Low-altitude aircraft route real-time meteorological risk early warning method and system

Abstract

The invention discloses a low-altitude aircraft route real-time meteorological risk early warning method and system, and belongs to the technical field of flight risk identification. The method is used for solving the technical problem that the missing report rate of sudden weather phenomena such as wind shear and micro downburst in an existing scheme is high. The three-dimensional wind shear and turbulence intensity is identified through bimodal cooperation, and the limitation of traditional single-parameter identification is broken through; through independent modeling of normal rigidity and tangential damping, the risk characteristics of vertical airflow and horizontal wind shear are decoupled, so that the risk identification precision can be effectively improved; a height self-adaptive threshold value can effectively reduce the false alarm rate of a low-altitude complex terrain scene; according to the nonlinear attenuation control, through a tanh function, a dynamic response that the larger the risk overshoot is, the stronger the control force is is, and the risk energy propagation depth can be effectively reduced; by using the amplification effect of fractional differential on weak disturbance, the detection rate of risk abrupt change can be effectively improved, and missing report of sudden risks is avoided.

Description

Technical Field

[0001] This invention relates to the field of flight risk identification technology, specifically to a method for real-time meteorological risk warning of low-altitude aircraft routes. Background Technology

[0002] Real-time weather risk warning for low-altitude aircraft routes refers to a comprehensive safety assurance system that uses real-time, high-precision weather monitoring and forecasting technology to conduct dynamic risk assessments of the planned or ongoing routes of aircraft such as UAVs and electric vertical take-off and landing aircraft operating in low-altitude airspace below 300 meters or 1000 meters, and to issue timely warnings of potential weather threats.

[0003] When implementing existing technical solutions, the low coverage of low-altitude meteorological monitoring equipment makes it difficult to achieve full coverage of complex terrain areas, resulting in a high rate of missed reports of sudden meteorological phenomena such as wind shear and microbursts. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time meteorological risk warning method for low-altitude aircraft routes, which solves the technical problem of high underreporting rate of sudden meteorological phenomena such as wind shear and microbursts in existing solutions.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for real-time weather risk warning of low-altitude aircraft routes, comprising:

[0007] Based on the real-time meteorological data collected by the sensors on board the aircraft and the preset numerical forecast field, the three-dimensional wind shear vector field and turbulence intensity scalar field are identified.

[0008] A time-varying early warning model for vertical airflow direction and horizontal wind shear direction is constructed, and the risk index corresponding to the assessment of risk propagation intensity is calculated. The risk index is analyzed using a preset risk propagation depth constraint mechanism, and nonlinear decay control is dynamically triggered.

[0009] The identified disturbance field features are preprocessed, and the meteorological time series are enhanced using fractional differential operators. A risk index prediction model is constructed based on the enhanced features to generate a risk mutation estimate for the next M seconds, and a risk mutation early warning is dynamically provided.

[0010] Preferably, the real-time meteorological data collected by the sensors on the aircraft are spatiotemporally aligned with the preset numerical forecast field, and sensor noise is eliminated by Kalman filtering to generate a fused observation vector.

[0011] Preferably, the force-meteorological coupling matrix is ​​calculated based on the current observation vector and the wind shear estimate from the previous moment;

[0012] When updating the gradient-corrected covariance, a meteorological field gradient correction term is introduced to update the covariance matrix.

[0013] Preferably, based on the current coupling matrix and covariance matrix, the three-dimensional wind shear vector field is calculated using the fractional recursive least squares method;

[0014] The scalar field of turbulence intensity is calculated by using the time series derivative of the three-dimensional wind shear vector field and the weighted average within a preset time window.

[0015] Preferably, the normal risk stiffness K in the vertical airflow direction is calculated and obtained. n (t) and the tangential risk damping B in the direction of horizontal wind shear. t (t), the expression involved is:

[0016] K n (t)=K n0 ·exp( μ ·w w (t)); where K n0 The reference stiffness is μ; the vertical wind shear sensitivity coefficient is w. w (t) represents the current vertical wind shear, corresponding to the vertical component of the three-dimensional wind shear vector field Vw(t);

[0017] Among them, B t0 As the reference damping; I t0 The reference turbulence intensity is β; the damping attenuation exponent is I. t (t) represents the scalar field of turbulence intensity.

[0018] Preferably, the normal risk stiffness K n (t) and tangential risk damping B t (t) is coupled into a unified risk field, and a time-varying risk coupling matrix is ​​constructed to achieve a quantitative characterization of asymmetric characteristics;

[0019] When comprehensively assessing the intensity of risk propagation, the maximum eigenvalue of the time-varying risk coupling matrix C(t) is calculated and obtained; the risk index I(t) is obtained by normalizing the maximum eigenvalue.

[0020] Preferably, the preset risk propagation depth constraint mechanism includes a dynamic threshold for the risk index, which is automatically adjusted according to the flight scenario. The relevant expression is:

[0021] Where, δ lim (t) is the dynamic threshold; δ0 is the baseline threshold; h(t) is the real-time flight altitude; H max θ represents the maximum flight altitude; θ is the altitude correction factor.

[0022] If the risk index I(t) > δ lim (t), triggering nonlinear decay control.

[0023] Preferably, the sample entropy under different derivative orders is calculated, and the derivative order that maximizes the sequence complexity is selected and marked as the optimal fractional order;

[0024] Fractional differential operators are used to enhance meteorological time series data, and an enhanced feature set is output.

[0025] Preferably, a risk index prediction model is constructed based on enhanced features to predict risk mutations over M seconds. The risk mutation prediction value predicted by the risk index prediction model is converted into an executable early warning signal, and a risk mutation early warning prompt is dynamically provided.

[0026] A real-time weather risk warning system for low-altitude aircraft routes includes:

[0027] The dual-mode online identification module for meteorological disturbance fields: Based on the real-time meteorological data collected by the sensors on the aircraft and the preset numerical forecast field, it processes and analyzes the data to identify the three-dimensional wind shear vector field and the turbulence intensity scalar field.

[0028] Asymmetric risk propagation dynamic modeling module: Constructs a time-varying early warning model for vertical airflow direction and horizontal wind shear direction, calculates and obtains the risk index corresponding to assess the risk propagation intensity, analyzes the risk index using a preset risk propagation depth constraint mechanism, and dynamically triggers nonlinear decay control;

[0029] Fractional differential risk prediction module: It preprocesses the identified disturbance field features, uses fractional differential operators to enhance the meteorological time series, constructs a risk index prediction model based on the enhanced features, generates a risk mutation estimate for the next M seconds, and dynamically provides risk mutation early warning prompts.

[0030] Compared to existing solutions, the beneficial effects achieved by this invention are:

[0031] This invention overcomes the limitations of traditional single-parameter identification by using dual-modal collaborative identification of three-dimensional wind shear and turbulence intensity, achieving a full-dimensional characterization of meteorological disturbance fields. By correcting the covariance matrix using meteorological field gradients, it can effectively reduce wind shear estimation errors in complex terrain areas such as mountainous regions and urban canyons, and reduce the lag time of turbulence intensity identification, thus effectively improving terrain adaptation accuracy. By introducing fractional recursive least squares, it enhances the ability to suppress sensor noise and numerical forecast field errors, maintaining a stable identification accuracy in low signal-to-noise ratio scenarios.

[0032] This invention decouples the risk characteristics of vertical airflow and horizontal wind shear by independently modeling normal stiffness and tangential damping, which can effectively improve the accuracy of risk identification. Risk parameters and real-time correlation with wind shear and turbulence intensity can effectively shorten the warning lag time under severe convective weather. The height adaptive threshold can effectively reduce the false alarm rate in low-altitude complex terrain scenarios. Nonlinear decay control achieves a dynamic response with stronger control as the risk overshoot increases through the tanh function, which can effectively reduce the depth of risk energy propagation.

[0033] This invention uses fractional differential operators for processing, which can preserve the historical dependence characteristics of meteorological sequences and effectively reduce prediction errors compared to traditional integer differentials. Based on a multi-step prediction model using a fractional LSTM network, it achieves a leap from real-time monitoring to ultra-early warning, effectively extending the reaction time compared to traditional warning systems and reserving sufficient operational windows for evasive control. Through fine division of the time step, it can pinpoint the specific moment of risk mutations. Utilizing the amplification effect of fractional differentials on weak disturbances, it can effectively improve the detection rate of risk mutations and avoid missing sudden risks. Attached Figure Description

[0034] The invention will now be further described with reference to the accompanying drawings.

[0035] Figure 1 This is a flowchart illustrating the steps involved in implementing a real-time meteorological risk warning method for low-altitude aircraft routes according to the present invention.

[0036] Figure 2 This is a block diagram of a real-time meteorological risk warning system for low-altitude aircraft routes according to the present invention. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] Example 1, such as Figure 1 As shown, this invention provides a real-time weather risk warning method for low-altitude aircraft routes, comprising:

[0039] The dual-mode online identification module for meteorological disturbance fields: Based on real-time meteorological data collected by sensors onboard the aircraft and pre-set numerical forecast fields, it processes and analyzes the data to identify the three-dimensional wind shear vector field and turbulence intensity scalar field; the specific steps include:

[0040] The real-time meteorological data collected by the sensors on the aircraft are spatiotemporally aligned with the preset numerical forecast field. Sensor noise is eliminated by Kalman filtering to generate a fused observation vector Z(t), expressed as:

[0041] Z(t) = W s (t)·Z s (t)+W q (t)·Z q Z(t)+v(t); where Z s (t), Z q (t) represents the sensor observation vector and the numerical prediction field vector, respectively, with the sensor observation vector having a dimension of N. s ×1, N s The number of sensors is N; the dimension of the numerical prediction field vector is N. q ×1, N q To predict the field parameters, spatial interpolation is used to map them to the spacecraft's real-time position; W s (t), W q (t) represent the sensor data weight matrix and the numerical prediction field weight matrix, respectively. v(t) is a zero-mean Gaussian white noise vector; t is the time variable;

[0042] It should be explained that the sensors include, but are not limited to, pitot tubes, IMUs, and millimeter-wave radars;

[0043] The spatiotemporal alignment and Kalman filtering implemented are both existing conventional technical solutions, and the specific implementation steps will not be elaborated here;

[0044] Set the initial covariance matrix Γ(0), fractional order α, spatiotemporal forgetting factor λ, and meteorological field gradient. The initial estimate can be assigned based on the aircraft's design parameters and historical flight big data;

[0045] It should be noted that the default value of the spatiotemporal forgetting factor λ for mountainous scenes is 0.85, and the default value of the spatiotemporal forgetting factor λ for plains scenes is 0.95.

[0046] Meteorological field gradient The initial estimate is obtained based on the finite difference calculation of the numerical prediction field;

[0047] Based on the current observation vector Z(t) and the wind shear estimate Vw(t-1) from the previous moment, calculate the force-meteorological coupling matrix Φ(t):

[0048] Where It(t-1) is the turbulence intensity at the previous moment. Let be the partial derivative matrix of wind speed with respect to the observation vector. This is the sensitivity coefficient of turbulence intensity to the observation vector;

[0049] When performing gradient-corrected covariance updates, a meteorological field gradient correction term is introduced. Update the covariance matrix Γ(t):

[0050]

[0051] in, The Kronecker product enhances the sensitivity of identifying regions with abrupt topographic changes by utilizing the spatial gradient information of the numerical prediction field.

[0052] Based on the current coupling matrix Φ(t) and covariance matrix Γ(t), the three-dimensional wind shear vector field Vw(t) is calculated using the fractional recursive least squares method:

[0053] Vw(t)=Vw(t-1)+Γ(t)Φ(t)(Z(t)-Φ(t) T Vw(t-1));

[0054] The scalar field It(t) of turbulence intensity is calculated using the time series derivative of Vw(t) and the weighted average over a preset time window N:

[0055] Where A is the gain coefficient, ranging from 0.8 to 1.2, dynamically adjusted according to the type of aircraft; for example, 1.2 for helicopters and 0.9 for fixed-wing UAVs; w k For exponentially decaying weights,

[0056] k = 0 represents the current time, and k = N-1 represents the start time of the window. Let be the norm of the time derivative of the wind shear.

[0057] In this embodiment of the invention, by using dual-modal collaborative identification of three-dimensional wind shear and turbulence intensity, the limitations of traditional single-parameter identification are broken, and a full-dimensional characterization of meteorological disturbance fields is achieved. By correcting the covariance matrix through meteorological field gradient, the wind shear estimation error in complex terrain areas such as mountainous areas and urban canyons can be effectively reduced, and the lag time of turbulence intensity identification can be reduced, which can effectively improve the terrain adaptation accuracy. By introducing fractional recursive least squares method, the ability to suppress sensor noise and numerical forecast field errors is enhanced, and a stable identification accuracy is maintained in low signal-to-noise ratio scenarios.

[0058] The asymmetric risk propagation dynamic modeling module constructs a time-varying early warning model for vertical airflow direction and horizontal wind shear direction, calculates and obtains the risk index corresponding to the assessment of risk propagation intensity, analyzes the risk index using a preset risk propagation depth constraint mechanism, and dynamically triggers nonlinear attenuation control; the specific steps include:

[0059] Calculate and obtain the normal risk stiffness K in the vertical airflow direction. n (t) and the tangential risk damping B in the direction of horizontal wind shear. t (t), the expression involved is:

[0060] K n (t)=K n0 ·exp(μ·w w (t)); where K n0 is the reference stiffness; μ is the vertical wind shear sensitivity coefficient, with a default value of 0.8; w w (t) represents the current vertical wind shear, corresponding to the vertical component of the three-dimensional wind shear vector field Vw(t);

[0061] Among them, B t0 As the reference damping; I t0 The reference turbulence intensity is β; the damping attenuation exponent is 0.3 by default.

[0062] It needs to be explained that the normal risk stiffness K n (t) is the influence coefficient of vertical airflow disturbance on the normal load of the aircraft, which characterizes the cumulative stiffness of vertical risk;

[0063] Tangential risk damping B t (t) is the energy dissipation coefficient of horizontal wind shear on the attitude adjustment of the aircraft, which characterizes the buffering capacity of horizontal risks;

[0064] Normal risk stiffness K n (t) and tangential risk damping B t (t) is coupled into a unified risk field, and a time-varying risk coupling matrix C(t) is constructed to achieve a quantitative characterization of asymmetric characteristics. The relevant expressions are:

[0065] Wherein, γ is the cross-coupling coefficient, with a default value of 0.15, which limits the mutual interference between vertical and horizontal risks;

[0066] When comprehensively assessing the intensity of risk propagation, the maximum eigenvalue λ of the time-varying risk coupling matrix C(t) is calculated. max (t):

[0067]

[0068] The risk index I(t) is obtained by normalizing the largest eigenvalue:

[0069] Where, λ lim The upper limit of the safety characteristic value is set at 1000, determined based on the structural strength test of the aircraft.

[0070] The preset risk propagation depth constraint mechanism includes a dynamic threshold for the risk index, which is automatically adjusted according to the flight scenario. The relevant expression is:

[0071] Where, δ lim (t) is the dynamic threshold; δ0 is the baseline threshold, with a value of 0.8; h(t) is the real-time flight altitude; H max θ represents the maximum flight altitude; θ is the altitude correction factor, with a value of 0.2.

[0072] If the risk index I(t) > δ lim (t) triggers nonlinear decay control, specifically:

[0073] Vertical direction: deflected by aileron Δδ a ∝K n (t)·U(t), suppressing abrupt changes in normal load;

[0074] Horizontal direction: via rudder damping Δδ r ∝B t (t)·U(t), dissipating tangential risk energy;

[0075] Where U(t) is the control variable, and its calculation expression is:

[0076] Where U0 is the maximum control variable; tanh() is the hyperbolic tangent function.

[0077] It should be explained that the nonlinear response of small overshoot and large overshoot is achieved by using the tanh function, which avoids the overshoot problem of traditional linear control when there is a sudden change in risk. The control quantity transition is smooth, which can effectively reduce the disturbance to the attitude of the aircraft. The maximum control quantity U0 limits the upper limit of the output. Combined with the saturation characteristics of the tanh function, it prevents the actuators such as servos and motors from being damaged due to overload at the hardware level.

[0078] The control quantity U(t) transforms risk quantification indicators into specific control commands, realizing closed-loop management of meteorological risks from monitoring to suppression, and significantly improving the safety of aircraft under complex weather conditions.

[0079] In this embodiment of the invention, by independently modeling the normal stiffness and tangential damping, the risk characteristics of vertical airflow and horizontal wind shear are decoupled, which can effectively improve the accuracy of risk identification; risk parameters and real-time correlation with wind shear and turbulence intensity can effectively shorten the warning lag time under severe convective weather; the height adaptive threshold can effectively reduce the false alarm rate in low-altitude complex terrain scenarios; nonlinear decay control achieves a dynamic response with a larger risk overshoot and stronger control through the tanh function, which can effectively reduce the risk energy propagation depth.

[0080] Fractional Differential Risk Prediction Module: This module preprocesses the identified disturbance field features, enhances the meteorological time series data using fractional differential operators, constructs a risk index prediction model based on the enhanced features, generates a risk mutation prediction value for the next M seconds, and dynamically provides risk mutation early warning alerts. Specific steps include:

[0081] The three-dimensional wind shear sequence and turbulence intensity sequence of the identified disturbance field characteristics are obtained. The three-dimensional wind shear sequence and turbulence intensity sequence are preprocessed, including but not limited to data cleaning and normalization. These are all existing conventional technical solutions, and the specific implementation steps are not described here.

[0082] Extract a sliding window sequence of length L = 200, corresponding to 20 seconds of historical data, and use it as the input for the fractional derivative;

[0083] To determine the optimal differential order α∈(0,1) and maximize the ability to capture the long-term memory of meteorological sequences, α∈{0.1, 0.2, ..., 0.9} with a step size of 0.1.

[0084] Calculate the sample entropy for different derivative orders α, select the derivative order α that maximizes the sequence complexity, and mark it as the optimal fractional order:

[0085] Where, α op t is the optimal fractional order; For the argmax operator; SampEn(D α x(t) is the sample entropy; D α is a fractional differential operator; x(t) is the original feature sequence, which includes three-dimensional wind shear components and turbulence intensity;

[0086] When using fractional differential operators to enhance meteorological time series data, fractional differential calculations are performed:

[0087] Where Γ(1-α) is the gamma function; τ is the integral dummy variable, representing a historical time point; τ is the integration kernel; x′(τ) is the first derivative of the original sequence x(τ) with respect to the historical time τ;

[0088] Convert the calculated numerical values ​​into discrete calculations:

[0089] Where T is the sampling interval, with a default value of 0.1; The coefficients are fractional binomial coefficients;

[0090] When outputting the enhanced feature set, the relevant expression is:

[0091] F(t) = [Dα u w (t),D α v w (t),D α w w (t),D α I t (t)] T ;in,

[0092] u w (t), v w (t), w w (t) represents the three-dimensional vector components of the spatial rate of change of the atmospheric wind field, specifically the wind shear components of the x-axis (vertical axis), y-axis (horizontal axis), and z-axis (vertical axis) of the aircraft coordinate system, which respectively reflect the wind field changes in the front-back, left-right, and vertical directions of the aircraft.

[0093] A risk index prediction model is constructed based on the enhanced feature F(t) to predict the risk index for M seconds; M can be 120, corresponding to 12 time steps, each step being 10 seconds.

[0094] A fractional-order LSTM network is used, with an input layer, hidden layers, and an output layer; specifically:

[0095] Input layer: contains 4-dimensional augmented features F(t) and 5-dimensional historical risk indices I(t-1), I(t-2), ..., I(t-5), for a total input dimension of 9;

[0096] Hidden layers: 2 layers of fractional LSTM units, 32 neurons per layer, with memory gating using fractional gradient updates;

[0097] Output layer: Fully connected layer with linear activation, outputting the estimated risk of mutations in the next 12 steps.

[0098] It needs to be explained that a fractional LSTM network is an experimental and research model that integrates the mathematical concepts of fractional calculus into the classic LSTM architecture. Its core purpose is to utilize the memory properties of fractional operators to capture long-term and complex dependencies in sequence data better than standard LSTMs. It extends the concepts of differentiation and integration to non-integer orders, such as 0.5 and 1.7. Fractional derivatives have memory or heredity, which means that the derivative at the current time step depends not only on the current state but also on the states at all past time steps, which is reflected by an integral with decaying weights.

[0099] The loss function involved in training is:

[0100] Where i is the time step, and each increment of i by 1 represents a 10-second shift in the prediction time.

[0101] By accumulating from i=1 to i=12, the entire prediction interval of 120 seconds is covered, ensuring that error calculations are performed for risk mutations from the short term of 10 seconds to the medium term of 120 seconds.

[0102] Furthermore, the optimizer is the Adam optimizer with a learning rate of 10. -4 The number of iterations is 500 rounds;

[0103] Predicted risk mutation estimates When converted into an executable warning signal, Δt = 10s, ..., 120s;

[0104] If there exists a time Δt that satisfies This will trigger a risk mutation warning.

[0105] In this embodiment of the invention, processing with fractional differential operators can preserve the historical dependence characteristics of meteorological sequences, effectively reducing prediction errors compared to traditional integer differentials. A multi-step prediction model based on a fractional LSTM network enables a leap from real-time monitoring to ultra-early warning, effectively extending reaction time compared to traditional warning systems and reserving sufficient operational windows for evasive control. Fine division of the time step allows for pinpointing the specific moment of risk mutations. Utilizing the amplification effect of fractional differentials on weak disturbances effectively improves the detection rate of risk mutations, avoiding missed reports of sudden risks.

[0106] Example 2, as Figure 2 As shown, a real-time weather risk warning system for low-altitude aircraft routes includes:

[0107] The dual-mode online identification module for meteorological disturbance fields: Based on the real-time meteorological data collected by the sensors on the aircraft and the preset numerical forecast field, it processes and analyzes the data to identify the three-dimensional wind shear vector field and the turbulence intensity scalar field.

[0108] Asymmetric risk propagation dynamic modeling module: Constructs a time-varying early warning model for vertical airflow direction and horizontal wind shear direction, calculates and obtains the risk index corresponding to assess the risk propagation intensity, analyzes the risk index using a preset risk propagation depth constraint mechanism, and dynamically triggers nonlinear decay control;

[0109] Fractional differential risk prediction module: It preprocesses the identified disturbance field features, uses fractional differential operators to enhance the meteorological time series, constructs a risk index prediction model based on the enhanced features, generates a risk mutation estimate for the next M seconds, and dynamically provides risk mutation early warning prompts.

[0110] In the several embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0111] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0112] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0113] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for real-time weather risk warning of low-altitude aircraft routes, characterized in that, include: Based on the real-time meteorological data collected by the sensors on board the aircraft and the preset numerical forecast field, the three-dimensional wind shear vector field and turbulence intensity scalar field are identified. A time-varying early warning model for vertical airflow direction and horizontal wind shear direction is constructed, and the risk index corresponding to the assessment of risk propagation intensity is calculated. The risk index is analyzed using a preset risk propagation depth constraint mechanism, and nonlinear decay control is dynamically triggered. The identified disturbance field features are preprocessed, and the meteorological time series are enhanced using fractional differential operators. A risk index prediction model is constructed based on the enhanced features to generate a risk mutation estimate for the next M seconds, and a risk mutation early warning is dynamically provided.

2. The method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 1, characterized in that, The real-time meteorological data collected by the sensors on the aircraft are spatiotemporally aligned with the preset numerical forecast field. Sensor noise is eliminated by Kalman filtering to generate a fused observation vector.

3. The method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 2, characterized in that, Based on the current observation vector and the wind shear estimate from the previous moment, the force-meteorological coupling matrix is ​​calculated. When updating the gradient-corrected covariance, a meteorological field gradient correction term is introduced to update the covariance matrix.

4. The method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 3, characterized in that, Based on the current coupling matrix and covariance matrix, the three-dimensional wind shear vector field is calculated using the fractional recursive least squares method. The scalar field of turbulence intensity is calculated by using the time series derivative of the three-dimensional wind shear vector field and the weighted average within a preset time window.

5. A method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 4, characterized in that, Calculate and obtain the normal risk stiffness K in the vertical airflow direction. n (t) and the tangential risk damping B in the direction of horizontal wind shear. t (t), the expression involved is: K n (t)=K n0 ·exp(μ·w w (t)); where K n0 The reference stiffness is μ; the vertical wind shear sensitivity coefficient is w. w (t) represents the current vertical wind shear, corresponding to the vertical component of the three-dimensional wind shear vector field Vw(t); Among them, B t0 As the reference damping; I t0 The reference turbulence intensity is β; the damping attenuation exponent is I. t (t) represents the scalar field of turbulence intensity.

6. The method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 5, characterized in that, Normal risk stiffness K n (t) and tangential risk damping B t (t) is coupled into a unified risk field, and a time-varying risk coupling matrix is ​​constructed to achieve a quantitative characterization of asymmetric characteristics; When comprehensively assessing the intensity of risk propagation, the maximum eigenvalue of the time-varying risk coupling matrix C(t) is calculated; The risk index I(t) is obtained by normalizing the largest eigenvalue.

7. A method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 6, characterized in that, The preset risk propagation depth constraint mechanism includes a dynamic threshold for the risk index, which is automatically adjusted according to the flight scenario. The relevant expression is: Where, δ lim (t) is the dynamic threshold; δ0 is the baseline threshold; h(t) is the real-time flight altitude; H max θ represents the maximum flight altitude; θ is the altitude correction factor. If the risk index I(t) > δ lim (t), triggering nonlinear decay control.

8. A method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 7, characterized in that, Calculate the sample entropy under different derivative orders, select the derivative order that maximizes the sequence complexity and mark it as the optimal fractional order; Fractional differential operators are used to enhance meteorological time series data, and an enhanced feature set is output.

9. A method for real-time meteorological risk warning of low-altitude aircraft routes according to claim 8, characterized in that, Based on enhanced features, a risk index prediction model is constructed to predict risk changes over M seconds. The risk change prediction value of the risk index prediction model is transformed into an actionable early warning signal, and a risk change early warning prompt is dynamically provided.

10. A real-time weather risk warning system for low-altitude aircraft routes, characterized in that, include: The dual-mode online identification module for meteorological disturbance fields: Based on the real-time meteorological data collected by the sensors on the aircraft and the preset numerical forecast field, it processes and analyzes the data to identify the three-dimensional wind shear vector field and the turbulence intensity scalar field. Asymmetric risk propagation dynamic modeling module: Constructs a time-varying early warning model for vertical airflow direction and horizontal wind shear direction, calculates and obtains the risk index corresponding to assess the risk propagation intensity, analyzes the risk index using a preset risk propagation depth constraint mechanism, and dynamically triggers nonlinear decay control; Fractional differential risk prediction module: It preprocesses the identified disturbance field features, uses fractional differential operators to enhance the meteorological time series, constructs a risk index prediction model based on the enhanced features, generates a risk mutation estimate for the next M seconds, and dynamically provides risk mutation early warning prompts.

Images