Airspeed estimation method, device, apparatus, and computer-readable storage medium

Abstract

This application discloses an airspeed estimation method, apparatus, device, and computer-readable storage medium. Specifically, the method comprises: reading a first state vector and a target state control variable of a target device at a current moment; inputting the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed based on the nonlinear relationship between the state vector and the state control variable of the target device; determining an observation vector corresponding to the second state vector, and adjusting the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector; determining a drag speed based on a drag state parameter in the target state vector, and determining an estimated speed based on a speed state parameter in the target state vector; and estimating a target airspeed of the target device at the next moment based on the difference between the estimated speed and the drag speed. In this manner, the airspeed of a flying vehicle can be accurately estimated, thereby improving the safety of the flying vehicle.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an airspeed estimation method, apparatus, device, and computer-readable storage medium. Background Art

[0002] With the advent of the concept of flying cars, the automotive industry has invested in research and development. During this development phase, safety has drawn significant attention. Flying cars contain structural components, and the wear and wind resistance of these components are closely related to flight speed. Therefore, to improve flying car safety, it is necessary to estimate the flight speed. Related technologies calculate the flight speed of a flying car by measuring airflow pressure data using a pitot tube and calculating the flight speed based on this airflow pressure data.

[0003] However, when the relevant technology calculates the flight speed through the air pressure data of the pitot tube, the instability of the airflow can easily interfere with the pitot tube's ability to perceive the air pressure, affecting the reliability of the air pressure data, thereby causing the calculated flight speed to be inaccurate and reducing the safety of the flying car. Summary of the Invention

[0004] Embodiments of the present application provide an airspeed estimation method, apparatus, device, and computer-readable storage medium, which can accurately estimate the airspeed of a flying car and improve the safety of the flying car.

[0005] The present invention provides an airspeed estimation method, comprising:

[0006] Reading the first state vector and target state control amount of the target device at the current moment;

[0007] Inputting the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed based on a nonlinear relationship between the state vector and the state control variable of the target device;

[0008] Determining an observation vector corresponding to the second state vector, and adjusting the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector;

[0009] determining a windage speed according to the windage state parameter in the target state vector, and determining an estimated speed according to the speed state parameter in the target state vector;

[0010] A target airspeed of the target device at a next moment is estimated based on the difference between the estimated speed and the wind resistance speed.

[0011] Accordingly, an embodiment of the present application provides an airspeed estimation device, comprising:

[0012] a reading unit, configured to read a first state vector and a target state control variable of the target device at a current moment;

[0013] an input unit, configured to input the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed based on a nonlinear relationship between the state vector and the state control variable of the target device;

[0014] an adjustment unit, configured to determine an observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector in an unscented Kalman filtering manner to obtain a target state vector;

[0015] a determining unit, configured to determine a windage speed according to the windage state parameter in the target state vector, and to determine an estimated speed according to the speed state parameter in the target state vector;

[0016] An estimating unit is configured to estimate a target airspeed of the target device at a next moment based on a difference between the estimated speed and the windage speed.

[0017] In some embodiments, the airspeed estimation device further includes a construction unit for:

[0018] Obtaining motion data of the target device at historical moments;

[0019] generating, based on the motion data, a state vector and a state control variable associated with the target device at a historical moment, wherein the state vector and the state control variable have a nonlinear corresponding relationship;

[0020] A dynamic model is constructed based on the nonlinear correspondence between the state vector and the state control quantity.

[0021] In some embodiments, the building block is further used to:

[0022] extracting the thrust coefficient, drag coefficient, and speed of the target device at a historical moment from the motion data;

[0023] Decomposing the thrust coefficient, wind resistance coefficient, and speed in multiple directions according to the three-dimensional spatial relationship to obtain the thrust sub-coefficient, wind resistance sub-coefficient, and directional speed of the target device in each direction;

[0024] Determine the attitude angle corresponding to the target device according to the thruster coefficient, and determine the angular rate corresponding to the attitude angle in each direction according to the Euler angle transformation relationship;

[0025] Constructing a state vector of the target device at a historical moment according to the attitude angle, directional speed and windage coefficient in each direction;

[0026] The wind sub-coefficient associated with the wind resistance sub-coefficient in the corresponding direction is read, and the state control quantity of the target device at the historical moment is constructed according to the angular velocity, the wind sub-coefficient and the thrust sub-coefficient.

[0027] In some embodiments, the adjustment unit is further configured to:

[0028] Based on an unscented transformation, a plurality of key points corresponding to the second state vector are obtained, and a predicted state quantity corresponding to each key point is calculated according to the target state control quantity and the state quantity of each key point;

[0029] Calculating the state variance of multiple key points according to the predicted state quantity and the state probability distribution relationship;

[0030] Determine the target observation value corresponding to each predicted state quantity according to the observation vector, and calculate the observation value variance between the multiple target observation values ​​according to the state probability distribution relationship;

[0031] The second state vector is adjusted according to the difference between the state variance and the observation value variance to obtain a target state vector.

[0032] In some embodiments, the adjustment unit is further configured to:

[0033] performing a nonlinear transformation on the second state vector to obtain a Gaussian distribution curve of the second state number;

[0034] The next moment is used as the midpoint of the state, and key points are symmetrically selected on the Gaussian distribution curve to obtain multiple key points corresponding to the second state vector, each key point corresponding to a state quantity. In some embodiments, the adjustment unit is further used to:

[0035] Performing weighted summation on the plurality of predicted state quantities to obtain state means corresponding to the plurality of key points;

[0036] Determining a state difference between each predicted state quantity and the state mean;

[0037] According to the state probability distribution relationship, the state variance corresponding to the multiple state difference values ​​is calculated.

[0038] In some embodiments, the adjustment unit is further configured to:

[0039] Calculating the covariance between the state variance and the observation value variance according to the state probability distribution relationship;

[0040] Determining a gain coefficient according to a ratio between the covariance and the observed value variance;

[0041] The second state vector is adjusted according to the gain coefficient to obtain an adjusted target state vector.

[0042] In addition, an embodiment of the present application also provides a computer device, including a processor and a memory, the memory storing a computer program, and the processor is used to run the computer program in the memory to implement the steps in the airspeed estimation method provided in the embodiment of the present application.

[0043] In addition, an embodiment of the present application further provides a computer-readable storage medium, which stores a plurality of instructions suitable for loading by a processor to execute the steps of any one of the airspeed estimation methods provided in the embodiment of the present application.

[0044] In addition, embodiments of the present application further provide a computer program product, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of any of the airspeed estimation methods provided in embodiments of the present application.

[0045] The embodiment of the present application can read the first state vector and target state control quantity of the target device at the current moment; input the first state vector and the target state control quantity into the dynamic model to obtain the second state vector at the next moment, and the dynamic model is constructed by the nonlinear relationship between the state vector and the state control quantity of the target device; determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector according to the unscented Kalman filtering method to obtain the target state vector; determine the windage speed according to the windage state parameters in the target state vector, and determine the estimated speed according to the speed state parameters in the target state vector; estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the windage speed. Therefore, this solution can first determine the state of the target device at the current moment and the control quantity of the input system, and then use the pre-built dynamic model to combine the current state and the control quantity to perform a nonlinear prediction of the state at the next moment, and obtain the second state of the target device at the next moment. Then, the second state is adjusted by the observed vector. Finally, the corresponding wind resistance speed and the estimated speed of the target device are determined according to the adjusted target state, and the difference between the estimated speed and the wind resistance speed is used as the airspeed of the target device at the next moment. In this way, the state of the target device at the next moment is adaptively calculated according to the state of the target device at the current moment through an iterative method, so as to calculate the airspeed of the target device, and the airspeed of the flying car can be accurately estimated, thereby improving the safety of the flying car. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0047] Figure 1 Schematic diagram of a scenario of an airspeed estimation system provided in an embodiment of the present application;

[0048] Figure 2 A schematic flow chart of the steps of the airspeed estimation method provided in an embodiment of the present application;

[0049] Figure 3 1 is a schematic flow chart of another step of the airspeed estimation method provided in an embodiment of the present application;

[0050] Figure 4 Schematic diagram of the process of constructing a kinetic model provided in the embodiment of the present application;

[0051] Figure 5 1 is a flow chart of the unscented Kalman filtering process provided by an embodiment of the present application;

[0052] Figure 6 Schematic diagram of the structure of the airspeed estimation device provided in an embodiment of the present application;

[0053] Figure 7 It is a structural diagram of the computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0055] The embodiments of the present application provide an airspeed estimation method, apparatus, device, and computer-readable storage medium. The embodiments of the present application will be described from the perspective of an airspeed estimation device. The airspeed estimation device can be integrated into a computer device, which can be a terminal device, specifically a terminal device carried on a means of transportation, i.e., a vehicle-mounted terminal, which can be a terminal on an aircraft. Furthermore, the terminal device can also be other types of devices, such as a television, a smartphone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a smart wearable device, and the like. Furthermore, but not limited to, the terminal device can be a television, a smartphone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a smart wearable device, and the like.

[0056] For example, see Figure 1 , which is a schematic diagram of a scenario for an airspeed estimation system provided in an embodiment of the present application. This system is suitable for use in scenarios involving real-time speed estimation of equipment and tools (such as flying cars and aircraft), and is not limited to applications involving airspeed estimation in real or simulated flight of aircraft / flying cars. This scenario includes a terminal or a server.

[0057] Specifically, the terminal can be a vehicle-mounted terminal, used to read the first state vector and target state control quantity of the target device at the current moment; input the first state vector and the target state control quantity into a dynamic model to obtain a second state vector at the next moment, and the dynamic model is constructed by a nonlinear relationship between the state vector and the state control quantity of the target device; determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector in an unscented Kalman filtering manner to obtain a target state vector; determine the windage speed according to the windage state parameters in the target state vector, and determine the estimated speed according to the speed state parameters in the target state vector; and estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the windage speed.

[0058] It should be noted that when the airspeed estimation system includes a server, a communication connection can be established between the vehicle-mounted terminal and the server. Using the aircraft as the target device, the vehicle-mounted terminal on board can read the target device's first state vector and target state control variable at the current moment and send these to the server. The server can then input the first state vector and target state control variable into a dynamic model to obtain a second state vector at the next moment. The dynamic model is constructed based on the nonlinear relationship between the target device's state vector and state control variable. The server then determines the observation vector corresponding to the second state vector and adjusts the second state vector using an unscented Kalman filter based on the observation vector to obtain a target state vector. The server then determines the windage velocity based on the windage state parameters in the target state vector and determines the estimated velocity based on the velocity state parameters in the target state vector. The target airspeed of the target device at the next moment is estimated based on the difference between the estimated velocity and the windage velocity. The target airspeed data is then sent to the vehicle-mounted terminal on the target device, enabling the target device to respond to the target airspeed data, such as by providing a voice announcement of the airspeed.

[0059] For example, taking the airspeed estimation of a flying car during flight as an example, first, the first state vector of the flying car at the current moment is read, which may include the roll angle δ, the heading angle Pitch angle θ, speed in three-dimensional coordinate direction Drag coefficient in three-dimensional coordinate direction In addition, the read state control quantity specifically includes the angular velocity (p, q, r) in the three-dimensional coordinate direction, the simulated wind force coefficient (w x 、w y 、w z ) etc. Then, the second state vector at the next moment is calculated based on the state vector and the state control quantity through a nonlinear dynamic model, and the corresponding observation vector or measurement vector of the flying car is determined. The second state vector is corrected according to the observation vector. Specifically, the second state vector can be adjusted by the Unscented Kalman Matrix (UKF) to obtain the adjusted target state vector. Finally, the windage coefficient of the flying car in each direction is read from the target state vector to determine the corresponding windage speed; and the speed state of the flying car in each direction is read from the target state vector to determine the corresponding estimated speed of the flying car in terms of power. Then, the airspeed of the flying car at the next moment is obtained by subtracting the windage speed from the estimated speed. The above is only an example and is not intended to be a limiting method for implementing this application.

[0060] It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments.

[0061] In the embodiment of the present application, the airspeed estimation device will be described from the perspective of the airspeed estimation device, and the airspeed estimation device can be integrated into a computer device such as a terminal device or a server. Figure 2 , Figure 2 This is a schematic flow chart of the steps of an airspeed estimation method provided in an embodiment of the present application. Taking a terminal device as an example, the terminal device is a terminal mounted on a vehicle. When the processor on the terminal device executes the program instructions corresponding to the airspeed estimation method, the specific process of the airspeed estimation method is as follows:

[0062] 101. Read the first state vector and target state control value of the target device at the current moment.

[0063] In an embodiment of the present application, in order to accurately estimate the vehicle's speed, the vehicle's state and control parameters at adjacent moments can be used to determine the speed. For example, the vehicle's state parameters (such as the operating state) at the previous moment are combined with the control parameters to adaptively infer the state parameters at the current moment. Alternatively, the vehicle's state parameters at the next moment are inferred based on the vehicle's state parameters at the current moment and the control parameters. The inferred state parameters can then be used to estimate the vehicle's speed at the current or next moment, eliminating the need for excessive reliance on speed-related sensors, thereby improving the accuracy of the vehicle's speed estimation.

[0064] The target device may be a moving vehicle, such as a flying vehicle, an aircraft, an aerial photography tool, etc. It is understood that the target device is equipped with an onboard terminal, which can be understood as a terminal with computing capabilities, and can calculate the status of adjacent moments in real time based on the device status and control quantity.

[0065] The first state vector can be a multidimensional array representing the target device, which can contain multiple different types of state parameters. For example, if a flying car is used as the target device, its corresponding state parameters may include speed, attitude angle, and drag coefficient. The drag coefficient can be understood as the resistance of the wind on the projection of the flying car. In addition, the state parameters can be decomposed according to the direction relationship of the three-dimensional space. For example, the direction of the xyz axis in the three-dimensional space coordinate system is used as an example. The speed is V, x 、V y 、V z , decompose the attitude angle into roll angle, pitch angle, and heading angle. Similarly, decompose the drag coefficient into the wind resistance coefficients in the x, y, and z directions, and construct a multi-dimensional state vector with the above state parameters.

[0066] The target state control quantity can be an input control quantity of the target device, and can be understood as a control parameter of the target device, which is used to control the target device to maintain or change its current state. For example, taking a flying car as the target device, in order to control the flying car to navigate according to a certain flight route, the control instructions input by the user can be read, or the control instructions in the automatic driving mode can be read and used as the state control quantity, or converted into the state control quantity. Exemplarily, taking the flying car as an example, the target state control quantity can also be in the form of a multidimensional array, and the control parameters in the array are related to the state parameters in the state vector. For example, the target state control quantity can include the angular rate of the attitude angle, the simulated wind coefficient, and the thrust coefficient. Similarly, the angular rate, the simulated wind coefficient, and the thrust coefficient can be decomposed according to the three-dimensional spatial direction, such as obtaining the angular rates corresponding to the roll angle, pitch angle, and heading angle, respectively, and obtaining the wind sub-coefficients w in the three directions respectively. x 、w y 、w z , respectively get the thrust coefficient T in three directions bx 、T by 、T bz , etc. In this way, the target device can be controlled to change the corresponding state through the corresponding state control parameters in the array, such as changing the attitude angle of the target device through the angular rate in the state control amount.

[0067] Specifically, in order to accurately estimate the real-time speed of a vehicle, the embodiments of the present application can determine the speed based on the vehicle's state and control variables at adjacent moments. Taking a flying car as an example, an array of the flying car's state parameters at the current moment, namely a first state vector, is obtained, and a control variable input by the flying car at the current moment, namely a target state control variable, is obtained. Subsequently, the various state parameters of the flying car at the next moment can be inferred based on the first state vector and the target state control variable to obtain the state of the flying car. This eliminates the need to rely solely on speed sensors, thereby improving the accuracy of estimating the vehicle's speed.

[0068] Through the above method, the real-time status and input control variables of the target device can be obtained, so that the state of the target device at the next moment can be inferred based on the current state and input control variables, thereby realizing the process of adaptive real-time calculation. There is no need to rely too much on sensors to measure speed, which improves the reliability of estimating the speed of the target device.

[0069] 102. Input the first state vector and the target state control variable into the dynamic model to obtain the second state vector at the next moment.

[0070] In an embodiment of the present application, in order to infer the state of the target device at the next moment based on the state of the target device at the current moment and the input control variables, it can be specifically achieved through a dynamic model. The dynamic model can be understood as the kinematic model corresponding to the target device, which can be used to combine the current state of the target device and the input control variables to infer the state of the target device at the next moment.

[0071] The dynamic model can be a nonlinear kinematic model of the target device, which is constructed by the nonlinear relationship between the state vector of the target device and the state control quantity, and can reflect the nonlinear relationship between the state of the target device in the motion mode and the control variable. It should be noted that during the motion process of the target device, due to the existence of some factors, the state conversion result between the actual state of the device and the input control variable is nonlinear. For example, the resistance factor of the motion environment on the target device, the conversion rate of the target device itself in the process of converting the control variable to power, and other factors will cause a nonlinear relationship between the state and the control quantity. In this regard, the embodiment of the present application constructs a nonlinear dynamic model by combining the state and control quantity of the target device for subsequent inference of the state of the target device during the motion process.

[0072] In the embodiments of the present application, the relationship between thrust and the operating rate of the power structure, and the relationship between speed and drag, can be considered to construct a nonlinear dynamic model based on these two relationships. For example, taking a flying car with a rotor assembly as an example, the rotor assembly is the propeller structure on the flying car (or aircraft). The flying car includes rotor assemblies in at least three spatial dimensions. For example, using the xyz three-dimensional coordinate system as an example, the flying car may include propeller assemblies on the x-axis, y-axis, and z-axis. In this case, the nonlinear dynamic model of the flying car can be constructed based on the two relationships between thrust and the rotational speed of the rotor assembly, and speed and drag. It is understood that in order to be able to infer the state of the target device based on the nonlinear dynamic model, the construction of the nonlinear dynamic model is generally completed before step 102.

[0073] In some embodiments, a nonlinear dynamic model can be constructed using historical motion data of the target device. For example, before step 102, the following steps can be included:

[0074] (102.A.1) Obtaining motion data of the target device at historical moments;

[0075] (102.A.2) Generate a state vector and a state control variable associated with the target device at a historical moment based on the motion data, where the state vector and the state control variable have a nonlinear correspondence relationship;

[0076] (102.A.3) Construct a dynamic model based on the nonlinear correspondence between the state vector and the state control variable.

[0077] The motion data refers to the historical operational data of the target device, reflecting its operational status. For example, in the case of a flying car, the motion data could be its test flight data, i.e., its driving data during the testing phase. This data could include thrust, wind resistance, speed, and other data, without limitation.

[0078] The state vector may be an array representing the multidimensional state of the target device during its historical operation, and may include multiple different types of state parameters. Furthermore, the state control variable may be the input control variable of the target device during historical operation, which can be understood as a control parameter of the target device used to control the target device to maintain or change its current state. It should be noted that details regarding the state vector and state control variable can be found in the aforementioned first state vector and target state control variable, respectively, and are not detailed here.

[0079] Specifically, to construct a dynamic model with a nonlinear relationship, the target device's historical motion data can be used to extract the state vector and state control variable of the target device at that historical moment. This allows the dynamic model to be constructed based on the state vector, state control variable, and the nonlinear relationship between the two. For example, taking a flying car as an example, flight test data of the flying car at a historical time (such as the test flight phase) can be obtained. Based on this test flight data, the state vector and state control variable with a nonlinear relationship can be extracted to construct the corresponding dynamic model of the flying car.

[0080] In some embodiments, the state vector may include state parameters representing the attitude angle, drag coefficient, and speed of the target device at the corresponding time, and the state control variable may include input control variables such as attitude angle change rate, wind force coefficient, and thrust coefficient. Then, step (102.A.2) may include:

[0081] (102.A.2.1) Extract the thrust coefficient, drag coefficient, and speed of the target device at the historical moment from the motion data;

[0082] (102.A.2.2) Decompose the thrust coefficient, drag coefficient, and speed in multiple directions according to the three-dimensional spatial relationship to obtain the thrust sub-coefficient, drag sub-coefficient, and directional speed of the target device in each direction;

[0083] (102.A.2.3) Determine the attitude angle corresponding to the target device based on the thrust coefficient, and determine the angular rate corresponding to the attitude angle in each direction according to the Euler angle transformation relationship;

[0084] (102.A.2.4) Construct the state vector of the target device at the historical moment based on the attitude angle, directional velocity and windage coefficient in each direction;

[0085] (102.A.2.5) Read the wind force coefficient associated with the wind resistance coefficient in the corresponding direction, and construct the state control quantity of the target device at the historical moment based on the angular rate, wind force coefficient, and thrust coefficient.

[0086] The three-dimensional spatial relationship may be a spatial relationship or a directional relationship of a three-dimensional coordinate system. Specifically, the three-dimensional directional relationship may be represented by the directions of the x, y, and z axes.

[0087] The thrust coefficient can be a coefficient representing the power of the target device. It can be understood as the power coefficient converted from the target device's power system and used to propel the target device. For example, taking a flying car as an example, the flying car may have a rotor assembly, i.e., a power structure. The rotor assembly can be arranged on the flying car according to a three-dimensional coordinate system. For example, rotor assemblies are arranged on the x, y, and z axes respectively. The rotation of these three rotors will be converted into the power of the flying car, i.e., thrust. Therefore, the relationship between the rotor speed and thrust can be used as a nonlinear factor. It should be noted that the thrust coefficient can be decomposed according to the x, y, and z axes to obtain the thrust sub-coefficients of the target device in the x, y, and z directions respectively.

[0088] It's understandable that because thrust is related to the rotational speed of the power structure (e.g., the rotor), the decomposed thrust coefficients are each related to the rotor power (rotational speed) in the corresponding direction of the target device. These power in different directions affects the target device's attitude angle in that direction. For example, the power (thrust) in the x-axis direction affects the target device's attitude angle in that direction, and similarly, the power in the y- and z-axis directions. It's important to note that, under the influence of the control variable thrust (power), the acceleration of the attitude angle affects the rate of change of the angle (angular velocity) in the corresponding direction. That is, the change in attitude angle is determined by the corresponding angular velocity, and the relationship between attitude angle and angular velocity is also nonlinear.

[0089] In addition, the speed of the target device can be decomposed in three dimensions to obtain the directional speeds in different directions. Taking the xyz three-dimensional coordinate system as an example, the directional speeds V on the x, y, and z axes can be obtained respectively. x 、V y 、V z , the sub-velocities in different directions all have nonlinear relationships with the drag coefficient and angular velocity. It should be noted that the drag coefficient is related to the wind force coefficient. The drag coefficient can be decomposed in three dimensions to obtain wind resistance sub-coefficients, which represent the state of wind speed acting on the target device in different directions. Similarly, the wind force coefficient can be decomposed in three dimensions to obtain wind force sub-coefficients.

[0090] Specifically, when extracting the state vector and state control variable of the corresponding target device at a historical moment from the motion data, the thrust coefficient, drag coefficient, speed, and other data of the target device can be extracted based on the motion data. The thrust coefficient, drag coefficient, and speed can be decomposed according to the directional relationship of the three-dimensional coordinate system to obtain the thrust sub-coefficient, wind resistance sub-coefficient, and directional velocity in each direction. The attitude angle corresponding to each thrust sub-coefficient in each direction and the angular velocity corresponding to each attitude angle can be determined. Then, the attitude angle, directional velocity, and wind resistance sub-coefficient corresponding to the three directions can be used as state parameters to construct the state vector of the target device at the historical time. In addition, the wind force sub-coefficient corresponding to each wind resistance sub-coefficient can be determined in a manner similar to wind simulation. The angular velocity, wind force sub-coefficient, and thrust sub-coefficient can be used as control variables, i.e., control parameters, to construct an array of state control parameters of the target device at the historical time, i.e., the state control variable. In this way, a state vector and state control variable with a nonlinear relationship are obtained. A nonlinear dynamic model can be constructed based on the state vector and state control variable with a nonlinear relationship.

[0091] In an embodiment of the present application, after obtaining the first state vector and target state control variable of the target device at the current moment, a dynamic model can be used to infer the state of the target device at the next moment based on the state of the target device at the current moment and the input control variable. Specifically, the first state vector and target state control variable can be input into the nonlinear dynamic model, and the dynamic model performs state inference based on the first state vector and the target state control variable to obtain the second state vector at the next moment. In this way, the state of the target device at the next moment can be reliably inferred using the nonlinear dynamic model.

[0092] Through the above method, the state of the target device at the next moment can be inferred by combining the nonlinear characteristics of the dynamic model with the target state vector and the target state control quantity at the current moment, so that the speed of the target device at the next moment can be estimated by using the nonlinear characteristics in the subsequent implementation without relying entirely on the data of the speed sensor, which is reliable.

[0093] 103. Determine an observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector.

[0094] In an embodiment of the present application, in order to improve the accuracy in estimating the speed of the target device at the next moment, after calculating the second state vector of the target device at the next moment, the state observation value of the target device at the next moment can be measured, and the observation value can be used to correct the second state vector through unscented Kalman filtering to improve the accuracy of the state vector expressing the state of the target device at the next moment, and thus, the accuracy of the estimated speed can be subsequently improved.

[0095] Among them, the observation vector can be a vector corresponding to multi-dimensional measurement data. Specifically, by measuring some state data of the target device, an observation vector is formed and used to correct the second state vector. Specifically, it can be processed by an unscented Kalman filter (UKF) to obtain a corrected target state vector.

[0096] The Unscented Kalman Filter (UKF) is a Kalman filter for nonlinear systems. It combines the unscented transform (UT) with the standard Kalman filter system. Through the unscented transform, the nonlinear system equations are adapted to the standard Kalman system under the linear assumption. It should be noted that the Unscented Kalman Filter has the characteristics of high estimation accuracy and good stability.

[0097] In some embodiments, the second state vector may be corrected by using an unscented Kalman filter. Specifically, in step 103, "adjusting the second state vector according to the observation vector using an unscented Kalman filter to obtain a target state vector" may include:

[0098] (103.1) Based on the unscented transformation, obtain multiple key points corresponding to the second state vector, and calculate the predicted state quantity corresponding to each key point based on the target state control quantity and the state quantity of each key point;

[0099] (103.2) Calculating the state variance of multiple key points based on the predicted state quantity and the state probability distribution relationship;

[0100] (103.3) Determine the target observation value corresponding to each predicted state quantity based on the observation vector, and calculate the observation value variance between multiple target observation values ​​according to the state probability distribution relationship;

[0101] (103.4) The second state vector is adjusted according to the difference between the state variance and the observation variance to obtain the target state vector.

[0102] Among them, the unscented transformation is the core of the unscented Kalman filter. The unscented transformation refers to sampling key points near the estimated point (the state point of the second state vector), and the key points obtained by sampling represent the probability distribution of the second state vector approximated by Gaussian density.

[0103] The key point may be a sampling point in the unscented Kalman filter processing process, namely, a sigma point.

[0104] Specifically, the above steps (103.1) to (103.5) are the processing process of the unscented Kalman filter. For ease of understanding, each step will be described in detail below, as follows:

[0105] It should be noted that before step (103.1), if this is the first time to perform the unscented Kalman filter process, the filter needs to be initialized. Subsequently, the state parameters at the current moment are used as the initial values ​​of the next moment and iterated. The filter initialization is expressed as follows:

[0106]

[0107]

[0108] 1. Regarding step (103.1), the main step is to perform an untraceable transformation on the second state vector. Specifically, step (103.1) may include: performing a nonlinear transformation on the second state vector to obtain a Gaussian distribution curve of the second state number; using the next moment as the state midpoint, symmetrically selecting key points on the Gaussian distribution curve to obtain multiple key points corresponding to the second state vector, each key point corresponding to a state quantity. In addition, a scaling parameter of the sampling point can be determined, and using the next moment as the state midpoint, symmetrically selecting key points on the Gaussian distribution curve according to the scaling parameter to obtain multiple key points corresponding to the second state vector.

[0109] For example, the symmetric sampling strategy is used to select key points, namely Sigma points, to obtain the Sigma point set {χ i}, i = 1, 2, ..., L, where L represents the number of sampling points and the parameter λ is a scaling parameter that determines the distance between the sampling point and the mean. The Sigma point set represents the Gaussian probability distribution after nonlinear transformation. Due to the symmetry of the Gaussian distribution, when selecting key points, the minimum number of sampling points is 3 points, one of which is the mean and the other two are distributed on both sides of the mean, so as to more accurately obtain the statistical characteristics of the system (mean and variance). In this way, through the unscented transformation in the unscented Kalman filter, sampling is performed near the Gaussian distribution of the state point of the second state vector to obtain multiple key points, sigma points, and determine the system state quantity of each key point. The system state quantity is expressed as follows:

[0110]

[0111] Among them, χ i Represents the system state quantity of the corresponding key point, P χ (k) represents the variance of the state at time k, Represents the mean value of the system state quantity at time k.

[0112] Furthermore, the step (103.1) of "calculating the predicted state quantity corresponding to each key point based on the target state control quantity and the state quantity of each key point" means predicting the state quantity from the current time k to time k+1 based on the system state quantity and input control quantity of the key point. The predicted state quantity can be specifically expressed as follows:

[0113] Y i (k+1|k)=f(χ i ,U(k)), where i=1...2L

[0114] Among them, Y i (k+1|k) represents the predicted state quantity at (k+1|k) based on each sampling point and the control input; i Represents the system state quantity of the i-th key point; U(k) represents the input control quantity, that is, the target state control quantity.

[0115] 2. Regarding step (103.2), it may include: (103.2.1) performing weighted summation on multiple predicted state quantities to obtain the state mean corresponding to multiple key points; (103.2.2) determining the state difference between each predicted state quantity and the state mean, and calculating the state variance corresponding to the multiple state differences according to the state probability distribution relationship.

[0116] The state mean refers to the mean of the predicted state quantities based on multiple key points. When calculating the state mean, it is necessary to calculate the mean of each predicted state quantity and the corresponding weight W. i Perform weighted summation to obtain the state mean. It should be noted that the weights to be multiplied for each predicted state quantity are inconsistent, but the sum of multiple weights can be 1.

[0117] Specifically, step (103.2.1) can be expressed as follows:

[0118]

[0119] Furthermore, the state variance between multiple predicted state quantities is calculated in (103.2.2), which can be specifically expressed as follows:

[0120]

[0121] Among them, P x (k+1|k) represents the state variance from time k to time k+1. i Indicates the weight corresponding to each predicted state quantity.

[0122] 3. Regarding step (103.3) of "determining the target observation value corresponding to each predicted state quantity based on the observation vector", it can be expressed as follows:

[0123] Vi (k+1|k)=H(Y i (k+1|k),U(k))i=1...2

[0124] Among them, V i (k+1|k) represents the target observation value, H(Y i (,) represents the observation function.

[0125] Furthermore, in step (103.3), "calculating the observation variance between the multiple target observations according to the state probability distribution relationship" may include: first, calculating the observation mean between the multiple target observations, specifically expressed as:

[0126]

[0127] Furthermore, the variance of the observation value is calculated based on the mean of the observation value, which can be expressed as follows:

[0128]

[0129] Among them, P zz (k+1|k) represents the variance of the observations.

[0130] Fourth, since the predicted state mean and variance at time k are not necessarily accurate, the mean and covariance between the predicted state quantity and the observed value can be further calculated based on the observed value at time k; then, in order to correct the predicted state mean and variance based on time k according to the gain calculated by the covariance, the final state mean probability P(k+1|k) and state variance probability P(k+1|k) are obtained.

[0131] Specifically, step (103.4) may include:

[0132] (103.4.1) According to the state probability distribution relationship, calculate the covariance between the state variance and the observation variance. It can be expressed as follows:

[0133]

[0134] Among them, the P xz (k+1|k) represents the covariance, Represents the transposed matrix.

[0135] (103.4.2) Determine the gain coefficient based on the ratio between the covariance and the variance of the observation value. Specifically, the gain coefficient of the unscented Kalman filter is determined based on the ratio between the covariance and the variance of the observation value, which is specifically expressed as follows:

[0136] L(k+1)=P xz (k+1|k)P zz (k+1|k)-1

[0137] Wherein, L(k+1) represents the gain coefficient of the state at time k+1.

[0138] (103.4.3) The second state vector is adjusted according to the gain coefficient to obtain an adjusted target state vector. Specifically, the system state equation and covariance matrix are updated according to the gain coefficient.

[0139] Among them, the updated system state variance equation is:

[0140]

[0141] Through the updated state variance equation, the state estimate at time k+1 can be calculated based on the gain of the unscented Kalman filter at time k, that is, the target state vector is obtained, which is specifically expressed as follows:

[0142]

[0143] Here, δ(k+1) represents the roll angle at time k+1, and θ(k+1) represents the pitch angle at time k+1. represents the navigation angle at time k+1, Indicates the direction velocity in the x-axis direction at time k+1, Indicates the direction velocity in the y-axis direction at time k+1, Indicates the direction velocity in the z-axis direction at time k+1, represents the windage coefficient in the x-axis direction at time k+1, represents the windage coefficient in the y-axis direction at time k+1, Indicates the windage coefficient in the z-axis direction at time k+1.

[0144] Among them, the updated covariance matrix is:

[0145] P x (k+1)=P x (k+1|k)-L(k+1)P zz (k+1|k)L(k+1) T

[0146] At this point, the unscented Kalman filter process is complete. The above process can be iterated to accurately represent the state of the target device at the next moment. It should be noted that the above "k moment" represents the "current moment" described in the embodiments of this application, and "k+1 moment" represents the "next moment" described in the embodiments of this application.

[0147] In order to improve the accuracy of estimating the speed of the target device at the next moment through the above method, the state observation value of the target device at the next moment can be measured and subjected to unscented Kalman filtering to improve the accuracy of the state vector expressing the state of the target device at the next moment.

[0148] 104. Determine a wind resistance speed according to the wind resistance state parameter in the target state vector, and determine an estimated speed according to the speed state parameter in the target state vector.

[0149] After obtaining the target state vector, the embodiment of the present application can read the grouped wind resistance state parameters in the target state vector to determine the wind resistance speed, and read the speed state parameters in the target state vector to determine the estimated speed of the target device, so as to subsequently calculate the final speed of the target device based on the estimated speed and the wind resistance speed.

[0150] The wind resistance state parameter may include wind resistance coefficients in multiple directions, such as wind resistance coefficients in the x, y, and z axis directions at the next moment. When calculating wind resistance speed, it can be expressed as:

[0151] The speed state parameter may include the directional speed in multiple directions, such as the directional speed in the x, y, and z axes at the next moment. When calculating the estimated speed, it can be expressed as: Vb=norm[VwxbVwybVwzb].

[0152] In the above manner, the windage speed and the estimated speed of the target device can be determined according to the relevant state parameters in the target state vector, so that the final speed of the target device can be calculated based on the estimated speed and the windage speed.

[0153] 105. Estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the wind resistance speed.

[0154] In the embodiments of the present application, after obtaining the estimated speed and wind resistance speed, the target device's final target airspeed can be obtained by subtracting the wind resistance speed from the estimated speed. For example, taking a flying car as the target device, the target state vector obtained through unscented Kalman filtering can accurately represent the target device's state at the next moment. Therefore, the target device's estimated speed and the wind resistance speed in the environment can be determined based on the relevant state parameters in the target state vector. Thus, by subtracting the wind resistance speed from the estimated speed, the target airspeed of the flying car at the next moment can be reliably estimated.

[0155] Through the above method, the state parameters of the vehicle at the next moment are calculated based on the state parameters and control parameters of the vehicle at the current moment; then, the vehicle speed at the current moment or the next moment is estimated based on the calculated state parameters, without relying too much on sensors related to measuring speed, thereby improving the accuracy and reliability of estimating the vehicle speed.

[0156] By implementing any one implementation or combination of implementations in the embodiments of the present application, an application scenario of the airspeed estimation process can be realized.

[0157] As can be seen from the above, the embodiment of the present application can read the first state vector and the target state control quantity of the target device at the current moment; input the first state vector and the target state control quantity into the dynamic model to obtain the second state vector at the next moment, and the dynamic model is constructed by the nonlinear relationship between the state vector and the state control quantity of the target device; determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector in an unscented Kalman filter manner to obtain

[0158] The target state vector is determined; the drag speed is determined based on the drag state parameters in the target state vector, and the estimated speed is determined based on the speed state parameters in the target state vector; and the target airspeed of the target device at the next moment is estimated based on the difference between the estimated speed and the drag speed. Therefore, this solution can first determine the state of the target device at the current moment and the control variable input to the system, and then use a pre-built dynamic model to combine the current state and the control variable to perform a nonlinear prediction of the state at the next moment, thereby obtaining the second state of the target device at the next moment. Then, the second state is adjusted based on the observed vector. Finally, the corresponding drag speed and the estimated speed of the target device are determined according to the adjusted target state, and the difference between the estimated speed and the drag speed is used as the airspeed of the target device at the next moment. In this way, the state of the target device at the next moment is adaptively calculated based on the current state of the target device through an iterative method, so as to calculate the airspeed of the target device, accurately estimating the airspeed of the flying car and improving the safety of the flying car.

[0159] The method described in the above embodiment will be further described in detail below with examples.

[0160] The present embodiment takes the airspeed estimation device as an example to further describe the airspeed estimation method provided by the present embodiment. Figure 3 This is another step flow chart of the airspeed estimation method provided in the embodiment of the present application. Figure 4 is a schematic diagram of the process of constructing a kinetic model provided in the embodiment of the present application, Figure 5 This is a flow chart of the unscented Kalman filtering process provided by the embodiment of the present application. Figure 3-5 Provide a description.

[0161] In the embodiments of this application, the airspeed estimation method will be described from the perspective of an airspeed estimation device. This airspeed estimation device can be integrated into a computer device such as an onboard terminal, which can be a terminal on a flying car. When the processor on the onboard terminal executes the program instructions corresponding to the data transmission method, the specific process of the airspeed estimation method is as follows:

[0162] 201. The vehicle-mounted terminal obtains movement data at historical moments.

[0163] The motion data refers to the historical operational data of the target device, reflecting its operational status. For example, in the case of a flying car, the motion data could be its test flight data, i.e., its driving data during the testing phase. This data could include thrust, wind resistance, speed, and other data, without limitation.

[0164] It should be noted that, taking a flying car as an example, the flying car may have a rotor assembly, or power structure, arranged on the flying car according to a three-dimensional coordinate system. For example, with rotor assemblies arranged on the x, y, and z axes, the rotation of these three rotors will be converted into the flying car's power, or thrust. The relationship between rotor speed and thrust can be used as a nonlinear factor. Therefore, the test flight data can be used to calibrate the relationship between thrust and speed. The calibration process is as follows: the relationship between rotor speed and thrust can be determined through bench testing. This involves giving the motor different speed commands, then measuring the thrust at different speeds using a pressure sensor, and finally calibrating using curve fitting or interpolation tables.

[0165] In addition, the drag coefficient of the flying car's fuselage is calibrated and corrected using test flight data. The specific process is: using wind tunnel testing as an example, wind is blown on the flying car in different directions and at different speeds. The drag is measured using pressure sensors, and the drag coefficient, such as the wind force coefficient, is calculated based on the air density and windward area in the drag formula.

[0166] 202. The vehicle-mounted terminal generates a state vector and a state control quantity associated with the target device at a historical moment based on the motion data.

[0167] The state vector can be an array representing the multidimensional state of the target device during its historical operation, and can contain multiple different types of state parameters. For example, if a flying car is used as the target device, its corresponding state parameters may include speed, attitude angle, and drag coefficient. The drag coefficient can be understood as the resistance of the wind on the projection of the flying car. In addition, the state parameters can be decomposed according to the direction relationship in three-dimensional space. For example, the direction of the xyz axis in the three-dimensional space coordinate system is used as an example, and the speed is V, x、V y 、V z , decompose the attitude angle into roll angle, pitch angle, and heading angle. Similarly, decompose the drag coefficient into the drag coefficients in the x, y, and z directions, and construct a multi-dimensional state vector with the above state parameters.

[0168] The state control quantity can be the input control quantity of the target device in the historical time, which can be understood as the control parameter of the target device, and is used to control the target device to maintain or change the current state. For example, taking a flying car as an example, the target state control quantity can also be in the form of a multidimensional array. The control parameters in the array are related to the state parameters in the state vector. For example, the target state control quantity may include the angular rate of the attitude angle, the simulated wind coefficient, and the thrust coefficient. Similarly, the angular rate, the simulated wind coefficient, and the thrust coefficient can be decomposed according to the three-dimensional spatial direction, such as obtaining the angular rates p, q, and r corresponding to the roll angle, pitch angle, and heading angle, respectively, and obtaining the wind sub-coefficients w in the three directions, respectively. x 、w y 、w z , respectively get the thrust coefficient T in three directions bx 、T by 、T bz , etc. In this way, the target device can be controlled to change the corresponding state through the corresponding state control parameters in the array, such as changing the attitude angle of the target device through the angular rate in the state control amount.

[0169] Specifically, when extracting the state vector and state control quantity of the corresponding target device at a historical moment from the motion data, the thrust coefficient, drag coefficient, speed and other data of the target device can be extracted respectively based on the motion data, and the thrust coefficient, drag coefficient and speed can be decomposed respectively according to the directional relationship of the three-dimensional coordinate system to obtain the thrust sub-coefficient, wind resistance sub-coefficient and directional speed in each direction, and the attitude angle corresponding to the thrust sub-coefficient in each direction and the angular velocity corresponding to each attitude angle can be determined. Then, the attitude angle, directional velocity and wind resistance sub-coefficient corresponding to the three directions are used as state parameters to construct the state vector of the target device at the historical time; in addition, the wind force sub-coefficient corresponding to each wind resistance sub-coefficient can be determined in accordance with the method of simulating wind force, and the angular velocity, wind force sub-coefficient and thrust sub-coefficient can be used as control variables, i.e., control parameters, to construct an array of state control parameters of the target device at the historical time, i.e., the state control quantity.

[0170] There is a nonlinear relationship between the state vector and the state control variable. For ease of understanding, let's take a flying car or aircraft as the target device. The motion data can be the test flight data of the flying car. The nonlinear relationship between the state vector and the state control variable is described below as follows:

[0171] (1) Since thrust is related to the rotational speed of the power structure (such as the rotor), the decomposed thrust coefficients are related to the rotor power (rotational speed) in the direction corresponding to the target device, as shown below: 2

[0173] Thrust=Kω

[0174] Among them, Thrust represents the thrust coefficient, ω represents the rotor speed, and K is a parameter calibrated according to test flight data.

[0175] (2) Decompose the thrust coefficient according to the x, y, and z axes to obtain the thrust coefficient T of the target device in the x, y, and z directions respectively. bx 、T by 、T bz , which is expressed as follows:

[0176]

[0177] Where δ is the roll angle, θ is the pitch angle, is the yaw angle.

[0178] (3) Since the thrust coefficient can be a coefficient representing the power of the target device, it can be understood as the power coefficient converted from the power system of the target device and used to propel the target device to move. Specifically, in the process of propulsion of the target device, the relationship between thrust, resistance, and acceleration needs to be considered, which can be specifically expressed as follows:

[0179]

[0180] Among them, Cross(,) represents cross product, is the rotation matrix from the navigation system to the aircraft system, a bx , a by , a bz They represent the acceleration in three directions of the aircraft system, m is the mass of the flying car, g is the acceleration due to gravity, is the projection of wind speed on the body axis, w b is the angular velocity of the three-dimensional body.

[0181] Among them, C dx , C dy , C dz is the drag coefficient calibrated by flight test data, ρ is the air density, S is the frontal area, then the system drag coefficient C D The specific expressions are as follows:

[0182]

[0183] in, It is the rotation matrix from the navigation system to the machine system, which can be understood as the transformation matrix formed by rotating around the x, y, and z axes of the target device. It is specifically expressed as follows:

[0184]

[0185] Combining the descriptions of (1), (2) and (3) above, according to the relationship between the rate of change of the Euler angle and the angular velocity, we can obtain:

[0186]

[0187] Among them, p, q, and r represent the angular rates of the roll angle, pitch angle, and heading angle, respectively.

[0188] Combining the descriptions of (1), (2) and (3) above, the speed change rate It can be expressed as:

[0189]

[0190] In this way, parameters such as attitude angle, directional speed, wind resistance coefficient, angular velocity, wind force coefficient, and thrust coefficient are obtained. Specifically, the attitude angle, directional speed, and wind resistance coefficient are used as state vectors, and the angular velocity, wind force coefficient, and thrust coefficient are used as state control quantities. Thus, a state vector and state control quantity with a nonlinear relationship are obtained.

[0191] 203. The vehicle-mounted terminal constructs a dynamic model based on the nonlinear correspondence between the state vector and the state control quantity.

[0192] The dynamic model is constructed from the nonlinear relationship between the state vector of the target device and the state control variable. This dynamic model can be a nonlinear kinematic model of the target device, which can reflect the nonlinear relationship between the state of the target device in the motion mode and the control variable. Specifically, during the motion of the target device, due to factors such as resistance and / or power conversion, the state transition between the actual state of the target device and the input control variable is nonlinear. Therefore, a nonlinear dynamic model is constructed by combining the state of the target device and the control variable for subsequent inference of the state of the target device during motion.

[0193] Specifically, a dynamic model can be constructed based on the nonlinear relationship of kinematics. The dynamic model can be represented by a discretized nonlinear equation f(), as follows:

[0194]

[0195] Where T is the interval of sampling time points, such as the time interval between the current moment and the next moment. It can be understood that the target device has a specific posture, speed and acceleration corresponding to each time interval; where w x 、w y 、w z It is Gaussian white noise, which can be understood as a state of wind resistance (wind force) that is conducive to Gaussian modified distribution simulation, specifically a Gaussian function.

[0196] Combined with the above nonlinear equation f(), without considering noise, let the current moment be moment k, and the next moment be moment k+1, so the nonlinear equation representing the second state vector can be expressed as follows:

[0197] X(k+1)=f(X(k),U(k))

[0198] Among them, X(k+1) represents the state estimation value at time k+1, that is, the second state vector, and U(k) represents the state control quantity.

[0199] Furthermore, considering the factor of system noise, the recursive equation of the state is determined, and the time interval T is used to describe the time from time k to time k+1. The nonlinear equation representing the motion state of the target device is:

[0200] X(k+1|k)=f(X(k),U(k))+F(k+1|k)W(k)

[0201] Where W(k) represents the system noise of the motion model, which is p-dimensional zero-mean white noise and represents the wind factor; F(k) is the interference input matrix; X(k+1|k) represents the state of the target device from time k to time k+1.

[0202] 204. The vehicle-mounted terminal reads the first state vector and the target state control variable at the current moment.

[0203] Specifically, the first state vector may be a multi-dimensional array representing the target device, which may include multiple state parameters of different types. The first state vector may be expressed as:

[0204]

[0205] Among them, δ(k) represents the roll angle at time k, θ(k) represents the pitch angle at time k, represents the navigation angle at time k, represents the velocity in the x-axis direction at time k, Indicates the direction velocity in the y-axis direction at time k, Indicates the direction velocity in the z-axis direction at time k, represents the windage coefficient in the x-axis direction at time k, represents the windage coefficient in the y-axis direction at time k, It represents the windage coefficient in the z-axis direction at time k.

[0206] Specifically, the target state control quantity can be an input control quantity of the target device, and can be understood as a control parameter of the target device, which is used to control the target device to maintain or change the current state. The target state control quantity can be expressed as:

[0207]

[0208] Where p(k) represents the angular rate of the roll angle at time k, q(k) represents the angular rate of the pitch angle at time k, and r(k) represents the angular rate of the heading angle at time k; w x (k) represents the wind force coefficient in the x-axis direction at time k, w y (k) represents the wind force coefficient in the y-axis direction at time k, w z (k) represents the wind force coefficient in the z-axis direction at time k; T bx (k) represents the thrust coefficient in the x-axis direction at time k, T by (k) represents the thrust coefficient in the y-axis direction at time k, T bz (k) represents the thruster coefficient in the z-axis direction at time k.

[0209] 205. The vehicle-mounted terminal inputs the first state vector and the target state control variable into the dynamic model to obtain the second state vector at the next moment.

[0210] In an embodiment of the present application, after obtaining the first state vector and target state control variable of the target device at the current moment, a dynamic model can be used to infer the state of the target device at the next moment based on the state of the target device at the current moment and the input control variable. Specifically, the first state vector and target state control variable can be input into the nonlinear dynamic model, and the dynamic model performs state inference based on the first state vector and the target state control variable to obtain the second state vector at the next moment. In this way, the state of the target device at the next moment can be reliably inferred using the nonlinear dynamic model.

[0211] 206. The vehicle-mounted terminal determines an observation vector corresponding to the second state vector.

[0212] In an embodiment of the present application, in order to improve the accuracy in estimating the speed of the target device at the next moment, after calculating the second state vector of the target device at the next moment, the state observation value of the target device at the next moment, i.e., the observation vector, can be measured.

[0213] Specifically, the observation vector can be a vector corresponding to multi-dimensional measurement data. Specifically, the observation vector is formed by measuring some state data of the target device to correct the second state vector. It should be noted that the observation vector can be calculated using the observation equation. Without considering noise, the observation equation can be expressed as:

[0214] Z(k+1)=H(X(k),U(k))

[0215] Here, Z(k+1) represents the observation vector, which represents the system observation value at time k+1, and H() represents the observation function.

[0216] Furthermore, considering the noise, the observation equation of the state can be expressed as:

[0217] Z(k+1)=H(X(k),I(k))+v(k)

[0218] Among them, ν(k) represents the observation noise generated during the motion process, which is also set to a Gaussian white noise sequence.

[0219] 207. The vehicle-mounted terminal adjusts the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector.

[0220] In an embodiment of the present application, after the observation vector is obtained, it can be processed by an unscented Kalman filter (UKF) to correct the second state vector using the observation vector to obtain a corrected target state vector.

[0221] Among them, the unscented Kalman filter (UKF) belongs to the Kalman filter of nonlinear systems. It is a combination of the unscented transform (UT) and the standard Kalman filter system. Through the lossless transform, the nonlinear system equation is adapted to the standard Kalman system under the linear assumption. The unscented Kalman filter has the characteristics of high estimation accuracy and good stability.

[0222] For ease of understanding, the following describes the processing of the observation vector and the second state vector in combination with the unscented Kalman filter, as follows:

[0223] (1) If it is the first time to perform the unscented Kalman filter process, the filter needs to be initialized, and then the state parameters at the current moment are used as the initial values ​​of the next moment and iterated.

[0224] Among them, the filter initialization is expressed as follows:

[0225]

[0226]

[0227] (2) Perform an untraceable transformation on the second state vector. Specifically, the untraceable transformation can be understood as selecting key points, i.e., Sigma points, using a symmetric sampling strategy. For example, the second state vector is re-simulated and mapped into 2n+1 points. The mapping method is a symmetric adoption strategy, and the Sigma point set {χ i}, i = 1, 2, …, L, where L represents the number of sampling points. At least three sampling points (key points) can be obtained. The parameter λ is a scaling parameter that determines the distance between the sampling point and the mean.

[0228] Through the unscented transformation in the unscented Kalman filter, sampling is performed near the Gaussian distribution of the state point of the second state vector to obtain multiple key points, sigma points, and determine the system state quantity of each key point. The system state quantity is expressed as follows:

[0229]

[0230] Among them, χ i Represents the system state quantity of the corresponding key point, P χ (k) represents the variance of the state at time k, Represents the mean value of the system state quantity at time k.

[0231] (3) Based on the system state quantity and input control quantity of the key point, the state quantity from the current time k to the time k+1 is predicted, that is, the predicted state quantity, which is specifically expressed as follows:

[0232] Y i (k+1|k)=f(χ i ,U(k)), where i=1...2L

[0233] Among them, Yi i (k+1|k) represents the predicted state from time k to time k+1, χ i Represents the system state quantity of the i-th key point; U(k) represents the input control quantity, that is, the target state control quantity.

[0234] Then, the mean of the predicted state quantities of multiple key points is calculated. When calculating the state mean, it is necessary to calculate the mean of each predicted state quantity and the corresponding weight W. i Perform weighted summation to obtain the state mean, which is specifically expressed as follows:

[0235]

[0236] in, Represents the state mean of multiple key points from time k to k+1, W i Represents the corresponding predicted state quantity and the corresponding weight.

[0237] Then, according to the state probability distribution relationship, the state variance of multiple predicted state quantities is calculated, which is specifically expressed as follows:

[0238]

[0239] Among them, P x (k+1|k) represents the state variance from time k to time k+1. i Indicates the weight corresponding to each predicted state quantity.

[0240] (4) Determine the state observation value from time k to time k+1, which is specifically expressed as follows:

[0241] V i (k+1|k)=H(Y i (k+1|k),U(k)) i=1...2

[0242] Among them, V i (k+1|k) represents the target observation value, H(Y i (,) represents the observation function.

[0243] Then, the mean value between multiple state observations is calculated as follows:

[0244]

[0245] Among them, the Represents the mean of the state observation values ​​of the target device at time k+1.

[0246] Furthermore, according to the state probability distribution relationship, the variance of the observation value is calculated according to the mean of the observation value, which can be expressed as follows:

[0247]

[0248] Among them, P zz (k+1|k) represents the variance of the observations.

[0249] Furthermore, according to the state probability distribution relationship, the covariance between the state variance and the observation variance is calculated, which is expressed as follows:

[0250]

[0251] Among them, the P xz (k+1|k) represents the covariance, Represents the transposed matrix.

[0252] (5) According to the covariance and the variance of the state observation value, the gain coefficient of the unscented Kalman filter is calculated, which is:

[0253] L(k+1)=P xz (k+1|k)P zz (k+1|k) -1

[0254] Wherein, L(k+1) represents the gain coefficient of the state at time k+1.

[0255] (6) The state variance equation is updated according to the gain coefficient, so that the state equation including the gain factor is specifically:

[0256]

[0257] And, the covariance matrix is ​​updated according to the gain coefficient, which is specifically expressed as follows:

[0258] P x (k+1)=P x (k+1|k)-L(k+1)P zz (k+1|k)L(k+1) T

[0259] At this point, the unscented Kalman filter process is completed. By iterating the above (1)-(6) processes, the iterative state variance equation is obtained when representing the target state vector at time k+1, which is as follows:

[0260]

[0261] Then, the wind resistance speed and the estimated speed of the target device can be calculated based on the wind resistance state parameter and the speed state parameter in the target state vector.

[0262] 208. The vehicle-mounted terminal determines the windage speed according to the windage state parameter in the target state vector, and determines the estimated speed according to the speed state parameter in the target state vector.

[0263] Specifically, the wind resistance state parameter may include wind resistance coefficients in multiple directions, such as wind resistance coefficients in the x, y, and z axis directions at the next moment. Vwzbk+1. When calculating windage speed, the details are as follows:

[0264]

[0265] in, Indicates the wind resistance speed of the target device at the next moment.

[0266] Specifically, the speed state parameter may include directional speeds in multiple directions, such as the directional speeds in the x, y, and z axes at the next moment. Vzbk+1. When calculating the estimated speed, it can be specifically expressed as follows:

[0267]

[0268] Among them, V b Indicates the speed of the target device at the next moment.

[0269] At this point, the speed and wind resistance speed of the target device at the next moment can be estimated, so that the final speed of the target device can be calculated based on the estimated speed and wind resistance speed.

[0270] 209. The vehicle terminal estimates the target airspeed at the next moment based on the difference between the estimated speed and the wind resistance speed.

[0271] In the embodiments of the present application, after obtaining the estimated speed and wind resistance speed, the target device's final target airspeed can be obtained by subtracting the wind resistance speed from the estimated speed. For example, taking a flying car as the target device, the target state vector obtained through unscented Kalman filtering can accurately represent the state of the target device at the next moment. Therefore, the estimated speed of the target device and the wind resistance speed in the environment can be determined based on the relevant state parameters in the target state vector. Thus, by subtracting the wind resistance speed from the estimated speed, the target airspeed of the flying car at the next moment can be estimated, as specifically expressed as follows:

[0272]

[0273] Among them, V airspeed It indicates the estimated actual true airspeed of the target device, that is, the target airspeed.

[0274] By executing the above steps, a nonlinear dynamic model of a multi-rotor flying car can be established. Based on the sensor fusion information and the sensitivity of aerodynamic drag to airspeed, an untraceable transformation is then used. Through regular sampling and weighting, the estimation error caused by linear approximation is reduced, and more accurate airspeed estimation can be achieved, thereby enhancing flight safety.

[0275] As can be seen from the above, the embodiment of the present application can first determine the state of the target device at the current moment and the control quantity of the input system, and use a pre-constructed dynamic model in combination with the current state and the control quantity to perform a nonlinear prediction of the state at the next moment, thereby obtaining a second state of the target device at the next moment. Then, the second state is adjusted by the observed vector. Finally, the corresponding wind resistance speed and the estimated speed of the target device are determined according to the adjusted target state, and the difference between the estimated speed and the wind resistance speed is used as the airspeed of the target device at the next moment. In this way, the state of the target device at the next moment is adaptively calculated according to the state of the target device at the current moment through an iterative method, so as to calculate the airspeed of the target device, and the airspeed of the flying car can be accurately estimated, thereby improving the safety of the flying car.

[0276] The implementation process or description of the above embodiment is identical or similar to the description of the previous implementation. For details, please refer to the description of the previous implementation and no further details will be given here.

[0277] In order to better implement the above method, an embodiment of the present application further provides an airspeed estimation device, which can be integrated into a computer device, such as a vehicle-mounted terminal or other computer device.

[0278] For example, Figure 6 As shown, the airspeed estimation device may include a reading unit 601 , an input unit 602 , an adjustment unit 603 , a determination unit 604 and an estimation unit 605 .

[0279] A reading unit 601 is used to read the first state vector and target state control value of the target device at the current moment;

[0280] An input unit 602 is configured to input the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed based on a nonlinear relationship between the state vector of the target device and the state control variable;

[0281] An adjustment unit 603 is configured to determine an observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector;

[0282] a determination unit 604, configured to determine a windage speed according to the windage state parameter in the target state vector, and to determine an estimated speed according to the speed state parameter in the target state vector;

[0283] The estimation unit 605 is configured to estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the windage speed.

[0284] In some embodiments, the airspeed estimation device also includes a construction unit for: obtaining motion data of the target device at a historical moment; generating a state vector and a state control quantity associated with the target device at the historical moment based on the motion data, and there is a nonlinear correspondence between the state vector and the state control quantity; and constructing a dynamic model based on the nonlinear correspondence between the state vector and the state control quantity.

[0285] In some embodiments, the construction unit is further used to: extract the thrust coefficient, drag coefficient and speed of the target device at the historical moment from the motion data; decompose the thrust coefficient, drag coefficient and speed in multiple directions according to the three-dimensional spatial relationship, and obtain the thrust sub-coefficient, wind resistance sub-coefficient and directional speed of the target device in each direction; determine the attitude angle corresponding to the target device according to the thrust sub-coefficient, and determine the angular velocity corresponding to the attitude angle in each direction according to the Euler angle transformation relationship; construct the state vector of the target device at the historical moment according to the attitude angle, directional velocity and wind resistance sub-coefficient in each direction; read the wind force sub-coefficient associated with the wind resistance sub-coefficient in the corresponding direction, and construct the state control quantity of the target device at the historical moment according to the angular velocity, wind force sub-coefficient and thrust sub-coefficient.

[0286] In some embodiments, the adjustment unit 603 is also used to: obtain multiple key points corresponding to the second state vector based on the untraceable transformation, and calculate the predicted state quantity corresponding to each key point based on the target state control quantity and the state quantity of each key point; calculate the state variance of multiple key points according to the state probability distribution relationship based on the predicted state quantity; determine the target observation value corresponding to each predicted state quantity according to the observation vector, and calculate the observation value variance between multiple target observation values ​​according to the state probability distribution relationship; adjust the second state vector according to the difference between the state variance and the observation value variance to obtain the target state vector.

[0287] In some embodiments, the adjustment unit 603 is further used to: perform a nonlinear transformation on the second state vector to obtain a Gaussian distribution curve of the second state number; take the next moment as the midpoint of the state, symmetrically select key points on the Gaussian distribution curve to obtain multiple key points corresponding to the second state vector, and each key point corresponds to a state quantity.

[0288] In some embodiments, the adjustment unit 603 is further used to: perform weighted summation on multiple predicted state quantities to obtain state means corresponding to multiple key points; determine the state difference between each predicted state quantity and the state mean; and calculate the state variance corresponding to multiple state differences according to the state probability distribution relationship.

[0289] In some embodiments, the adjustment unit 603 is further used to: calculate the covariance between the state variance and the observation value variance according to the state probability distribution relationship; determine the gain coefficient according to the ratio between the covariance and the observation value variance; and adjust the second state vector according to the gain coefficient to obtain the adjusted target state vector.

[0290] As can be seen from the above, the embodiment of the present application can first determine the state of the target device at the current moment and the control quantity of the input system, and use a pre-constructed dynamic model in combination with the current state and the control quantity to perform a nonlinear prediction of the state at the next moment, thereby obtaining a second state of the target device at the next moment. Then, the second state is adjusted by the observed vector. Finally, the corresponding wind resistance speed and the estimated speed of the target device are determined according to the adjusted target state, and the difference between the estimated speed and the wind resistance speed is used as the airspeed of the target device at the next moment. In this way, the state of the target device at the next moment is adaptively calculated according to the state of the target device at the current moment through an iterative method, so as to calculate the airspeed of the target device, and the airspeed of the flying car can be accurately estimated, thereby improving the safety of the flying car.

[0291] The specific implementation of the above operations can be found in the previous embodiments and will not be described in detail here.

[0292] The present application also provides a computer device, such as Figure 7 , which shows a schematic diagram of the structure of the computer device involved in the embodiment of the present application, specifically:

[0293] The computer device may include one or more processing core processors 701, one or more computer readable storage media memories 702, a power supply 703, an input unit 704 and other components. Those skilled in the art will understand that Figure 7 The computer device structure shown in the figure does not constitute a limitation on the computer device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently.

[0294] Processor 701 is the control center of the computer device. It connects the various components of the entire computer device using various interfaces and lines. By running or executing software programs and / or modules stored in memory 702 and accessing data stored in memory 702, it performs various functions of the computer device and processes data, thereby providing overall monitoring of the computer device. Optionally, processor 701 may include one or more processing cores; preferably, processor 701 may integrate an application processor and a modem processor, wherein the application processor primarily processes the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into processor 701.

[0295] The memory 702 can be used to store software programs and modules. The processor 701 executes various functional applications and airspeed estimation by running the software programs and modules stored in the memory 702. The memory 702 may mainly include a program storage area and a data storage area. The program storage area may store an operating system, at least one application required for a function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 702 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 702 may also include a memory controller to provide the processor 701 with access to the memory 702.

[0296] The computer device also includes a power supply 703 for supplying power to various components. Preferably, the power supply 703 can be logically connected to the processor 701 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The power supply 703 can also include one or more DC or AC power supplies, a recharging system, a power failure detection circuit, a power converter or inverter, a power status indicator, and other arbitrary components.

[0297] The computer device may further include an input unit 704, which may be configured to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0298] Although not shown, the computer device may further include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 701 in the computer device will load the executable files corresponding to one or more application processes into the memory 702 according to the following instructions, and the processor 701 will run the application stored in the memory 702 to implement various functions as follows:

[0299] Read the first state vector and target state control quantity of the target device at the current moment; input the first state vector and the target state control quantity into a dynamic model to obtain a second state vector at the next moment, wherein the dynamic model is constructed by a nonlinear relationship between the state vector and the state control quantity of the target device; determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector; determine the windage speed based on the windage state parameters in the target state vector, and determine the estimated speed based on the speed state parameters in the target state vector; and estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the windage speed.

[0300] The specific implementation of the above operations can be found in the previous embodiments and will not be described in detail here.

[0301] As can be seen from the above, the embodiment of the present application can first determine the state of the target device at the current moment and the control quantity of the input system, and use a pre-constructed dynamic model to combine the current state and the control quantity to perform a nonlinear prediction of the state at the next moment, thereby obtaining a second state of the target device at the next moment. Then, the second state is adjusted by the observed vector. Finally, the corresponding wind resistance speed and the estimated speed of the target device are determined according to the adjusted target state, and the difference between the estimated speed and the wind resistance speed is used as the airspeed of the target device at the next moment. In this way, the state of the target device at the next moment is adaptively calculated according to the state of the target device at the current moment through an iterative method, so as to calculate the airspeed of the target device, thereby accurately estimating the airspeed of the flying car and improving the safety of the flying car.

[0302] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be accomplished by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0303] To this end, an embodiment of the present application provides a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the steps of any of the airspeed estimation methods provided in the embodiments of the present application. For example, the instructions can execute the following steps:

[0304] Read the first state vector and target state control quantity of the target device at the current moment; input the first state vector and the target state control quantity into a dynamic model to obtain a second state vector at the next moment, wherein the dynamic model is constructed by a nonlinear relationship between the state vector and the state control quantity of the target device; determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector; determine the windage speed based on the windage state parameters in the target state vector, and determine the estimated speed based on the speed state parameters in the target state vector; and estimate the target airspeed of the target device at the next moment based on the difference between the estimated speed and the windage speed.

[0305] The specific implementation of the above operations can be found in the previous embodiments and will not be repeated here.

[0306] The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0307] The present application also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the airspeed estimation method provided in various optional implementations of the above embodiments.

[0308] Since the instructions stored in the computer-readable storage medium can execute the steps in any of the airspeed estimation methods provided in the embodiments of the present application, the beneficial effects that can be achieved by any of the airspeed estimation methods provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0309] The above is a detailed introduction to an airspeed estimation method, device, equipment and computer-readable storage medium provided in the embodiments of the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present application, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present application.

Claims

1. A method for estimating airspeed, characterized in that: include: Read the first state vector and target state control quantity of the target device at the current moment; Inputting the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed by a nonlinear relationship between the state vector and the state control variable of the target device; Determining an observation vector corresponding to the second state vector, and adjusting the second state vector according to the observation vector using an unscented Kalman filter method to obtain a target state vector; determining a windage speed according to the windage state parameter in the target state vector, and determining an estimated speed according to the speed state parameter in the target state vector; estimating a target airspeed of the target device at a next moment based on a difference between the estimated speed and the wind resistance speed; The step of adjusting the second state vector according to the observation vector in an unscented Kalman filtering manner to obtain a target state vector includes: Based on an unscented transformation, a plurality of key points corresponding to the second state vector are obtained, and a predicted state quantity corresponding to each key point is calculated according to the target state control quantity and the state quantity of each key point; Calculating the state variance of multiple key points according to the predicted state quantity and the state probability distribution relationship; Determine the target observation value corresponding to each predicted state quantity according to the observation vector, and calculate the observation value variance between the multiple target observation values ​​according to the state probability distribution relationship; The second state vector is adjusted according to the difference between the state variance and the observation value variance to obtain a target state vector.

2. The method according to claim 1, characterized in that Before inputting the first state vector and the target state control variable into the dynamics model, the method further includes: Obtaining motion data of the target device at historical moments; generating, based on the motion data, a state vector and a state control variable associated with the target device at a historical moment, wherein the state vector and the state control variable have a nonlinear corresponding relationship; A dynamic model is constructed based on the nonlinear correspondence between the state vector and the state control quantity.

3. The method according to claim 2, characterized in that The step of generating, based on the motion data, a state vector and a state control quantity associated with the target device at a historical moment, includes: extracting the thrust coefficient, drag coefficient, and speed of the target device at a historical moment from the motion data; Decomposing the thrust coefficient, wind resistance coefficient, and speed in multiple directions according to the three-dimensional spatial relationship to obtain the thrust sub-coefficient, wind resistance sub-coefficient, and directional speed of the target device in each direction; Determine the attitude angle corresponding to the target device according to the thruster coefficient, and determine the angular rate corresponding to the attitude angle in each direction according to the Euler angle transformation relationship; Constructing a state vector of the target device at a historical moment according to the attitude angle, directional speed and windage coefficient in each direction; The wind sub-coefficient associated with the wind resistance sub-coefficient in the corresponding direction is read, and the state control quantity of the target device at the historical moment is constructed according to the angular velocity, the wind sub-coefficient and the thrust sub-coefficient.

4. The method according to claim 1, wherein The obtaining, based on the unscented transformation, a plurality of key points corresponding to the second state vector includes: performing a nonlinear transformation on the second state vector to obtain a Gaussian distribution curve of the second state vector; Taking the next moment as the state midpoint, key points are symmetrically selected on the Gaussian distribution curve to obtain multiple key points corresponding to the second state vector, each key point corresponding to a state quantity.

5. The method according to claim 1, wherein The step of calculating the state variance of the plurality of key points according to the predicted state quantity and the state probability distribution relationship includes: Performing weighted summation on the plurality of predicted state quantities to obtain state means corresponding to the plurality of key points; Determining a state difference between each predicted state quantity and the state mean; According to the state probability distribution relationship, the state variance corresponding to the multiple state difference values ​​is calculated.

6. The method according to claim 1, characterized in that The step of adjusting the second state vector according to the difference between the state variance and the observation value variance to obtain a target state vector includes: Calculating the covariance between the state variance and the observation value variance according to the state probability distribution relationship; Determining a gain coefficient according to a ratio between the covariance and the observed value variance; The second state vector is adjusted according to the gain coefficient to obtain an adjusted target state vector.

7. An airspeed estimation device, characterized in that: include: A reading unit, configured to read a first state vector and a target state control variable of a target device at a current moment; an input unit, configured to input the first state vector and the target state control variable into a dynamic model to obtain a second state vector at a next moment, wherein the dynamic model is constructed based on a nonlinear relationship between the state vector and the state control variable of the target device; An adjustment unit is used to determine the observation vector corresponding to the second state vector, and adjust the second state vector according to the observation vector in an unscented Kalman filter manner to obtain a target state vector; the adjustment unit is also used to obtain multiple key points corresponding to the second state vector based on an unscented transformation, and calculate the predicted state quantity corresponding to each key point according to the target state control quantity and the state quantity of each key point; calculate the state variance of multiple key points according to the state probability distribution relationship based on the predicted state quantity; determine the target observation value corresponding to each predicted state quantity according to the observation vector, and calculate the observation value variance between the multiple target observation values ​​according to the state probability distribution relationship; adjust the second state vector according to the difference between the state variance and the observation value variance to obtain a target state vector; a determining unit, configured to determine a windage speed according to the windage state parameter in the target state vector, and to determine an estimated speed according to the speed state parameter in the target state vector; An estimating unit is configured to estimate a target airspeed of the target device at a next moment based on a difference between the estimated speed and the windage speed.

8. A computer device, characterized in that: The invention comprises a processor and a memory, wherein the memory stores a computer program, and the processor is used to run the computer program in the memory to implement the steps of the airspeed estimation method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium is computer-readable and stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor to execute the steps of the airspeed estimation method according to any one of claims 1 to 6.

Images