "influence magnitude" of the parameters and overload, the higher the initial weight. Purpose: Learnable weight vectors (dimension and Consistency) is used to correct prior knowledge during training. Initially, the prior importance of "wind speed" was considered low, but the model found that "sudden changes in wind speed" in turbulent data were a strong signal of a sharp increase in overload. This will automatically increase the corresponding weight, thus boosting the actual weight of the feature; ultimately, the feature weighting factor... It will strengthen the attention weight of "features that have a significant impact on overload" and weaken the interference of irrelevant features (such as cabin temperature).

[0130] The piecewise weighted Huber loss function, designed with asymmetric loss to address the need for "higher costs for prediction errors in dangerous overload scenarios":

[0131] ;

[0132] in (Prediction error) (Amplify the penalties for dangerous scenarios). True overload value (the actual normal overload experienced by the aircraft, measured in gravitational acceleration). ), It is the overload value predicted by the model. This measures the deviation between the predicted and actual values. A positive value indicates that the predicted value is too low (underestimating the danger), while a negative value indicates that the predicted value is too high (over-warning). The larger the absolute value, the more severe the error. Threshold The "dangerous overload judgment standard" extracted from flight safety regulations defines a "dangerous scenario" when the actual overload exceeds 1.5g, requiring the model to make more accurate predictions.

[0133] Training and optimization details: Input: 30-second historical sequence (300 time points, 21-dimensional features: 12-dimensional EDR + 8-dimensional environment + 1-dimensional historical overload).

[0134] During the model training and optimization phase, the input data has a specific format and dimensions. The input is a 30-second historical sequence. Since the data acquisition frequency is assumed to be 10Hz (i.e., 10 data points per second), the 30 seconds contain 300 data points. These data points cover 21 features: 12 dimensions from the EDR (Flight Data Recorder), including key flight state parameters such as aircraft speed, altitude, and attitude angles; 8 dimensions of environmental features, such as temperature, air pressure, humidity, and wind speed in different directions; and 1 dimension of historical overload data, recording the overload conditions experienced by the aircraft in the past. Output: Overload sequence for the next 3 seconds (30 time points). Optimizer: AdamW (initial learning rate 0.001, weight decay to suppress overfitting), early stopping strategy (stopping if the verification loss does not decrease after 5 epochs).

[0135] To further improve model training performance and prevent overfitting, an early stopping strategy is adopted. During training, the model's loss on the validation set is closely monitored. If the validation loss does not decrease for five consecutive epochs (one epoch refers to a complete training run of the entire training dataset), the early stopping mechanism is triggered, halting model training.

[0136] Quantify the number of times the overload exceeds the threshold per unit time to assess the intensity of the turbulence:

[0137] ;

[0138] Overload frequency (times / hour), such as 30 times / hour corresponds to "moderate turbulence". : The number of times the overload exceeds the threshold within the statistical window (comfort threshold ±0.3g, structural safety threshold 1.2g). : Statistical window duration (e.g., 1 hour).

[0139] For the purpose of simplicity, the method steps disclosed in the above embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0140] like Figure 2 As shown, the present invention also provides an aircraft overload prediction system, comprising:

[0141] The multi-source parameter acquisition and standardization processing module 201 is configured to respond to the acquired flight environment parameters, positioning parameters, and EDR parameters, calculate air density based on the flight environment parameters using the wet air state equation, and calculate the true airspeed based on the flight environment parameters and the positioning parameters using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle; the positioning parameters include ground speed and heading angle; and the EDR parameters include engine speed and fuel flow rate.

[0142] The core physical quantity conversion module 202 is configured to convert the atmospheric temperature into thermodynamic temperature, calculate the standard engine thrust based on the EDR parameters, call the wing geometry parameters, obtain the angle of attack in the real-time flight status parameters, combine the air density, thermodynamic temperature and engine standard thrust, calculate the actual engine thrust through the environmental correction model, and combine the air density, real airspeed, wing geometry parameters and angle of attack to calculate lift and drag through the aerodynamic model.

[0143] The attitude-based overload calculation module 203 is configured to obtain the pitch attitude angle in the real-time flight status parameters, decompose the actual engine thrust, lift and drag into the normal direction according to the pitch attitude angle, and calculate the instantaneous overload value based on the balance relationship between the synthesized total normal resultant force and gravity.

[0144] The overload timing prediction module 204 is configured to use historical timing parameters, flight environment parameters and instantaneous overload values ​​as input sequences to predict future overload sequences through a deep learning model.

[0145] The turbulence intensity quantification assessment module 205 is configured to count the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculate the overload frequency per unit time, and output the turbulence intensity index.

[0146] It is worth noting that although only some basic functional modules are disclosed in the embodiments of this invention, it does not mean that the composition of this system is limited to the above-mentioned basic functional modules. On the contrary, what this embodiment intends to express is that, based on the above-mentioned basic functional modules, those skilled in the art can arbitrarily add one or more functional modules in combination with existing technology to form an infinite number of embodiments or technical solutions. That is to say, this system is open rather than closed. The fact that this embodiment only discloses a few basic functional modules should not be considered as the scope of protection of the claims of this invention being limited to the disclosed basic functional modules. At the same time, for the convenience of description, the above device is described separately according to its functions as various units and modules. Of course, in implementing this invention, the functions of each unit and module can be implemented in one or more software and / or hardware.

[0147] like Figure 3 As shown, the present invention also provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an aircraft overload prediction method.

[0148] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. For example... Figure 3 The structure shown in this embodiment of the invention includes an electronic device comprising one or more processors 710 and a memory 720; the processors 710 in this electronic device may be one or more. Figure 3 Taking a processor 710 as an example; memory 720 is used to store one or more programs; the one or more programs are executed by the one or more processors 710, so that the one or more processors 710 implement an aircraft overload prediction method as described in any one of the embodiments of the present invention.

[0149] The electronic device may also include an input device 730 and an output device 740.

[0150] The processor 710, memory 720, input device 730, and output device 740 in this electronic device can be connected via a bus or other means. Figure 3 Taking the example of a connection between China and Israel via a bus.

[0151] The memory 720 in this electronic device serves as a computer-readable storage medium, capable of storing one or more programs. These programs can be software programs, computer-executable programs, or modules, such as the program instructions / modules corresponding to the aircraft overload prediction method provided in this embodiment of the invention. The processor 710 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 720, thereby implementing the aircraft overload prediction method described in the above embodiment.

[0152] The memory 720 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 720 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 720 may further include memory remotely located relative to the processor 710, which can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0153] Input device 730 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the electronic device. Output device 740 may include display devices such as a display screen.

[0154] The present invention also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of an aircraft overload prediction method.

[0155] Specifically, the computer storage medium in this embodiment of the invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be—but is not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting aircraft overload, characterized in that, include: In response to the acquired flight environment parameters, positioning parameters, and EDR parameters, the air density is calculated based on the flight environment parameters using the wet air state equation, and the true airspeed is calculated based on the flight environment parameters and the positioning parameters using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle; the positioning parameters include ground speed and heading angle; and the EDR parameters include engine speed and fuel flow rate. The atmospheric temperature is converted into thermodynamic temperature. The standard engine thrust is calculated based on the EDR parameters. The wing geometry parameters are called to obtain the angle of attack in the real-time flight status parameters. The actual engine thrust is calculated through an environmental correction model by combining the air density, thermodynamic temperature and standard engine thrust. The lift and drag are calculated through an aerodynamic model by combining the air density, real airspeed, wing geometry parameters and angle of attack. The pitch attitude angle in the real-time flight status parameters is obtained, and the actual engine thrust, lift and drag are decomposed into the normal direction according to the pitch attitude angle. The instantaneous overload value is calculated based on the balance relationship between the total normal resultant force and gravity after synthesis. The input sequence is composed of historical EDR parameters, flight environment parameters, and instantaneous overload values. A deep learning model is used to predict future overload sequences. Within a set time window, count the number of times the instantaneous overload value exceeds the overload threshold, calculate the overload frequency per unit time, and output the turbulence intensity index.

2. The aircraft overload prediction method according to claim 1, characterized in that, In response to the acquired flight environment parameters, positioning parameters, and EDR parameters, the air density is calculated based on the flight environment parameters using the wet air state equation, and the true airspeed is calculated based on the flight environment parameters and the positioning parameters using vector synthesis, further including: Atmospheric temperature is converted to Kelvin temperature using a thermodynamic temperature scale to generate thermodynamic temperature. Based on atmospheric pressure, relative humidity, and thermodynamic temperature, the partial pressure of dry air and the partial pressure of water vapor are separated by the law of partial pressure of moist air. The density components of dry air and water vapor are calculated separately based on the gas constant of dry air and the gas constant of water vapor, and then superimposed to generate the density of moist air. By using the trigonometric relationship between heading angle and wind direction angle, the wind speed vector is decomposed into a headwind component parallel to the heading and a crosswind component perpendicular to the heading. The true airspeed is then calculated based on the vector difference between the ground speed and the wind speed vector components.

3. The aircraft overload prediction method according to claim 1, characterized in that, Based on the EDR parameters, the standard engine thrust is calculated. The wing geometry parameters are then used to obtain the angle of attack from the real-time flight parameters. Combining the air density, thermodynamic temperature, and standard engine thrust, the actual engine thrust is calculated using an environmental correction model. Finally, combining the air density, true airspeed, wing geometry parameters, and angle of attack, lift and drag are calculated using an aerodynamic model. Further steps include: Based on engine speed and fuel flow, the standard state thrust value is calculated through a pre-calibrated engine thrust mapping relationship; Standard environmental parameters are obtained as a benchmark. The ratio of actual air density to standard air density and the ratio of actual thermodynamic temperature to standard thermodynamic temperature are calculated. Combined with the engine temperature influence coefficient, the standard state thrust value is corrected by a nonlinear scaling function to generate the actual engine thrust value. The pre-stored wing reference area is called, and the current lift coefficient is calculated using the lift coefficient function based on the real-time angle of attack. The product of air density and the square of the true airspeed is defined as dynamic pressure, which is multiplied by the lift coefficient and the wing reference area to obtain lift. Based on the lift coefficient, the total drag coefficient is calculated using the induced drag model, which is multiplied by the dynamic pressure and the wing reference area to obtain drag. It outputs the actual thrust, lift, and drag of the engine.

4. The aircraft overload prediction method according to claim 1, characterized in that, The pitch attitude angle is obtained from the real-time flight status parameters. The actual engine thrust, lift, and drag are decomposed into the normal direction according to the pitch attitude angle. The instantaneous overload value is calculated based on the balance relationship between the synthesized total normal force and gravity. This further includes: Based on the pitch angle, thrust and drag are decomposed into axial components parallel to the flight path and normal components perpendicular to the flight path, respectively. The positive normal components of lift and thrust perpendicular to the flight path are added together, and the negative normal component of drag is subtracted to generate the total normal resultant force. The vector difference between the total normal resultant force and gravity is calculated as the scalar ratio of gravity to the total normal resultant force. This is the instantaneous overload value.

5. The aircraft overload prediction method according to claim 1, characterized in that, The input sequence is composed of historical EDR parameters, flight environment parameters, and instantaneous overload values. A deep learning model is used to predict future overload sequences, further including: By aligning historical EDR parameters, flight environment parameters, and instantaneous overload values ​​with a unified timestamp, a three-dimensional feature tensor containing time dimension, feature dimension, and parameter dimension is constructed. Using the three-dimensional feature tensor as input, local temporal fluctuation features are extracted through a multi-layer causal convolutional network, and a local feature vector is output. The local feature vectors are input into an encoder with causal masking, and global physical correlations ranging from seconds to hours are captured through a multi-head attention mechanism; In the attention score calculation, the time decay factor and the feature importance factor are integrated to optimize the distribution of time decay weight and feature importance weight; The encoder output is mapped to a future overload prediction sequence for consecutive future time steps through a fully connected layer.

6. The aircraft overload prediction method according to claim 5, characterized in that, In the attention score calculation, the time decay factor and feature importance factor are integrated to optimize the distribution of time decay weights and feature importance weights, further including: The time decay weight is calculated using a negative exponential function based on the time interval between the current moment and historical moments. The importance weights of features are initialized based on the physical correlation strength of overload based on the input feature set, wherein the input feature set includes flight environment parameters, EDR parameters and flight state parameters; Multiply the time decay weight by the feature importance weight and use it as a bias term in the attention score calculation; A piecewise weighted loss function is used to impose a nonlinear penalty on the prediction error of the overload scenario.

7. The aircraft overload prediction method according to claim 1, characterized in that, Within a set time window, count the number of times the instantaneous overload value exceeds the overload threshold, calculate the overload frequency per unit time, and output the turbulence intensity index, which further includes: A first threshold is set to quantify the degree of deviation in occupant comfort, and a second threshold is set to quantify structural safety risks; Within a sliding window of a set time length, detect discrete events in which the instantaneous overload value exceeds the first threshold or the second threshold. The number of discrete events exceeding the first and second thresholds per unit time is counted to generate a standardized overload frequency value. Based on the preset range where the overload frequency value is located, it is mapped to the corresponding bump intensity level output.

8. An aircraft overload prediction system, characterized in that, include: The multi-source parameter acquisition and standardization processing module is configured to respond to acquired flight environment parameters, positioning parameters, and EDR parameters. Based on the flight environment parameters, it calculates air density using the wet air state equation. Based on the flight environment parameters and the positioning parameters, it calculates true airspeed using vector synthesis. The flight environment parameters include atmospheric temperature, air pressure, relative humidity, wind speed vector, and wind direction angle. The positioning parameters include ground speed and heading angle. The EDR parameters include engine speed and fuel flow rate. The core physical quantity conversion module is configured to convert the atmospheric temperature into thermodynamic temperature, calculate the standard engine thrust based on the EDR parameters, call the wing geometry parameters, obtain the angle of attack in the real-time flight status parameters, combine the air density, thermodynamic temperature and engine standard thrust, calculate the actual engine thrust through the environmental correction model, and combine the air density, real airspeed, wing geometry parameters and angle of attack to calculate lift and drag through the aerodynamic model. The attitude-based overload calculation module is configured to obtain the pitch attitude angle in the real-time flight status parameters, decompose the actual engine thrust, lift and drag into the normal direction according to the pitch attitude angle, and calculate the instantaneous overload value based on the balance relationship between the synthesized total normal resultant force and gravity. The overload timing prediction module is configured to use historical timing parameters, flight environment parameters, and instantaneous overload values ​​as input sequences to predict future overload sequences through a deep learning model. The turbulence intensity quantification assessment module is configured to count the number of times the instantaneous overload value exceeds the overload threshold within a set time window, calculate the overload frequency per unit time, and output the turbulence intensity index.

9. An electronic device, characterized in that, include: The system includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of the method according to any one of claims 1 to 7.

Images