An aircraft state estimation method and device, electronic equipment and storage medium

Abstract

This application provides a method, apparatus, electronic device, and storage medium for aircraft state estimation. The method includes: acquiring flight state data of the aircraft; determining the load parameters corresponding to the observability conditions based on the flight state data; estimating the load deviation to obtain a load deviation estimate; restricting the update of the load state in response to the load deviation estimate satisfying a preset convergence condition, and estimating the wind speed to obtain a wind speed estimate. This method, by determining load observability based on vertical dynamic excitation and performing load deviation estimation and wind speed estimation in stages, and restricting its state update after load convergence, transforms the coupled joint estimation problem into a decoupled sequential estimation problem. This suppresses the mutual absorption phenomenon between wind speed and load in residual interpretation, improving the estimation stability under all operating conditions.

Description

Technical Field

[0001] This application relates to the field of aircraft technology, and more specifically, to an aircraft state estimation method, apparatus, electronic device, and storage medium. Background Technology

[0002] In dynamics-based state estimation of multi-rotor or vertical takeoff and landing aircraft, dynamic measurements of the aircraft (such as acceleration, thrust, and velocity) are typically used to jointly estimate wind speed and load (or load deviation). Since both load changes and wind disturbances are external forces or inertial forces acting on the aircraft, they are coupled in the same measurement equation. That is, load deviation and wind speed terms jointly affect the dynamic residuals through flight maneuvers (such as changes in acceleration and thrust).

[0003] However, the aforementioned joint estimation method has the following drawbacks: the observability of load estimation strongly depends on vertical or thrust-direction maneuvering excitation, while wind speed estimation mainly relies on horizontal velocity and aerodynamic drag effects. When the aircraft is in a weak excitation condition (such as hovering or constant-speed cruise), the load's contribution to the measurement is insufficient, leading to mutual absorption between the load and wind speed in the residual interpretation. That is, the estimator tends to use wind speed or noise to explain the measurement error caused by the load, and vice versa, thus causing drift, jitter, or even divergence in the estimation results. Although existing technologies have proposed various joint filtering algorithms, they lack strategies for stage division and state locking based on observability conditions, making it difficult to achieve stable decoupled estimation of wind speed and load in engineering. Summary of the Invention

[0004] The purpose of this application is to provide an aircraft state estimation method, apparatus, electronic device, and storage medium to solve the above-mentioned problems.

[0005] In a first aspect, embodiments of this application provide an aircraft state estimation method, the method comprising: acquiring flight state data of the aircraft; determining load parameters corresponding to satisfying observable conditions based on the flight state data; wherein the observable conditions are determined based on vertical dynamic excitation in the flight state data; estimating load deviation to obtain a load deviation estimate; and, in response to the load deviation estimate satisfying a preset convergence condition, restricting the updating of the load state and estimating wind speed to obtain a wind speed estimate.

[0006] In the implementation of the above scheme, load observability is determined based on vertical dynamic excitation, and load deviation estimation and wind speed estimation are performed in stages. After load convergence, its state update is restricted, transforming the coupled joint estimation problem into a decoupled sequential estimation problem. This suppresses the mutual absorption phenomenon between wind speed and load in residual interpretation and improves the estimation stability under all operating conditions. On the other hand, using vertical dynamic excitation as the criterion for determining load observability ensures that load estimation is activated only within the window of sufficient maneuver excitation, avoiding the unobservable or weakly observable problem of load estimation under weak excitation conditions such as hovering or constant speed cruise, and reducing the problems caused by insufficient observability. On the one hand, the drift is estimated; on the other hand, by limiting the update rate of the load state or directly fixing the load value after the load deviation estimation meets the convergence condition, the original joint estimation problem is reduced to a low-dimensional estimation problem only for wind speed, and the load deviation and wind speed are decoupled in the dynamic measurement equation to prevent the load estimation from interfering with the wind speed estimation. On the other hand, a two-stage mode switching mechanism is constructed based on observability conditions, convergence conditions and state constraint strategies, which provides clear online estimation logic rules for load and wind speed for the flight control mission manager, solves the problem of lack of engineering feasibility of traditional joint estimation schemes, and realizes long-term stable operation under real flight data.

[0007] In one implementation of the first aspect, limiting the update of the load state includes: fixing the value of the load state to the sum of the nominal mass and the estimated load deviation, and removing the load state from the estimated state vector; or setting the process noise covariance of the load state to zero; or limiting the update amplitude or update rate of the load state by gating the filter gain.

[0008] In the implementation of the above scheme, by providing multiple locking implementation methods, the measurement coupling between load state and wind speed estimation is effectively blocked after the load deviation estimation converges, ensuring that the wind speed estimation stage is not disturbed by load drift. This allows the above aircraft state estimation method to adapt to the differentiated requirements of different flight missions for the degree of load state solidification and adaptive capability. On the other hand, by fixing the load state value to the sum of the nominal mass and the load deviation estimate and removing the load state from the estimated state vector, the original coupled joint estimation problem is reduced to a low-dimensional estimation problem only for wind speed, preventing the load estimation from interfering with the wind speed estimation, while reducing the computational complexity and memory usage of the state estimator. Furthermore, by setting the process noise covariance of the load state to zero or gating the filter gain, the load state is basically locked while allowing it to undergo limited adaptive correction. This takes into account the slow thermal drift or measurement deviation correction requirements that may exist in the load, and avoids the loss of the ability to correct slow parameter drift after the load state is completely solidified.

[0009] In one implementation of the first aspect, the method further includes: monitoring anomaly indicators when estimating the wind speed; wherein the anomaly indicators are used to quantify the degree of mismatch between the estimated load deviation and flight dynamics measurements, or to quantify the degree of change in the load parameters; the degree of mismatch is used to quantify the degree of deviation between the dynamic measurements predicted based on the estimated load deviation and the actual dynamic measurements in the flight status data; and in response to the anomaly indicators meeting preset re-entry conditions and the load parameters meeting the observability conditions, re-executing the load deviation estimation.

[0010] In the implementation of the above scheme, by monitoring abnormal indicators and triggering a re-entry mechanism, the two-stage decoupled estimation strategy of load and wind speed can adapt to load change conditions such as loading and dropping, as well as estimation deviations caused by gusts, thus achieving long-term stable operation of online state estimation instead of failure after a one-time estimation. On the other hand, by quantifying the degree of mismatch between the estimated load deviation value and flight dynamics measurements, or quantifying the degree of change in load parameters, the inconsistency between the actual load and the current locked value can be detected in a timely manner, avoiding residual divergence caused by continuing to use an incorrect fixed load value for wind speed estimation after a sudden change in load. Furthermore, by logically combining the triggering of abnormal indicators with the satisfaction of load observability conditions, the necessity and feasibility of re-estimating the load are ensured, preventing estimation jitter or convergence failure caused by re-estimating the load under weak excitation or unsuitable flight conditions.

[0011] In one implementation of the first aspect, determining the load parameters corresponding to the observability conditions based on the flight state data includes: acquiring the vertical dynamic excitation of the aircraft; calculating an excitation index based on the vertical dynamic excitation; wherein the excitation index is used to quantify the vertical excitation intensity of the aircraft; accumulating the duration for which the excitation index satisfies a preset amplitude condition; and determining that the load parameters satisfy the observability conditions when the duration reaches a first preset duration, and acquiring the load parameters corresponding to the observability conditions.

[0012] In the implementation of the above scheme, the time-domain cumulative judgment mechanism of vertical dynamic excitation ensures that load estimation is activated only within the excitation window where load observability is sufficient, providing a reliable switching trigger condition for the two-stage decoupling strategy and avoiding estimation divergence caused by insufficient observability. On the other hand, using vertical dynamic excitation as the criterion for determining load observability is consistent with the physical characteristics that load deviation mainly affects aircraft dynamics through vertical thrust and acceleration, improving the consistency between observability judgment and the physical essence of load estimation. Furthermore, by using the duration of cumulative excitation index satisfying amplitude conditions rather than instantaneous judgment, false excitation signals caused by instantaneous noise or brief maneuvers are effectively filtered out, avoiding convergence failure or estimation jitter caused by load estimation under non-continuous excitation conditions.

[0013] In one implementation of the first aspect, the preset convergence condition includes: within a second preset time period, the convergence statistical characteristics of the load deviation estimate continuously satisfy a preset statistical characteristic threshold.

[0014] In the implementation of the above scheme, the stability and reliability of the load deviation estimate are determined by a preset convergence condition, providing a clear quantitative criterion for switching from the load estimation stage to the wind speed estimation stage, and preventing subsequent wind speed estimation deviations caused by prematurely locking the load in the non-converged state. On the other hand, by utilizing the method that the convergence statistical characteristics continuously meet the threshold within the second preset time period, the time-domain hysteresis mechanism avoids misjudgments caused by instantaneous noise or estimation fluctuations, thus improving the robustness of the convergence determination. Furthermore, multi-dimensional convergence determination based on convergence statistical characteristics can adapt to the output characteristics of different estimation algorithms, enhancing the decoupling between the convergence criterion and the specific implementation method.

[0015] In one implementation of the first aspect, estimating the load deviation to obtain a load deviation estimate includes: estimating the load deviation based on the vertical residual measurement equation to obtain a load deviation estimate; or, estimating the load deviation using a joint state to obtain a load deviation estimate; or, using the wind speed term in the measurement equation as a measurement noise term and fitting the load deviation estimate using the recursive least squares method.

[0016] In the implementation of the above scheme, by providing multiple load deviation estimation methods such as vertical residual measurement, joint state estimation, or recursive least squares (RLS), the two-stage decoupling strategy can be compatible with different state estimation algorithm frameworks, enhancing the portability and engineering applicability of the above aircraft state estimation method in different flight control systems. On the other hand, by estimating load deviation based on the vertical residual measurement equation, and utilizing the physical characteristic that load mainly affects vertical dynamics, the vertical residual measurement with lower sensitivity to wind speed is selected for priority estimation, reducing the impact of wind speed disturbance on the accuracy of load estimation. Furthermore, when using joint state estimation, the wind speed term in the measurement equation is treated as a measurement noise term for recursive least squares fitting, achieving rapid identification of load deviation while maintaining the simplicity of the algorithm.

[0017] In one implementation of the first aspect, the method further includes: when switching from load deviation estimation to wind speed estimation, or from wind speed estimation to load deviation estimation, determining whether the duration of the current estimation mode reaches a preset minimum dwell time; if the duration of the current estimation mode meets the preset minimum dwell time, then switching the estimation mode.

[0018] In the implementation of the above scheme, by judging whether the duration of the current estimation mode reaches the preset minimum dwell time when switching between load deviation estimation and wind speed estimation, it is ensured that both stages have sufficient stable running time, avoiding frequent mode switching caused by transient fluctuations in conditions, and ensuring the long-term operational stability of the two-stage decoupling strategy. On the other hand, by using the preset minimum dwell time as a switching threshold, the mode jitter phenomenon caused by the fluctuation of excitation conditions or abnormal indicators near the threshold is effectively suppressed, reducing the waste of computing resources and degradation of estimation performance caused by the estimator repeatedly switching between load estimation and wind speed estimation. Furthermore, requiring the duration of the current estimation mode to meet the minimum dwell time before switching is allowed provides the necessary time window for the full convergence of the load deviation estimate or the stable tracking of the wind speed estimate, preventing incomplete estimation or residual divergence caused by premature exit from the current mode.

[0019] In one implementation of the first aspect, the wind speed estimation includes: in response to detecting gust conditions, performing the wind speed estimation after increasing the estimation bandwidth of the wind speed estimation or increasing the process noise covariance of the wind speed estimation.

[0020] In the implementation of the above scheme, by detecting gust conditions and adaptively adjusting the dynamic response parameters of wind speed estimation, the wind speed estimation can automatically switch estimation characteristics under different conditions of steady-state wind and transient gusts, taking into account both steady-state estimation accuracy and dynamic tracking capability, and improving estimation performance under all wind conditions. On the other hand, when gust conditions are detected, the estimation bandwidth is increased or the process noise covariance is increased, which enhances the sensitivity and response speed of the filter to rapid wind changes and avoids the wind speed estimation deviation caused by the response lag of the fixed low-bandwidth estimator during gusts. Furthermore, the estimation dynamic characteristics are only temporarily improved when gust conditions are detected, rather than permanently changing the estimator parameters, which ensures the smoothness and noise suppression capability of wind speed estimation during non-gust periods, and realizes the optimized allocation of estimation bandwidth and computing resources under different wind conditions.

[0021] Secondly, embodiments of this application provide an aircraft state estimation device, comprising: The data acquisition module is used to acquire the flight status data of the aircraft; The load parameter acquisition module is used to determine the load parameters corresponding to the observable conditions based on the flight status data; wherein the observable conditions are determined based on the vertical dynamic excitation in the flight status data. The load estimation module is used to estimate the load deviation to obtain the estimated load deviation value; The wind speed estimation module is used to limit the update of the load state in response to the load deviation estimation value satisfying the preset convergence condition, and to estimate the wind speed to obtain the wind speed estimate value.

[0022] Thirdly, embodiments of this application provide an electronic device, including: a processor, a memory, and a communication bus, wherein the processor and the memory communicate with each other through the communication bus; the memory stores computer program instructions that can be executed by the processor, and the computer program instructions are read and executed by the processor to perform the method provided in the first aspect or any possible implementation of the first aspect.

[0023] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when read and executed by a processor, perform the method provided in the first aspect or any possible implementation thereof.

[0024] Fifthly, embodiments of this application provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the method provided by the first aspect or any possible implementation of the first aspect.

[0025] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims and drawings. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart illustrating the aircraft state estimation method provided in this application embodiment; Figure 2 A flowchart illustrating an aircraft state estimation method in a specific application scenario provided in this application embodiment; Figure 3 This is a schematic diagram of the excitation window detection process provided in an embodiment of this application; Figure 4 This is a schematic diagram of the load limit status update strategy provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the aircraft state estimation device provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0028] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0029] In dynamics-based state estimation of multi-rotor or vertical takeoff and landing aircraft, residual equations are typically constructed using aircraft dynamic measurements to jointly estimate wind speed and load (or load deviation). The measurement relationship can be abstracted as follows:

[0030] in, This refers to the load tolerance. These are excitation terms related to flight maneuvers (such as acceleration, thrust changes, etc.). Represents the resistance mapping matrix; Indicates wind speed; This represents measurement noise and unmodeled terms. The above measurement structure couples the load deviation and wind speed term through the same measurement equation, and both affect the dynamic residual.

[0031] However, the aforementioned measurement structure inherently suffers from coupling absorption problems. Specifically, when the aircraft is in a weakly excited state (such as hovering or constant-speed cruise), the excitation term... The load deviation contributes insufficiently to the measurement, leading to load deviation. The observability decreases. At this point, the estimator tends to attribute the residual error to wind speed. or noise This causes the load deviation estimate to drift toward the wind speed term; conversely, under gust conditions, sudden changes in wind speed may be misinterpreted as load changes, leading to mutual absorption between wind speed and load in the residual interpretation, resulting in systematic bias in the estimation results.

[0032] Furthermore, the observability conditions for load estimation and wind speed estimation dynamically change and conflict with flight maneuvers. Load estimation mainly relies on vertical or thrust-direction maneuvering excitations, while wind speed estimation mainly relies on horizontal velocity and aerodynamic drag effects. When using a single joint filter to simultaneously estimate both parameters under all flight conditions, it is difficult to achieve a stable balance between weakly excited hovering and strongly excited maneuvers, leading to unobservable or weakly observable parameters in some flight phases. The estimation covariance is also prone to ill-conditioned growth due to insufficient excitation. Moreover, existing solutions typically only describe the algorithmic structure of joint estimation, without providing explicit mode switching conditions, convergence criteria, and state locking strategies. The lack of engineering rules on when to estimate load and when to estimate wind speed results in the estimator being prone to estimation drift, parameter jitter, or even numerical divergence under real flight data due to mode switching or insufficient excitation, making it difficult to meet the engineering requirements for long-term stable operation.

[0033] In view of this, embodiments of this application provide an aircraft state estimation method. This method determines the observability of the load based on vertical dynamic excitation and performs load deviation estimation and wind speed estimation in stages. After the load converges, its state update is restricted, transforming the coupled joint estimation problem into a decoupled sequential estimation problem. This suppresses the mutual absorption phenomenon between wind speed and load in the residual interpretation, improving the estimation stability under all operating conditions. On the other hand, by using vertical dynamic excitation as the criterion for determining the observability of the load, it ensures that load estimation is activated only within the window of sufficient maneuver excitation, avoiding the unobservable or weakly observable problem of load estimation under weak excitation conditions such as hovering or constant speed cruise, and reducing the impact of... The problem addresses several issues: First, insufficient observability leads to estimation drift. Second, by limiting the update rate of the load state or directly fixing the load value after the load deviation estimation meets the convergence condition, the original joint estimation problem is reduced to a low-dimensional estimation problem focusing only on wind speed. This decouples the load deviation and wind speed in the dynamic measurement equations, preventing the load estimation from interfering with the wind speed estimation. Third, a mode switching mechanism is constructed based on observability conditions, convergence conditions, and state constraint strategies. This provides the flight control mission manager with clear online estimation logic rules for load and wind speed, solving the problem of the lack of engineering feasibility in traditional joint estimation schemes and achieving long-term stable operation under real flight data.

[0034] Please see Figure 1 The diagram illustrates a flowchart of the aircraft state estimation method provided in this application embodiment. The aircraft state estimation method provided in this application embodiment can be applied to electronic devices, which may include physical devices such as servers, PCs, tablets, or smartphones, or virtual devices such as virtual machines or containers. The electronic device can be a single device, a combination of multiple devices, or a cluster of a large number of devices. The aforementioned aircraft state estimation method may include: Step S110: Obtain the flight status data of the aircraft.

[0035] The aforementioned aircraft include various types of unmanned aerial vehicles (UAVs) that rely on rotors to generate lift or possess vertical takeoff and landing (VTOL) capabilities, including but not limited to multi-rotor UAVs, VTOL fixed-wing UAVs, and tiltrotor UAVs. These aircraft are typically equipped with onboard sensors such as inertial measurement units (IMUs), global navigation satellite system (GNSS) receivers, barometric altimeters, and airspeed indicators (ASIAs) to collect real-time flight status data. Because these aircraft often face the coupled effects of changes in payload (such as cargo delivery and retrieval operations) and complex wind disturbances (such as constant winds, gusts, and wind shear) during operations, their flight control and state estimation systems must identify payload deviations and wind speed parameters online to ensure flight quality and mission accuracy. Payload deviation refers to the deviation of the aircraft's total mass from its nominal mass, while wind speed parameters reflect the horizontal airflow velocity in the aircraft's environment.

[0036] The aforementioned flight status data refers to the set of measured physical quantities used to characterize the current motion state of an aircraft and the effects of the external environment, typically defined in the Earth-Fixed NED (North-East-Down) coordinate system. The following provides the raw measurement input for online estimation of load deviation and wind speed. Flight status data can include the following types of data: (1) navigation speed data, that is, the linear velocity vector of the aircraft's center of mass relative to the ground coordinate system; (2) gravity-free linear acceleration data, that is, the specific force measurement value of the aircraft after removing the influence of gravity acceleration, reflecting the kinematic characteristics under non-conservative external force; (3) thrust ground system component data, that is, the projection components of the total thrust generated by the rotor or power unit in each axis of the ground coordinate system; (4) attitude angle data, used to characterize the angular relationship between the aircraft's body coordinate system and the ground coordinate system, usually including roll angle, pitch angle and yaw angle; (5) aircraft aerodynamic and mass characteristic parameters such as drag parameters and nominal mass, where drag parameters are used to establish the mapping relationship between aerodynamic force and relative airspeed, and nominal mass is used as the benchmark reference value for load estimation.

[0037] The above data can be obtained in the following ways: navigation speed data and attitude angle data can be calculated by fusing the measurements from the Inertial Measurement Unit (IMU) and the Global Navigation Satellite System (GNSS) using extended Kalman filtering or complementary filtering algorithms; degravity acceleration data can be obtained by projecting accelerometer measurements onto the Earth-fixed coordinate system via an attitude transformation matrix and then subtracting the standard gravity acceleration vector; thrust ground-based component data can be obtained by combining the propeller thrust model with the thrust in the body coordinate system based on motor speed or ESC commands, and then projecting it onto the Earth-fixed coordinate system via an attitude transformation matrix; drag parameters and nominal mass are stored as known design parameters or prior calibration values ​​in the non-volatile memory of the flight control system for use by the real-time estimation module.

[0038] Step S120: Based on the flight status data, determine the load parameters corresponding to the observable conditions; wherein, the observable conditions are determined based on the vertical dynamic excitation in the flight status data.

[0039] The observability conditions of the aforementioned load parameters refer to the mathematical conditions under which the load deviation parameters are determined based on the current flight status data to ensure that they have sufficient excitation to be uniquely and accurately identified. Essentially, the observability conditions of load parameters reflect whether load changes have a distinguishable impact on the measurement residuals, thereby avoiding the phenomenon of parameters being unobservable or weakly observable due to insufficient information. The core function of step S120 above, which determines whether the load parameters meet the observability conditions, is to provide a theoretical feasibility guarantee for the initiation and execution of load deviation estimation, ensuring that load estimation (or load deviation estimation) is activated under conditions where the aircraft's dynamic response is sensitive to load changes. By selecting a time window with sufficient vertical dynamic excitation, the observability conditions can effectively suppress estimation drift caused by insufficient load contribution under weak maneuvering conditions such as hovering and constant speed cruise, preventing load and wind speed from coupling and absorbing each other in the residual interpretation, thereby improving the convergence speed and estimation accuracy of load deviation estimation, and laying a reliable parameter foundation for load state locking in the subsequent wind speed estimation stage.

[0040] Optionally, the above-mentioned determination of the load parameters corresponding to the observability conditions based on flight state data includes: acquiring the vertical dynamic excitation of the aircraft; calculating the excitation index based on the vertical dynamic excitation; wherein the excitation index is used to quantify the vertical excitation intensity of the aircraft; accumulating the duration for which the excitation index meets the preset amplitude condition; when the duration reaches the first preset duration, determining that the load parameters meet the observability conditions, and acquiring the load parameters corresponding to the observability conditions.

[0041] The aforementioned vertical dynamic excitation refers to the active control force or acceleration response generated in the vertical direction of the ground coordinate system. Vertical dynamic excitation can reflect the non-conservative external force action or inertial motion characteristics of the aircraft in the direction of gravity. Since load deviation directly manifests as a change in the total mass of the aircraft, and the change in mass mainly reflects its dynamic effects by influencing the vertical thrust-gravity balance relationship and the vertical acceleration response, vertical dynamic excitation can be used as a key physical quantity for determining the observability of load parameters. When the aircraft performs vertical maneuvers (such as climb, descent, hovering to level flight, etc.) or when the thrust changes, the vertical dynamic excitation is enhanced, and the contribution of load deviation to the vertical motion equation increases, thereby satisfying the continuous excitation conditions required for parameter identification. Vertical dynamic excitation can be acquired through various airborne sensors or control commands. For example, vertical acceleration can be measured by a triaxial accelerometer in the inertial measurement unit (IMU), projected onto the vertical direction of the ground-fixed NED coordinate system via an attitude transformation matrix, and thus obtains a linear acceleration measurement characterizing the aircraft's performance under vertical dynamic excitation. Thrust changes can be calculated based on motor speed commands or electronic speed controller (ESC) PWM signals, combined with the propeller thrust-speed characteristic curve, to obtain the single-propeller thrust. This is then synthesized into a total thrust using a rotor distribution matrix and projected onto the vertical direction of the ground-fixed coordinate system, thus obtaining the instantaneous change in vertical thrust. Control variable changes are obtained by monitoring the difference between flight control commands (such as throttle input, vertical speed command, or altitude command) within adjacent control cycles, reflecting the active excitation intent for the aircraft's vertical maneuvers. These methods of acquiring vertical dynamic excitation can be used individually or in weighted combinations to comprehensively quantify the aircraft's maneuver intensity in the vertical channel.

[0042] The aforementioned excitation indices refer to scalar or vector metrics used to quantify the intensity of vertical dynamic excitation on an aircraft. Through mathematical operations, they transform raw flight state data into comparable excitation level representations, providing a numerical basis for determining the observability conditions of load parameters. The design of excitation indices can take into account the physical nature of load estimation, namely, that mass deviation primarily affects aircraft dynamics through vertical thrust and acceleration.

[0043] Incentive indicators can be constructed in at least one of the following ways: The first method: based on the construction of vertical acceleration measurement; In this method, the excitation index is obtained by calculating the absolute value or norm of the vertical acceleration component in the Earth-fixed coordinate system, i.e. ,in, Indicates incentive indicators; This represents the linear acceleration component of the aircraft in the vertical direction of the NED coordinate system. This excitation index directly quantifies the intensity of the change in inertial force generated by the aircraft in the direction of gravity by calculating the absolute value of the vertical acceleration. It can effectively capture the differences in dynamic response caused by mass inertia during maneuvers such as climb, descent, or hovering to level flight, thus providing a physical basis for the observability determination of load parameters.

[0044] The second method: constructing based on thrust variation; In this approach, incentive indicators The expression can be: ,in, This represents the vertical increment of the spacecraft's total thrust between adjacent sampling times, i.e., the vertical thrust at the current time. Vertical thrust at the previous moment The absolute value of the difference. This excitation index reflects the thrust adjustment range required for the aircraft to perform vertical maneuvers by monitoring the intensity of the active control excitation of the power system in the vertical direction.

[0045] The third method: constructing based on changes in control variables; In this approach, incentive indicators The expression can be: ,in, Indicates the first k One control cycle (or moment) k Vertical control commands (such as throttle position, vertical speed command, or normalized thrust command). Indicates the first k -1 control cycle (or time) k -1) Vertical control commands (such as throttle position, vertical speed command or normalized thrust command). This refers to vector norm or absolute value operations. By calculating the change in vertical channel control input within adjacent control cycles, this index predicts the intensity of vertical maneuvers that the aircraft is about to perform or is currently performing. It can identify the load observability window in advance before the actual thrust response or significant changes in acceleration measurement, thus enhancing the timeliness of excitation detection.

[0046] When setting the preset amplitude conditions for the excitation index, the following factors can be considered: (1) Take into account the noise statistical characteristics of the airborne accelerometer and thrust measurement link to ensure that the threshold is higher than the sensor noise floor and the random acceleration fluctuation caused by atmospheric disturbance, so as to prevent the load estimation from being mistakenly triggered due to the instantaneous exceedance of the measurement noise in weak excitation conditions such as hovering or constant speed cruise. (2) The preset amplitude conditions should match the expected load deviation estimation resolution, that is, the excitation intensity should be sufficient to make the load change generate a signal contribution significantly higher than the noise level in the vertical dynamic measurement, so as to ensure that the signal-to-noise ratio of parameter identification meets the convergence requirements. (3) Take into account the physical maneuverability and safety constraints of the aircraft. The preset amplitude conditions should not exceed the acceleration or thrust change level within the normal operating maneuver range, so as to avoid forcing the aircraft to perform violent maneuvers beyond the mission requirements or safety boundaries in order to obtain the observability of the load, and ensure the feasibility and safety of the excitation detection mechanism in engineering applications.

[0047] The above scheme sets a first preset duration excitation window and constructs an entry threshold for the load estimation mode through a time-domain accumulation mechanism. This ensures that the load deviation estimation is triggered when the vertical dynamic excitation exhibits continuous stability rather than transient pulse characteristics. The excitation window setting requires that the duration for which the excitation index meets the preset amplitude condition must reach this first preset duration. This effectively identifies and eliminates false observable windows caused by sensor measurement noise, instantaneous gusts of wind disturbance, or brief operational jitter, preventing premature entry into the load estimation mode under conditions where the excitation intensity has not yet stabilized or only a transient dynamic response exists.

[0048] When setting the first preset duration, the following factors can be considered: (1) Take into account the dynamic response delay and excitation establishment process of the vertical channel of the aircraft, and ensure that the duration is sufficient to cover the transient stage from the action of the control command to the generation of stable vertical acceleration of the aircraft, so as to avoid starting the load estimation in the early stage before the dynamic response has fully unfolded; (2) Be able to effectively filter out the short-term excitation spikes caused by sensor measurement noise, thrust pulsation or instantaneous gust disturbance (for example, set the first preset duration to be greater than a certain number of sampling periods), and ensure that the load estimation stage is allowed only when the vertical dynamic excitation shows a continuous and stable characteristic, so as to prevent estimation jitter or divergence caused by false excitation windows.

[0049] Step S130: Estimate the load deviation to obtain an estimated load deviation value.

[0050] The aforementioned load deviation estimation refers to the deviation of the actual total mass of the aircraft from its nominal mass (i.e., mass deviation) based on the aircraft's vertical dynamic measurement data within the excitation window where load observability conditions are met. The process involves online parameter identification. The load deviation estimation process utilizes the dynamic coupling relationship between load changes and thrust and acceleration in the vertical channel. Under the condition that wind speed-related terms are treated as measurement interference or noise suppression, the deviation parameters characterizing load changes or mass drift are calculated, thereby obtaining reliable load deviation estimates. This provides an accurate parameter benchmark for load state locking and dimensionality reduction in the subsequent wind speed estimation stage.

[0051] The above step S130 can be performed by at least one of the following methods to estimate the load deviation: The first method: Estimate the load deviation based on the vertical residual measurement equation to obtain the estimated load deviation value; In this approach, a residual measurement equation is constructed by comparing the difference between the measured vertical acceleration and the predicted vertical dynamic response based on the nominal mass. Under ideal conditions of no wind and no mass deviation, the difference between vertical thrust and gravity should equal the product of nominal mass and vertical acceleration. When mass deviation exists, the actual vertical acceleration deviates systematically from the predicted value, and this deviation constitutes the residual signal. Its mathematical expression can be summarized as a mismatch term between the change in vertical thrust and the nominal mass multiplied by the vertical acceleration. This term has a linear relationship with the mass deviation and vertical excitation. This implementation fully utilizes the weak coupling characteristics between horizontal wind speed and the vertical dynamics of the aircraft. Under small attitude angle flight conditions, horizontal wind speed mainly affects the motion in the horizontal plane through aerodynamic drag. Its contribution to the vertical thrust demand and vertical acceleration is usually manifested as a small second-order quantity or a negligible term after attitude conversion. In contrast, mass deviation directly acts on the gravity and inertial terms of the vertical motion equation, forming a direct linear mapping relationship with vertical thrust and acceleration. Therefore, by limiting the measurement constraints to the vertical channel, the influence of horizontal wind speed on the measurement residuals is suppressed, and the load deviation becomes the main unknown quantity for explaining the vertical residuals, thus achieving passive isolation from wind disturbances.

[0052] In terms of parameter identification, the wind speed-related coupling terms in the measurement equation can be treated as bounded disturbances or included in the measurement noise covariance statistics. A recursive least squares or Kalman filter framework can be used to estimate the load deviation online. By minimizing the weighted sum of squares of the vertical residuals, or by recursively updating the load state using the mismatch between vertical thrust and acceleration, the estimated load deviation can be directly calculated. Since the vertical channel has strong controllability through direct adjustment of the total thrust, and wind disturbance coupling is relatively weak, the above implementation can effectively identify load parameters during vertical maneuvers without explicitly estimating the horizontal wind speed component.

[0053] The second method is to use joint state estimation to obtain the load deviation estimate. This implementation constructs an augmented state vector encompassing both load and wind speed states, and simultaneously estimates both types of parameters using a single filter framework. Within this framework, load deviation and horizontal wind speed components are jointly incorporated into the state-space model, and a mapping relationship between the state and measurements is established through a unified measurement equation (typically based on the dynamic residuals of the horizontal or vertical plane). This implementation avoids the complexity of designing separate estimators for load and wind speed, and allows for differentiated handling of the two states during the estimation process through parameter configuration.

[0054] To achieve parameter decoupling in the load estimation mode, this implementation can specifically configure the joint state estimator to reconstruct the filter gain allocation mechanism. One approach is to fix the wind speed state value to a preset initial value or the previous estimated value, i.e., freeze the wind speed state update channel, so that the measurement residual drives the load state correction only through the coupling relationship between the state transition matrix and the measurement matrix. Another approach is to increase the process noise covariance of the wind speed state, so that the filter treats the wind speed as a high-uncertainty or strong random walk process in the prediction step, thereby automatically reducing the sensitivity to wind speed measurement information in the update step and correspondingly increasing the filter gain weight of the load state. Alternatively, the measurement noise covariance can be increased to suppress the overall measurement update amplitude, but combined with the process noise setting of the load state, the relative update weight can still be tilted towards the load direction. The technical essence of the above configuration strategies lies in reconstructing the observability structure of the estimator, so that within the vertical excitation window where the load observability condition is satisfied, the measurement information mainly flows to the load state correction channel. By artificially increasing the uncertainty of wind speed or fixing its value, the filter attributes the measurement residuals primarily to load deviation rather than wind disturbance, thereby achieving preferential convergence of the load parameter within the mathematical framework of joint estimation. This soft-constraint mechanism allows the first stage of the two-stage strategy to be completed within a single filter, achieving temporal decoupling between load and wind speed estimation without explicitly switching the estimator structure.

[0055] The third approach is to treat the wind speed term in the measurement equation as the measurement noise term and use the recursive least squares method to fit the load deviation estimate. In the dynamic residual measurement equations in the horizontal or vertical plane, load deviation and wind speed terms typically coexist in a linear combination, meaning the residual signal simultaneously includes contributions from mass deviation and aerodynamic drag. This implementation removes the wind speed-related term (usually manifested as an aerodynamic term proportional to airspeed) from the estimated state and incorporates it into the measurement noise term. This suppresses it as part of the noise statistics during parameter identification, preventing it from participating as an independent parameter in the load deviation fitting process. By treating wind speed as colored noise or a random bias with specific covariance characteristics, the original joint estimation problem is simplified to a parameter identification problem solely for load deviation, thus avoiding the mutual coupling and absorption of the two physical quantities in the same measurement channel.

[0056] The recursive least squares method is applicable to the simplified parameter identification framework described above, particularly for performing fast fitting within a finite excitation window where load observability conditions are met. Since load deviation can be considered a constant or slowly varying parameter over short timescales, and the wind speed term has been reconstructed as part of the measurement noise, the load deviation estimate can be updated online in the continuously arriving measurement data stream by utilizing the exponential weighting or finite memory properties of recursive least squares. The optimal mass deviation estimate is solved by constructing a normal equation relating the vertical or horizontal dynamic residuals to the excitation signal, using the minimization of the weighted sum of squares as the criterion.

[0057] Furthermore, this implementation can define the measurement vector as the dynamic residual after attitude transformation, the regression vector as composed of vertical acceleration or equivalent excitation terms, and the aerodynamic fluctuations caused by wind speed as incorporated into the measurement noise covariance matrix. The recursive update process corrects the prior parameter estimates by calculating the gain matrix and continuously optimizes the load deviation estimate using newly arrived measurement data; when the excitation window ends or the convergence condition is met, the final estimate is output as the convergence result to the subsequent wind speed estimation stage. This approach statistically averages or suppresses the dynamic changes in wind speed, ensuring that the load estimation primarily responds to the systematic deviations caused by mass changes, rather than the transient fluctuations caused by gusts or steady-state winds.

[0058] It is understood that, in addition to the three methods mentioned above, the embodiments of this application can also employ various other methods for load deviation estimation. For example, an augmented state estimation method based on extended Kalman filtering can be used, treating the load deviation as part of the extended state vector. A Taylor expansion linearizes the nonlinear dynamic model, and parameter recursive estimation is achieved through covariance propagation and Kalman gain updates. Alternatively, an extended state observer can be used, treating the load deviation as part of the lumped disturbance for real-time observation and compensation, and utilizing high-gain characteristics to quickly converge to the true value. A parameter identification method based on model reference adaptation can also be used, constructing a reference model and an adaptive law driven by the actual output error to adjust the estimated load deviation value so that the closed-loop error tends to zero. These methods, mathematically, all aim to establish a mapping relationship between aircraft dynamic measurements and mass deviations. Online identification of load parameters can be achieved through different optimization criteria, observer structures, or adaptive mechanisms, as long as they can effectively extract load information and suppress wind speed coupling interference during the period when the vertical dynamic excitation meets the observability conditions.

[0059] Step S140: In response to the load deviation estimate satisfying the preset convergence condition, the update of the load state is restricted, and the wind speed is estimated to obtain the wind speed estimate.

[0060] The aforementioned preset convergence criteria refer to the quantitative standards used to assess whether the load deviation estimation process has reached a statistically stable state. They are primarily used to determine whether the current load deviation estimate has entered a statistically stable state and whether the fluctuation range is within an acceptable range. As the decision-making basis for switching from the load estimation stage to the wind speed estimation stage, the preset convergence criteria ensure that state locking and dimensionality reduction operations are only performed when the load parameter estimation process is sufficiently stable and the estimation results have a sufficient level of confidence. This prevents systematic biases in subsequent wind speed estimations caused by prematurely fixing load deviation estimates that are still in the convergence process.

[0061] Optionally, the aforementioned preset convergence condition includes: within a second preset time period, the convergence statistical characteristics of the load deviation estimate continuously satisfy a preset statistical characteristic threshold.

[0062] The aforementioned second preset duration is designed to provide a time-domain cumulative benchmark for determining the stability of the load deviation estimate, ensuring that the confirmation of the convergence state is based on continuous observation rather than instantaneous sampling. This effectively avoids misjudgments of the convergence state caused by estimation fluctuations, sensor noise, or brief external disturbances, ensuring the reliability of the transition from the load estimation stage to the wind speed estimation stage. By requiring the load deviation estimate to continuously meet the preset statistical characteristic threshold within the second preset duration, the load parameter exhibits stable statistical characteristics in the time domain, providing sufficient data support and confidence guarantees for subsequent state locking operations, preventing the entry into the locking state before the estimation has truly stabilized.

[0063] When setting the second preset duration, the following factors can be considered: (1) Consider the convergence dynamic characteristics of the load estimation algorithm to ensure that the second preset duration can cover the transition process required for parameter estimation to converge from the initial value to the steady state, thereby avoiding misjudging convergence when the estimation error is still in the decay stage. (2) The second preset duration should match the continuous characteristics of the excitation window to ensure that there is sufficient margin for convergence determination during the effective period of vertical dynamic excitation, and to prevent the inability to complete stability verification due to the early termination of excitation. (3) Take into account the ability to distinguish between estimation noise and real parameter fluctuations. If the duration is too short, the instantaneous satisfaction caused by noise may be misjudged as convergence. If the duration is too long, the switching time may be delayed or the opportunity to complete the determination within the limited excitation window may be missed. Therefore, a balance needs to be struck between response speed and determination reliability.

[0064] The convergence statistics of the above load deviation estimates can be expressed in any of the following forms: The first approach: quantification using the time-series rate of change criterion; The degree of convergence is assessed by monitoring the magnitude of changes in the estimated value within the sliding window; for example, calculating the estimated load deviation at the current moment. Compared with the estimated value N sampling periods ago The absolute value of the difference When the change amount remains less than or equal to a preset threshold for a second preset time period When the load deviation estimate has entered the steady-state convergence stage, the parameter update amplitude has decayed to an acceptable level.

[0065] The second form: characterization using an estimation uncertainty criterion; Using the elements of the covariance matrix output by the state estimation algorithm as a statistical confidence measure, when the covariance of the load deviation estimate... The value remains less than or equal to a preset threshold for a second preset duration. When the time is right, it indicates that the estimator's uncertainty about the parameter has been reduced to a sufficient level, the standard deviation of the estimation error is controlled within the predetermined range, and the accuracy requirements of state locking are met.

[0066] The third approach: verification is performed using the consistency criterion of measurement residuals; Dynamic measurement residuals based on load estimation mode y The statistical characteristics are used to assess the quality of the estimate when the statistical mean of the measurement residuals approaches zero and the variance is within a second preset time period. When the value is less than or equal to the preset threshold, it indicates that the estimated load deviation can fully explain the systematic deviation in the measurement data, and the residual mainly exhibits random noise characteristics rather than unmodeled dynamics, thus confirming the effectiveness of the load parameter estimation.

[0067] It is understandable that, in addition to the three forms mentioned above, the convergence statistical characteristics of the load deviation estimate can also be expressed in various other forms. For example, the chi-square consistency test can be used as a convergence statistical characteristic. By calculating the sum of squares of the normalized innovation sequence (innovation refers to the deviation between the actual measurement value at the current moment and the expected measurement value predicted by the measurement model based on the state estimation at the previous moment during the recursive update process of the state estimation) and comparing it with the critical value of the chi-square distribution, it can be determined whether the statistical characteristics of the measurement residual under the load estimation mode are consistent with the theoretical noise model, thereby verifying the correctness of the load estimation. Alternatively, the white noise test of the measurement residual sequence can be used. The autocorrelation function or spectral analysis method can be used to determine whether the innovation sequence exhibits unbiased and uncorrelated random characteristics, which can be used as an indirect criterion for filter convergence. The trace of the covariance matrix (that is, the sum of the variances of each component in the covariance matrix, which can characterize the overall dispersion of the data) or the condition number can also be used as a convergence measure to evaluate the overall reduction of estimation uncertainty or numerical stability. These statistical characteristics can all be used to quantify the convergence state of load deviation estimation, as long as they can continuously meet the preset threshold in the time domain and reflect the stability and reliability of the estimation.

[0068] Optionally, the above-mentioned limitation on the update of the load state includes: fixing the load state value to the sum of the nominal mass and the estimated load deviation value, and removing the load state from the estimated state vector; or setting the process noise covariance of the load state to zero; or limiting the update amplitude or update rate of the load state by gating the filter gain.

[0069] The following sections will describe the three methods for updating the load limit status: The first approach is to fix the load state value as the sum of the nominal mass and the estimated load deviation, and remove the load state from the estimated state vector. This method is a hard-locking method. Hard-locking refers to fixing the value of the parameter to be locked to a certain value during the state estimation process and physically removing it from the estimated state vector, so that the parameter will not be estimated again in subsequent estimation updates.

[0070] In the first method described above, after confirming the estimated load deviation... Once the convergence condition is met, the estimated value is compared with the pre-calibrated nominal mass of the aircraft. By performing algebraic superposition, a fixed estimate of the total mass is obtained. This mass value is maintained as a constant during subsequent wind speed estimation. Simultaneously, the load state component is removed from the state-space representation, and the dimensions of the state transition matrix, process noise covariance matrix, and measurement matrix are adjusted accordingly. This reduces the original joint state vector containing both load and wind speed to a low-dimensional subspace containing only the wind speed component. The reduced-dimensional horizontal surface dynamic measurement equations can be reconstructed as follows: ,in, This indicates the residual of the horizontal plane measurement. Represents linear acceleration in the horizontal direction. This represents the aerodynamic drag coefficient matrix on the horizontal plane. This represents the horizontal wind speed to be estimated. This indicates the measurement noise term. Due to load deviation... Since the load and wind speed are fixed as known constants, the left side of the equation forms an effective measurement input, and the right side is left with only wind speed as the only unknown state. The competition between load and wind speed in the original coupled equation is completely eliminated, and wind speed estimation is transformed into a standard low-dimensional state estimation problem based on determined mass parameters.

[0071] This approach physically removes the load state from the estimated state vector and fixes its value, decoupling the load deviation and wind speed in the dynamic measurement equation. This prevents the load estimation from interfering with the wind speed estimation, avoids the mutual absorption of parameters caused by ambiguity in the residual interpretation of the estimator, reduces the computational complexity and memory usage of the state estimator, improves numerical stability and real-time performance, ensures that the load parameter remains constant during long-term wind speed estimation, and eliminates the risk of filter divergence caused by excessively high state vector dimension or parameter correlation, thus achieving decoupling and stability in wind speed estimation.

[0072] The second approach is to set the process noise covariance under load conditions to zero. This method is a soft-locking implementation. Soft-locking means that during the state estimation process, the parameters to be locked in the estimated state vector are retained, but the response and update magnitude of the state to the measurement residual are limited by setting the process noise covariance of the parameter to zero or a minimum value, or by gating and limiting the filter gain, so that the estimated value remains basically constant in the subsequent estimation stage, while retaining a limited adaptive correction margin.

[0073] Unlike the strategy in hard locking that removes the load state from the estimated state vector, the second approach described above preserves the load bias during the wind speed estimation stage. As a component of the augmented state vector, but its corresponding diagonal elements in the process noise covariance matrix (or The value is set to zero or a very small amount close to zero, thereby mathematically constraining the growth of the prediction uncertainty of the state.

[0074] Within the Kalman filter framework, process noise covariance It characterizes the random walk strength or model uncertainty of the load state in the prediction step; when When set to zero, the prior covariance of the load state in the prediction step retains only the posterior covariance of the previous time step without adding the process noise increment, resulting in the Kalman gain of the corresponding state being greatly suppressed in the update step. This is due to the filter gain... It is directly proportional to the process noise covariance. This ensures that the load condition receives almost no correction information from the measurement residuals, and its estimated value remains at... By moving away from changes in dynamic measurements and remaining in the vicinity, the load state is effectively frozen while preserving the complete dimensional structure of the state vector.

[0075] This soft-locking mechanism maintains the decoupling effect of wind speed estimation while providing a limited adaptive margin for the long-term stability of the load state. It avoids the model mismatch accumulation problem that may be caused by complete fixation in the hard-locking strategy, and takes into account both estimation stability and the ability to track extremely slow parameter changes.

[0076] The third method is to limit the update amplitude or update rate of the load status by gating the filter gain. This method, also known as soft-locking, actively constrains the load estimate correction within the Kalman filter framework by gating the filter gain to limit the magnitude or rate of load status update. During the measurement update phase, a threshold threshold or saturation function is applied to the calculated load status Kalman gain. When the gain magnitude exceeds a preset threshold, it is truncated or scaled to a limited range. Alternatively, upper limits on the magnitude and rate of the status update are directly imposed, ensuring that the load deviation estimate can only adjust slowly within a small neighborhood during the wind speed estimation phase, rather than responding freely to the measurement residuals. This gating mechanism effectively suppresses sudden changes in load status caused by instantaneous measurement noise, sensor anomalies, or gust disturbances, preventing unpredictable jumps or drifts in the locked load deviation estimate while preserving the limited ability to track actual, slow changes in load status.

[0077] This approach, by explicitly limiting the update magnitude or rate, provides the estimator with a certain robustness margin while keeping the load state basically constant. It avoids the accumulation of model mismatch that may occur due to complete fixation under the hard-lock strategy, and also overcomes the implicit update risk that may still occur due to numerical errors or abnormal measurements when only the process noise covariance is set to zero.

[0078] It is understandable that, in addition to the three methods mentioned above, information matrix constraints based on the information filtering framework or state projection mechanisms based on constraint optimization can also be used to restrict the update of the load state. In the implementation of information filtering, the diagonal elements of the information matrix (i.e., the inverse of the covariance matrix) corresponding to the load state can be set to maximum values ​​to characterize that the state has extremely high prior certainty. Since the posterior estimate of information filtering is a normalized combination of information vector and information matrix, the extremely large information matrix elements greatly suppress the contribution of measurement residuals to the correction of the load state, effectively achieving soft locking of the load state. Another implementation method uses projective Kalman filtering or constraint state estimation to constrain the load state within a closed interval centered on the convergence value and with a preset threshold as the radius. When the standard filtering update causes the load deviation estimate to exceed this constraint interval, the projection operator forces the state estimate back to the boundary or constraint hyperplane, thereby limiting the update amplitude of the load state in geometric space and ensuring that it always remains within the allowable neighborhood of the convergence value, achieving a locking effect equivalent to gain gating but based on state space geometric constraints.

[0079] Optionally, the above-mentioned aircraft state estimation method may further include: monitoring anomaly indicators when estimating wind speed; wherein the anomaly indicators are used to quantify the degree of mismatch between the estimated load deviation and flight dynamics measurements, or to quantify the degree of change in load parameters; the degree of mismatch is used to quantify the degree of deviation between the dynamic measurements predicted based on the estimated load deviation and the actual dynamic measurements in the flight state data; in response to the anomaly indicators meeting the preset re-entry conditions and the load parameters meeting the observability conditions, re-executing the load deviation estimation.

[0080] The aforementioned re-entry mechanism for load estimation aims to adapt to dynamic load changes during flight missions. It ensures timely detection of mismatches between the locked load value and the actual mass under conditions of sudden mass changes such as loading, dropping, or load transfer. This allows for the reactivation of load deviation estimation to obtain updated mass parameters, thus avoiding residual divergence caused by continuous wind speed estimation based on an incorrect load baseline. The re-entry mechanism identifies the degree of mismatch between the load deviation estimate and flight dynamics measurements, or potential changes in load parameters, by monitoring abnormal indicators. Combined with the re-satisfaction of load observability conditions, it achieves a reverse switch from the wind speed estimation stage to the load estimation stage. This allows the state estimator to dynamically adjust its estimation focus based on actual conditions throughout the entire mission cycle, rather than permanently locking the load after a one-time identification. This enhances long-term operational stability and parameter tracking capabilities under complex operational scenarios.

[0081] The aforementioned anomaly indicators refer to metrics used during the wind speed estimation phase to quantify the consistency between previously converged and locked load deviation estimates and current flight dynamics measurements, or to characterize the magnitude of substantial changes in load parameters relative to locked values. By monitoring the statistical characteristics or state deviation of measurement residuals, anomaly indicators identify load estimation mismatches or sudden changes in actual mass caused by load loading, drop operations, or external disturbances. This provides a quantitative basis for determining whether to exit the current wind speed estimation mode and reactivate load deviation estimation, ensuring that a re-entry mechanism is triggered promptly to update mass parameters when the aircraft's mass characteristics undergo substantial changes. This avoids the accumulation of systematic deviations caused by continuously performing wind speed estimations on outdated load benchmarks.

[0082] The aforementioned abnormal indicators can be constructed using the measurement residual norm. By monitoring the amplitude level of the dynamic residual during the wind speed estimation stage, the degree of mismatch between the load lock-in value and the actual mass can be identified. This can be achieved by calculating the actual measured values. Compared with the measured values ​​based on the current state estimate and the locked load parameter prediction Deviation norm between ,in, express kThe flight dynamics measurement vector (such as acceleration or velocity) at any given moment; This represents the predicted measurement vector calculated based on the state prediction model and the locked load parameters; when this residual norm continuously exceeds a preset threshold... If the time exceeds a certain period, it indicates that the locked load value can no longer fully explain the current dynamic measurement, and a mismatch is determined to occur, triggering the re-entry condition.

[0083] The aforementioned anomaly indicators can also be constructed using the time-series rate of change of wind speed estimates. Changes in load parameters can be indirectly identified by monitoring the abrupt changes in the wind speed estimates. The norm of the difference between wind speed estimates at adjacent sampling times is calculated. ,in, express k The horizontal wind speed estimation vector at time t. express k The corresponding estimated value at time 1; when the rate of change exceeds a preset threshold Furthermore, when the aerodynamic response is inconsistent with that implied by the flight control input, it suggests that the sudden change in current wind speed may be due to a mismatch in the dynamic model caused by load changes rather than a change in the actual wind field.

[0084] The aforementioned abnormal indicators can also be constructed using the steady-state drift of thrust load or motor control quantities. By monitoring abnormal changes in the thrust required to maintain a specific flight state, changes in load parameters can be directly quantified. The mean drift of thrust command or motor load current within a specific time window can be statistically analyzed. ,in, This indicates the deviation between the current average thrust value and the historical average thrust value. When this drift continues to exceed the preset threshold, it indicates that the aircraft needs to systematically adjust its thrust in order to maintain the same motion state, suggesting that the total mass may have undergone sudden changes such as loading or jettisoning.

[0085] In addition to the criteria based on instantaneous amplitude mentioned above, outlier indicators can also be assessed using cumulative sum statistical tests or model validation consistency. For example, a cumulative sum control chart can be used to weight and accumulate the standardized residual series, and the small, slow shifts in load can be identified by monitoring the continuous increase of the cumulative statistic. Alternatively, a sliding window chi-square test can be used to evaluate the statistical distribution characteristics of the residual series, and model mismatch can be determined when the test statistic exceeds the chi-square distribution critical value, thereby detecting latent changes in load parameters earlier.

[0086] Optionally, the above-mentioned aircraft state estimation method may further include: when switching from load deviation estimation to wind speed estimation, or from wind speed estimation to load deviation estimation, determining whether the duration of the current estimation mode has reached a preset minimum dwell time; if the duration of the current estimation mode meets the preset minimum dwell time, then switching the estimation mode.

[0087] The purpose of setting a minimum dwell time for the estimation model in the above scheme is to provide a sufficient stable operating time window for the current estimation model, ensuring that the load deviation estimation or wind speed estimation has reached a sufficient degree of convergence and statistical stability before model switching, and avoiding frequent switching of the estimation model due to instantaneous condition satisfaction or short-term fluctuations. By forcing the current estimation model to run continuously for the minimum dwell time, the scheme effectively suppresses model oscillations caused by fluctuations in excitation conditions, convergence criteria, or abnormal indicators near critical values, ensuring a smooth transition and long-term operational reliability of load deviation estimation and wind speed estimation.

[0088] When setting the minimum dwell time, the following factors can be considered: (1) Take into account the convergence dynamic characteristics of both load deviation estimation and wind speed estimation, and ensure that the dwell time is greater than the typical convergence time required for the state estimation algorithm to recover from the initial value or disturbed state to the steady state, so as to avoid the degradation of parameter estimation quality caused by switching modes before the estimation is sufficiently stable. (2) The minimum dwell time should match the duration of the excitation conditions under each estimation mode, and ensure that there is sufficient margin for estimation convergence and verification during the vertical excitation window or the active period of wind speed change, so as to prevent the loss of sufficient parameter identification opportunities due to premature switching. (3) The minimum dwell time should fully consider the transient disturbances or numerical shocks that may occur during the mode switching process, and reserve buffer time to allow the filter or observer to re-establish stable tracking after switching, so as to avoid the risk of ill-conditioned growth of estimation covariance or numerical divergence caused by too frequent switching. (4) The minimum dwell time setting should also be coordinated with the overall time scale of the flight mission to ensure that the response to changes in operating conditions is not excessively delayed while meeting the requirements for estimation stability, and to achieve an engineering-acceptable balance between the agility of mode switching and the robustness of the estimation process.

[0089] Optionally, the above-mentioned aircraft state estimation method may further include: in response to the detection of gust conditions, performing wind speed estimation after increasing the estimation bandwidth of wind speed estimation or increasing the process noise covariance of wind speed estimation.

[0090] The aforementioned gust condition refers to an external disturbance condition affecting an aircraft where the wind speed vector's time-varying rate exceeds a preset steady-state threshold, causing unsteady aerodynamic loads and dynamic responses. Gust conditions can lead to sudden deviations in the aircraft's attitude and trajectory, significantly impacting measurement residuals and causing estimation lag in wind speed estimation algorithms based on steady-state assumptions. Gust conditions can be detected in various ways, such as monitoring the time-series variation characteristics of wind speed estimates. By analyzing the differences or short-term trends between wind speed estimates at adjacent times, rapid wind changes exceeding the normal steady-state fluctuation range can be identified. Another approach is to monitor the statistical characteristics of measurement residuals. Given a locked load condition, an abnormal increase in dynamic measurement residuals or abrupt changes in statistical characteristics may indicate drastic changes in external wind disturbances rather than a mismatch in internal model parameters. Furthermore, information from external sensors, such as airborne airspeed indicators or weather radar, can be combined to directly detect transient changes in incoming airflow velocity, thereby determining whether the aircraft is under gust conditions.

[0091] Under gusty conditions, wind speed exhibits high-frequency, rapidly changing, and non-stationary characteristics. If a low-bandwidth estimation strategy designed for steady-state winds is maintained, the estimator will experience tracking lag due to insufficient inertial response, causing the estimated wind speed to deviate from the actual wind field, thus affecting flight control accuracy and disturbance rejection performance. Increasing the bandwidth of wind speed estimation enhances the estimator's sensitivity and response speed to rapid wind changes, enabling it to capture transient peaks and directional changes in gusts in a timely manner, reducing control mismatch and position drift caused by estimation delays. Increasing the process noise covariance of wind speed estimation is an effective means of adjusting the prediction and measurement confidence weights within the filtering framework. In Kalman filtering or similar state estimation structures, the process noise covariance characterizes the increase in uncertainty of the wind speed state in the prediction step; increasing it means the model imparts stronger random walk characteristics to the wind speed, reducing the confidence of the prior estimate in the prediction step, thereby tilting the filter gain towards the measurement information at the current moment. By increasing the estimation bandwidth of wind speed estimation or increasing the adjustment of the process noise covariance of wind speed estimation, the estimator relies more on real-time dynamic measurements rather than historical predictions during gusts, enabling it to respond quickly to drastic changes in wind speed. This avoids gust estimation attenuation or phase delay caused by excessive smoothing, and improves the estimation fidelity and robustness in complex wind disturbance environments.

[0092] Furthermore, it is understandable that, during long-term operation, combining load mutation detection with mission events (such as grabbing, dropping, and other operational actions) or controller steady-state deviation can reliably trigger the re-entry mechanism by utilizing mission scheduling information or the inherent characteristics of the control loop. Among these, mission events provide deterministic prior timescales for mass changes, enabling the system to synchronously or preemptively activate load re-estimation at the physical moment of load mutation, avoiding response delays caused by waiting for the accumulation of dynamic residuals. Meanwhile, controller steady-state deviation reflects the steady-state offset of control input required to maintain the target's flight state and can be detected in the early stages of mass changes, thereby achieving early warning and rapid confirmation of load mutations. The combination of the two can improve the timeliness and accuracy of re-entry triggering and reduce the risk of missed or false detections that may be caused by relying solely on measurement residual statistics.

[0093] To facilitate understanding of the working principle of the above-described aircraft state estimation method, this application also provides a specific application example of this method in a certain application scenario. The details of this example are as follows: First, the parameters involved in the following content are defined: The coordinate system adopted is the Earth-fixed NED (North-East-Down) coordinate system. ; Indicates the linear velocity of the aircraft; This represents the linear acceleration after gravity has been removed; Indicates the thrust-ground component; Indicates the resistance parameter; Indicates nominal mass; Indicates mass deviation (load). Indicates horizontal wind speed; This represents the estimated wind speed. This represents the estimated load deviation. Indicates incentive indicators; Indicates the excitation threshold; Indicates the first preset duration; Indicates the convergence criterion; Indicates the second preset duration; Indicates the minimum dwell time; Indicates abnormal indicators; Indicates the current mode (MASS_MODE or WIND_MODE).

[0094] like Figure 2 As shown, in this application scenario, the above-mentioned aircraft state estimation methods mainly include: Step 1: Obtain the flight status data and parameters of the aircraft; Get navigation speed acceleration away from the line of gravity Thrust system components Resistance parameters and nominal quality , which serve as input measurements and model parameters for state estimation.

[0095] Step 2: Calculate incentive indicators And determine whether the load observability condition is met; like Figure 3 As shown, step two mainly includes: The excitation index is calculated based on vertical dynamic excitation. The excitation index can be vertical acceleration. Thrust change Or control quantity change .

[0096] Next, determine whether the incentive index meets the preset amplitude condition. , where the threshold The settings are adjusted based on acceleration noise and desired load resolution, and the cumulative duration must reach the first preset duration. , It can be greater than a certain number of sampling periods to avoid triggering by instantaneous noise; when the above conditions are met, it is determined that the load parameters meet the observable conditions and enter MASS_MODE.

[0097] Step 3: Perform load deviation estimation in MASS_MODE to obtain the estimated load deviation value; Within a window where load observability is high, based on the horizontal plane measurement equation Identify the coupling structure between load and wind speed. For example... Figure 4 As shown, step three above can prioritize the estimation of load using any of the following methods: Method A: Estimate load deviation based on vertical residual measurement equation, and use the low sensitivity of vertical channel to wind speed to suppress the impact of wind; Method B: Use a joint state estimator but freeze the wind speed state, increase wind process noise or increase measurement noise, so that the update mainly affects the load. Method C: Within the excitation window, the wind speed term in the measurement equation is treated as measurement noise, and the load deviation is fitted using the recursive least squares method.

[0098] Step 4: Determine whether the estimated load deviation meets the preset convergence conditions. ; Preset convergence conditions It can include: The rate of change criterion, i.e., within the sliding window ; Uncertainty criterion, i.e., estimating covariance ; The residual consistency criterion is that the mean of the residuals under the load mode is close to zero and the variance is... Less than the threshold.

[0099] In addition, to avoid jitter, a hysteresis mechanism can be added, that is, the above criteria must be met continuously for a second preset duration. Switching is only allowed after that.

[0100] Step 5: When the convergence condition is met and the duration reaches [a certain value], [the following steps are taken]. At that time, limit the update of the load status and enter WIND_MODE to perform wind speed estimation; After switching, either a hard lock or a soft lock strategy is used to restrict the update of the load status, wherein: Hard locking, fixing the load condition value to... Furthermore, by removing the load state from the estimated state vector, the joint estimation is reduced to a low-dimensional wind estimation problem. The measurement equation after dimensionality reduction is as follows: ; Soft locking preserves the load state but reduces the process noise covariance. Setting the value to zero (or close to zero), or limiting the update amplitude or rate of the load state by gating the filter gain, is suitable for scenarios with slow drift. Under the premise of load locking, any algorithm such as Kalman filtering, Extended State Observer (LESO), or low-pass filtering with analytical inverse calculation can be used to estimate the horizontal wind speed. .

[0101] Step 6: Continuously monitor abnormal indicators in WIND_MODE And determine whether the re-entry conditions of the load estimation mode are met; Anomaly indicators are used to quantify the degree of mismatch between the estimated load deviation and flight dynamics measurements, or to quantify the degree of change in load parameters. The re-entry conditions for the load estimation mode can take any of the following forms: residual norm in wind mode Continuously greater than the threshold ; Estimated sudden change in wind speed And it is inconsistent with the control quantity; Thrust load mean drift Exceeding the threshold.

[0102] When the abnormal indicator meets the preset re-entry condition, and at the same time the load excitation window is detected to meet the condition again ( And continue When MASS_MODE is triggered, the load is re-estimated.

[0103] In this application scenario, a minimum dwell time can be set for each estimation mode. Prevent frequent switching; In addition, a gust mode (GUST_MODE) can be added to improve the wind speed estimation bandwidth or increase the process noise when gust conditions are detected, thereby enhancing the tracking capability.

[0104] Step 7: Output the estimated load deviation value Horizontal wind speed estimate Current mode identifier and credibility metrics (such as covariance) Residual statistics (etc.), for use by the Flight Control Task Manager.

[0105] like Figure 5 As shown in the figure, this application embodiment also provides an aircraft state estimation device 200, including: Data acquisition module 210 is used to acquire flight status data of the aircraft; The load parameter acquisition module 220 is used to determine the load parameters corresponding to the observable conditions based on the flight state data; wherein the observable conditions are determined based on the vertical dynamic excitation in the flight state data. The load estimation module 230 is used to estimate the load deviation to obtain the estimated load deviation value; The wind speed estimation module 240 is used to limit the update of the load state in response to the load deviation estimation value satisfying the preset convergence condition, and to estimate the wind speed to obtain the wind speed estimation value.

[0106] It is understood that the aircraft state estimation device 200 provided in this application embodiment can be used to execute the aircraft state estimation method provided in this application embodiment. Its implementation principle and the resulting technical effects have been described in the foregoing method embodiment. For the sake of brevity, any part not mentioned in the device embodiment can be referred to the corresponding content in any of the foregoing method embodiments.

[0107] Figure 6 This is a schematic diagram of an electronic device provided in an embodiment of this application. (Refer to...) Figure 6 The electronic device 300 includes a processor 310, a memory 320, and a communication interface 330. These components are interconnected and communicate with each other via a communication bus 340 and / or other forms of connection mechanism (not shown).

[0108] The memory 320 includes one or more (only one is shown in the figure), which may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The processor 310 and other possible components may access the memory 320 to read and / or write data therein.

[0109] Processor 310 includes one or more (only one is shown in the figure), which can be an integrated circuit chip with signal processing capabilities. The processor 310 described above can be a general-purpose processor, including a central processing unit (CPU), a microcontroller unit (MCU), a network processor (NP), or other conventional processors; it can also be a special-purpose processor, including a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0110] Communication interface 330 includes one or more (only one is shown in the figure) that can be used to communicate directly or indirectly with other devices to exchange data. For example, communication interface 330 can be an Ethernet interface; it can be a mobile communication network interface, such as an interface for 3G, 4G, or 5G networks; or it can be other types of interfaces with data transmission and reception capabilities.

[0111] One or more computer program instructions may be stored in memory 320, and processor 310 may read and run these computer program instructions to implement the aircraft state estimation method provided in the embodiments of this application and other desired functions.

[0112] Understandable. Figure 6The structure shown is for illustrative purposes only; the electronic device 300 may also include components that are more advanced than those shown. Figure 6 The more or fewer components shown, or having the same Figure 6 The different configurations shown. Figure 6 The components shown can be implemented using hardware, software, or a combination thereof. For example, electronic device 300 can be a single server (or other device with computing power), a combination of multiple servers, a cluster of a large number of servers, etc., and can be either a physical device or a virtual device.

[0113] This application also provides a computer-readable storage medium storing computer program instructions. These computer program instructions are read and executed by a processor to perform the aircraft state estimation method provided in this application. For example, the computer-readable storage medium can be implemented as follows: Figure 6 The memory 320 in the electronic device 300, or a separate storage product (such as a USB flash drive, portable hard drive, etc.).

[0114] This application also provides a computer program product, which includes computer program instructions. These computer program instructions are read and executed by a processor to perform the aircraft state estimation method provided in this application. For example, these computer program instructions can be stored in... Figure 6 The memory 320 in the electronic device 300 is located inside the memory, or it is stored in a separate storage product (such as a USB flash drive, portable hard drive, etc.).

[0115] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0116] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0117] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0118] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for estimating the state of an aircraft, characterized in that, The method includes: Acquire flight status data of the aircraft; Based on the flight status data, the load parameters corresponding to the condition of being observable are determined; wherein, the observable condition is determined based on the vertical dynamic excitation in the flight status data; The load deviation is estimated to obtain the estimated load deviation value; In response to the load deviation estimate satisfying the preset convergence condition, the update of the load state is restricted, and the wind speed is estimated to obtain the wind speed estimate.

2. The aircraft state estimation method according to claim 1, characterized in that, The update of the load limit status includes: The load condition value is fixed as the sum of the nominal mass and the estimated load deviation value, and the load condition is removed from the estimated state vector; Alternatively, the process noise covariance of the load state can be set to zero. Alternatively, the update amplitude or update rate of the load state can be limited by gating the filter gain.

3. The aircraft state estimation method according to claim 1, characterized in that, The method further includes: When estimating the wind speed, anomaly indicators are monitored; wherein, the anomaly indicators are used to quantify the degree of mismatch between the estimated load deviation and the flight dynamics measurement, or to quantify the degree of change of the load parameters; the degree of mismatch is used to quantify the degree of deviation between the dynamics measurement predicted based on the estimated load deviation and the actual dynamics measurement in the flight status data; In response to the abnormal index meeting the preset re-entry condition and the load parameter meeting the observable condition, the load deviation estimation is re-executed.

4. The aircraft state estimation method according to claim 1, characterized in that, The determination of the load parameters corresponding to the observable conditions based on the flight status data includes: The vertical dynamic excitation of the aircraft is obtained from the flight status data; Based on the vertical dynamic excitation, an excitation index is calculated; wherein, the excitation index is used to quantify the vertical excitation intensity of the aircraft. The cumulative duration for which the incentive index meets the preset amplitude condition; When the duration reaches the first preset duration, it is determined that the load parameter meets the observability condition, and the load parameter corresponding to the observability condition is obtained.

5. The aircraft state estimation method according to claim 1, characterized in that, The preset convergence condition includes: within a second preset time period, the convergence statistical characteristics of the load deviation estimate continuously satisfy the preset statistical characteristic threshold.

6. The aircraft state estimation method according to any one of claims 1 to 5, characterized in that, The estimation of load deviation to obtain an estimated load deviation value includes: The load deviation is estimated based on the vertical residual measurement equation, and the estimated value of the load deviation is obtained. Alternatively, the load deviation can be estimated using joint state estimation to obtain the estimated value of the load deviation; Alternatively, the wind speed term in the measurement equation can be treated as the measurement noise term, and the estimated load deviation can be fitted using the recursive least squares method.

7. The aircraft state estimation method according to any one of claims 1 to 5, characterized in that, The method further includes: When switching from load deviation estimation to wind speed estimation, or vice versa, determine whether the duration of the current estimation mode has reached the preset minimum dwell time. If the duration of the current estimation mode meets the preset minimum dwell time, then the estimation mode is switched.

8. The aircraft state estimation method according to any one of claims 1 to 5, characterized in that, The estimation of wind speed includes: In response to the detection of gust conditions, the wind speed estimation is performed after either increasing the estimation bandwidth of the wind speed estimation or increasing the process noise covariance of the wind speed estimation.

9. An aircraft state estimation device, characterized in that, include: The data acquisition module is used to acquire the flight status data of the aircraft; The load parameter acquisition module is used to determine the load parameters corresponding to the observable conditions based on the flight status data; wherein the observable conditions are determined based on the vertical dynamic excitation in the flight status data. The load estimation module is used to estimate the load deviation to obtain the estimated load deviation value; The wind speed estimation module is used to limit the update of the load state in response to the load deviation estimation value satisfying the preset convergence condition, and to estimate the wind speed to obtain the wind speed estimate value.

10. An electronic device, characterized in that, include: A processor, a memory, and a communication bus, wherein the processor and the memory communicate with each other via the communication bus; The memory stores program instructions that can be executed by the processor, and the processor can execute the method as described in any one of claims 1 to 8 by calling the program instructions.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a computer, cause the computer to perform the method as described in any one of claims 1 to 8.

Images