A method and computing device for determining an attitude of an aircraft

Abstract

The embodiment of the specification provides a method and a computing device for determining an aircraft attitude, which involves a plurality of sensors arranged at target positions of a wing, a fuselage and a rotor platform of the aircraft, local measurement values of attitude change rates of each target position are determined according to output data of the sensors; a weight corresponding to each target position is determined according to a vibration state of each target position, the vibration state is determined according to the local measurement values, the local measurement values are weighted and fused according to the weights to determine a comprehensive measurement value, the aircraft attitude at a current time is determined by using a Kalman filtering method according to the comprehensive measurement value and a predicted state of the aircraft attitude at the current time, a deployment scheme of arranging the sensors at the target positions to form an observation network is proposed, and a method of fusing the measurement values of the target positions is further proposed on the basis, and the obtained comprehensive measurement value can reduce the interference of local deformation and obtain a relatively accurate aircraft attitude.

Description

Technical Field

[0001] The embodiments in this specification belong to the field of data processing technology, and more specifically, relate to a method and computing device for determining the attitude of an aircraft. Background Technology

[0002] Aircraft attitude describes the rotational relationship between the aircraft's body coordinate system and the Earth's coordinate system or inertial coordinate system. Typically, the aircraft's real-time angular velocity, linear acceleration, and magnetic field direction can be obtained through sensors, and then the aircraft's attitude can be calculated through integration and other methods.

[0003] In large eVTOL (electric vertical takeoff and landing) aircraft, especially those employing compound wing or tiltrotor configurations, lightweight materials such as carbon fiber are often used to reduce weight. This results in lower overall structural rigidity, making the airframe prone to deformation and oscillation during flight. Consequently, the measurement of the aircraft's attitude is affected by localized deformation, and the measurement results often fail to reflect the true attitude of the aircraft.

[0004] Therefore, there is an urgent need for a technical solution to determine the attitude of an aircraft, in order to at least partially solve the above-mentioned technical problems. Summary of the Invention

[0005] The embodiments in this specification are intended to provide a method and computing device for determining the attitude of an aircraft.

[0006] This specification provides, in one aspect, a method for determining the attitude of an aircraft, the method involving multiple sensors deployed at various target locations of the aircraft, including wings, fuselage, and rotor platform, the method comprising:

[0007] Based on the output data of the multiple sensors at the current moment, determine the local measurement value of the attitude change rate for each target part;

[0008] Based on the vibration state of each target part, the weight corresponding to each target part is determined. The vibration state is determined based on the local measurement values ​​of the target part at the current time and at each historical time.

[0009] Based on the weight corresponding to each target part, the local measurement values ​​are weighted and fused to determine the comprehensive measurement value of the attitude change rate at the current moment;

[0010] Based on the comprehensive measurement values ​​and the predicted state of the aircraft attitude at the current moment, the Kalman filter method is used to determine the aircraft attitude at the current moment, wherein the predicted state at the current moment is calculated based on the aircraft attitude determined at the previous moment.

[0011] A second aspect of this specification provides a computer-readable storage medium having a computer program stored thereon that, when executed in a computer, causes the computer to perform the method described in the first aspect.

[0012] A third aspect of this specification provides a computing device including a memory and a processor, wherein the memory stores executable code, and the processor, when executing the executable code, implements the method described in the first aspect.

[0013] This specification provides a fourth aspect of a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method as described in the first aspect.

[0014] The technical solution for determining the attitude of an aircraft provided in the embodiments of this specification proposes a deployment scheme in which sensors are deployed at each target location to form an observation network. Based on this, a method for fusing the measurement values ​​of each target location is further proposed. The resulting comprehensive measurement value can reduce the interference of local deformation and obtain a more accurate aircraft attitude. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments in this specification, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a technical concept for determining the attitude of an aircraft in one embodiment of this specification;

[0017] Figure 2 This is a flowchart illustrating a method for determining an aircraft attitude according to one embodiment of this specification;

[0018] Figure 3 This is a schematic diagram of the sensor deployment location in one embodiment of this specification;

[0019] Figure 4 This is a schematic diagram of the wavelet packet decomposition process in one embodiment of this specification. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0021] Determining the aircraft's attitude is a crucial step in the aircraft control process. Typically, during aircraft control, the aircraft's attitude, as part of the aircraft's state, is used along with various data such as the aircraft's position and speed in the current control cycle as input to the flight control rate, which then outputs control commands for the next control cycle.

[0022] Ideally, the aircraft attitude input to the flight control law should represent the overall attitude of the aircraft (which can also be understood as the attitude of the aircraft's center of gravity). Therefore, existing technologies typically deploy sensors near the aircraft's center of gravity to determine the aircraft attitude based on their output data. Taking an Inertial Measurement Unit (IMU) as an example, an IMU typically consists of a gyroscope, an accelerometer, and a magnetometer. Thus, when the sensor deployed on the aircraft is an IMU, the output data includes the angular velocity determined by the gyroscope, the linear acceleration determined by the accelerometer, and the direction of the magnetic field. The direction of the magnetic field allows the angular velocity and linear acceleration to be transformed from a coordinate system centered on the IMU to the inertial frame. Furthermore, by integrating the angular velocity and linear acceleration and combining this with the aircraft attitude from the previous control cycle, the aircraft attitude for the current control cycle can be obtained.

[0023] However, due to the inherent instability of the aircraft's structure (as mentioned earlier, deformation and oscillation are prone to occur during flight), on the one hand, the aircraft's center of gravity becomes unstable due to deformation disturbances, leading to an unstable relative position between the sensor and the center of gravity; on the other hand, the sensor's output data is also susceptible to oscillations, containing a large amount of noise. Consequently, the aircraft attitude obtained through existing methods is difficult to reflect the true aircraft attitude.

[0024] Figure 1 A schematic diagram illustrating a technical concept for determining an aircraft attitude according to one embodiment of this specification is shown. Figure 1As shown, this embodiment first identifies the parts of the aircraft that best reflect the true aircraft attitude under different flight conditions as target parts. Then, sensors are deployed at each target part, and the local measurement value of the aircraft attitude change rate at each target part is determined based on the sensor output data. Next, based on the local measurement value, the vibration state corresponding to each target part is determined. This vibration state also reflects the reliability of the local measurement value for each target part. Therefore, the weights corresponding to each target part are determined based on the vibration state. The local measurement values ​​are then fused using these weights to obtain a comprehensive measurement value that includes the attitude information of each target part while minimizing noise interference. Finally, this comprehensive measurement value is used to determine the aircraft attitude.

[0025] Based on the above technical concept. Figure 2 A flowchart illustrating a method for determining the attitude of an aircraft according to an embodiment of this specification is shown. The method involves multiple sensors deployed at various target locations of the aircraft, including the wings, fuselage, and rotor platform.

[0026] Figure 3 A schematic diagram showing the deployment location of the sensor in one embodiment of this specification is illustrated. For example... Figure 3 As shown in the diagram, area A represents the fuselage. Due to its relatively stable state and proximity to the aircraft's center of gravity, the fuselage is the most common sensor deployment location in current technology. Area B represents the wing. Due to its material and shape, the wing is a deformation-sensitive part of the aircraft. Deploying sensors on the wing can capture the aircraft's deformation state. In some other aircraft models, the arms are also deformation-sensitive parts and can be used as sensor deployment locations. Area C represents the rotor platform. The rotor platform is close to the vibration source of the aircraft—the rotor—and deploying sensors on the rotor platform can capture the aircraft's vibration state.

[0027] It should be noted that the number of sensors does not correspond one-to-one with the number of target locations. On the one hand, several sensors can be deployed for any given target location; on the other hand, target locations such as wings and rotor platforms may be located in more than one place within the aircraft. Figure 3 The aircraft shown has four rotor platforms, and several sensors can be deployed on each rotor platform.

[0028] This embodiment provides a specific sensor deployment scheme. Multiple IMUs are installed at each target location—the rotor platform, wings and arms, and fuselage (each rotor platform is considered an independent target location, and the wings and arms are treated similarly)—to form a space observation network. Specifically, 2-3 IMUs can be deployed at each target location to reduce the impact of a single IMU failure on the subsequent aircraft attitude determination process.

[0029] For the IMU located on the rotor platform, since it is close to the vibration source, it can be electromagnetically shielded and encapsulated, and a silicone shock absorber can be added to suppress the interference of high-frequency noise >200Hz on the IMU.

[0030] In addition, hardware-triggered synchronization methods, such as pulses per second (PPS), can be used to eliminate data delays during dynamic motion, ensuring that the sampling time deviation of all IMUs is <1ms. Furthermore, the IEEE 1588 protocol can be used to correct transmission delays, further improving the timestamp accuracy to ±10μs. This ensures that the time error between the output data collected by each sensor at the "same moment" meets user requirements under most conditions.

[0031] In this embodiment, the method for determining the aircraft's attitude can be executed by a computing device deployed on the aircraft or by the aircraft's control system. It is only necessary to ensure that the aircraft attitude obtained using this method can be applied in real-time to the determination of the aircraft's control commands; this specification does not impose any limitations here. The following only considers the computing device as the execution entity, and... Figure 2 The method shown is described below, and includes:

[0032] Step S201: Based on the output data of the multiple sensors at the current moment, determine the local measurement value of the attitude change rate for each target part.

[0033] First, the computing device acquires the output data from each sensor deployed on the aircraft at the current moment. As mentioned earlier, the sensor output data cannot directly represent the aircraft's attitude change rate. Therefore, the sensor output data needs to be processed to determine the local measurement value of the attitude change rate for each target location. The current moment is the time when this process is executed. Figure 2 The timing of the method shown.

[0034] Specifically, for any target location, the process of determining the local measurement of the attitude change rate may include a data conversion step and a data aggregation step.

[0035] In the data conversion step, for any sensor, the output data measured by the sensor may include the angular velocity and linear acceleration with the sensor itself as the center of the coordinate system. Therefore, the aforementioned angular velocity and linear acceleration need to be converted to the inertial frame according to the direction of the magnetic field measured by the sensor, and the triaxial angular velocity and triaxial acceleration in the corresponding inertial frame of the sensor are determined.

[0036] Thus, the output data of each sensor is converted to the inertial frame, which completes the data conversion step, and then the data aggregation step can be performed.

[0037] In the data aggregation step, the converted three-axis angular velocities and three-axis accelerations from various sensors deployed at the same target location are fused to determine the local measurement value of the attitude change rate for that target location. The aforementioned three axes are the pitch axis, roll axis, and yaw axis.

[0038] In some implementations, fault detection can be performed on each sensor before the data aggregation step. This allows the data aggregation step to fuse only the converted triaxial angular velocities and triaxial accelerations from the sensors that are not faulty at any given target location. Specifically, angular velocities corresponding to the same axis are fused together, and accelerations corresponding to the same axis are fused together.

[0039] To more accurately identify faulty sensors, this embodiment provides a fault detection method—preset the expected value and standard deviation of the output data for each target part. When the difference between any item in the output data of any sensor and the expected value is greater than n times the standard deviation, it can be determined that the sensor has failed.

[0040] The expected value and standard deviation of the output data for each target part can be determined through pre-conducted flight experiments on the aircraft. During the flight experiments, sensors can be used to collect output data from each target part under normal operating conditions. The collected output data is then processed to determine the expected value and standard deviation for each target part. The aforementioned parameter n can be set by the implementer; a smaller value indicates more stringent fault detection. Typically, n is set to 3.

[0041] Furthermore, for any target location, if all sensors at that target location malfunction, the local measurement values ​​of the target location can be interpolated using the local measurement values ​​of the sensors at several target locations closest to the target location, based on the pre-established rigid-flexible hybrid dynamics model of the aircraft.

[0042] Specifically, the interpolation process can be referenced using the following formula:

[0043] (1)

[0044] in, This is the mass matrix of the target region; This is the stiffness matrix of the target location; The damping matrix of the target part is given. The parameters in the aforementioned mass matrix, stiffness matrix, and damping matrix can all be determined through observation experiments on the aircraft. The external force acting on the target location can be determined based on local measurements of several target locations closest to the target location. and These represent the elastic deformation degrees of freedom (describing the local deformation of a flexible body relative to its undeformed state, such as bending, stretching, or vibrational displacement) and rigid body motion degrees of freedom (the rigid body motion of the entire structure, such as rotation angle and translational displacement) of the target location, respectively, to determine the corresponding... and This is equivalent to determining the local measurement value of the target location.

[0045] It should be noted that the time interval between two consecutive "moments" in the concepts of current moment, historical moment, and previous moment referred to in this specification can be determined according to the control cycle of the aircraft (the time required for the aircraft control system to complete one "aircraft state sampling-calculation-control command output" process).

[0046] It should also be noted that the sensor's sampling interval can be shorter than the control cycle, meaning that the sensor can output multiple data points between the current moment and the previous moment. In cases such as... Figure 2 In each step of the method shown, the output data of a sensor at the current moment can be determined based on multiple measurements of the sensor between the current moment and the previous moment.

[0047] Step S203: Determine the weight corresponding to each target part based on the vibration state of each target part. The vibration state is determined based on the local measurement values ​​of the target part at the current time and at each historical time.

[0048] After determining the local measurement value of any target part, the computing device can determine the vibration state of the target part based on the local measurement value of the target part at the current time and several consecutive historical times before the current time.

[0049] It should be noted that the local measurement values ​​of the target area mainly include two parts of signals: one is the motion state of the aircraft itself, and the other is the vibration (which can also be regarded as noise signal) of the part where the sensor is deployed. Among them, the motion state is the information that is expected to be extracted and utilized in this embodiment, while the vibration is the interference that needs to be eliminated.

[0050] According to existing technical solutions, even if noise separation is performed on local measurements based on the characteristics of the noise signal, it is impossible to completely eliminate the influence of noise on local measurements. Furthermore, in the subsequent prediction of aircraft attitude, noise will further affect the accuracy of the determined aircraft attitude. In other words, the vibration state of a target location reflects the reliability of the data for that target location.

[0051] Specifically, since vibration is the periodic motion of an object at a certain frequency, the vibration state of the target part can be determined by decomposing the triaxial acceleration in the local measurement values ​​of the target part.

[0052] In addition, since the effects of vibration are usually continuous, in this embodiment, when determining the vibration state of the target part, not only the local measurement value at the current moment is referenced, but also the local measurement values ​​of several consecutive historical moments before the current moment are introduced to more accurately determine the vibration state of the target part.

[0053] In this embodiment, the weight of each target part is determined based on its vibration state. A higher vibration intensity at a target part indicates lower data reliability, and therefore a lower weight is assigned to that target part.

[0054] Specifically, the vibration state corresponding to each target part can be normalized, and the weight corresponding to each target part can be determined based on the normalized vibration state.

[0055] In some implementations, for any target part, the triaxial vibration acceleration of the target part can be determined based on the local measurement value of the triaxial acceleration of the target part. Based on the triaxial vibration acceleration of the target part at the current time and at each historical time, the average vibration acceleration of the target part is determined as the vibration state of the target part. Furthermore, for any target part, the weight corresponding to the target part is determined based on the average vibration acceleration of the target part, wherein the larger the average vibration acceleration of the target part, the smaller the weight corresponding to the target part.

[0056] Specifically, the triaxial accelerations of the target location at the current moment and at each historical moment can be used as an acceleration signal sequence. This acceleration signal sequence can be decomposed to determine the vibration acceleration signal sequence, thereby determining the triaxial vibration acceleration of the target location at the current moment and at each historical moment. Methods such as the Fast Fourier Transform (FFT) can be used to transform the time-domain representation of the sequence to the frequency domain representation. Periodic features can be extracted in the frequency domain representation to determine the triaxial vibration acceleration. It should be noted that vibration acceleration is caused by the deformation vibration of the organism. For a target location, the greater the vibration acceleration at that location, the greater the degree of deformation vibration at that location. Correspondingly, the reliability of the local measurement value of that target location should be lower.

[0057] The following formula can be used to determine the weight of each target part:

[0058] (2)

[0059] (3)

[0060] (4)

[0061] in, Let be the vibration acceleration of the i-th target part at time t; These are the triaxial components of the vibration acceleration; This represents the average vibration acceleration of the i-th target location; This represents the total time interval between the current moment and all historical moments. For example, the current moment is... Each historical moment is ( (For the historical moment furthest from the current moment), then ; Let be the weight of the i-th target part; the total number of target parts is . indivual.

[0062] As described above, the average vibration acceleration reflects the vibration state of the target location over a period of time, including the current moment, thus indicating the reliability of the local measurement values ​​of the target location. To ensure the reliability of the fused comprehensive measurement values, the lower the reliability of the local measurement values, the smaller their corresponding weights should be. Therefore, in this embodiment, the greater the average vibration acceleration of a target location, the smaller the weight corresponding to that target location.

[0063] In determining the weights of the target locations, incorporating historical vibration acceleration can reduce the impact of instantaneous noise at the current moment on the accuracy of the current vibration acceleration. However, on the other hand, incorporating historical vibration acceleration can also reduce the influence of the current vibration acceleration on the average vibration acceleration, resulting in a certain degree of response lag.

[0064] Therefore, this embodiment takes into account the trade-offs between noise elimination and rapid response requirements under different flight states. In some implementations, the length of the time window corresponding to the current moment is determined according to the flight state of the aircraft. The flight state includes at least take-off, landing and cruise. Several target moments are determined in each historical moment according to the length of the time window. The average vibration acceleration of the target part is determined according to the triaxial vibration acceleration of the target part at the current moment and the several target moments.

[0065] In the aircraft's takeoff and landing phase, where high response speed is required, the target times can be taken from a shorter time window preceding the current time. However, in the aircraft's cruise phase, noise can be excluded as the target, and the target times can be taken from a longer time window preceding the current time. Specifically, the length of the time window can be set by the executor in this embodiment. In some implementations, the time window length can be set within the range [0.1s, 0.5s].

[0066] Step S205: Based on the weight corresponding to each target part, perform weighted fusion of each local measurement value to determine the comprehensive measurement value of the attitude change rate at the current moment.

[0067] After determining the weight corresponding to each target part, the local measurement values ​​can be weighted and fused. Since the local measurement values ​​come from parts such as the fuselage, wings, and rotor platform, if the aircraft body deforms and the center of gravity shifts, the local measurement values ​​of the wings and rotor platform can dynamically reflect the above deformation. Therefore, the comprehensive measurement value obtained by weighting and fusing the local measurement values ​​can more accurately represent the measurement value for the aircraft as a whole (center of gravity).

[0068] The comprehensive measurement values ​​determined through the above steps fully include the attitude information of each target part contained in each local measurement value. Compared with the measurement values ​​from the fuselage alone, they can reflect the true attitude of the aircraft after local deformation to a certain extent. On the other hand, the interference of noise information is reduced by dynamically determined weights, which can more accurately reflect the attitude of the aircraft at the current moment.

[0069] Step S207: Based on the comprehensive measurement value and the predicted state of the aircraft attitude at the current moment, the Kalman filter method is used to determine the aircraft attitude at the current moment, wherein the predicted state at the current moment is calculated based on the aircraft attitude determined at the previous moment.

[0070] The attitude change of an aircraft is a continuous process. If the current attitude of the aircraft is determined solely based on the output data of the sensors, it is equivalent to ignoring the influence of the aircraft's attitude at historical moments on the current moment, as well as the influence of the aircraft's motion state on its attitude. This may result in a certain degree of distortion in the measured aircraft attitude.

[0071] Therefore, the computing device can be cross-calibrated by combining the aircraft attitude from the previous moment with the integrated measurement value. After determining the integrated measurement value, the Kalman filter method is used to determine the aircraft attitude at the current moment.

[0072] Specifically, in this embodiment, step S207 may include a prediction process and an update process. During the prediction process, the predicted state of the aircraft attitude at the current moment is calculated based on the aircraft attitude at the previous moment. During the update process, the predicted state of the aircraft attitude at the current moment is corrected based on the comprehensive measurement values ​​to obtain the aircraft attitude at the current moment.

[0073] The prediction process can refer to the following formula:

[0074] (5)

[0075] (6)

[0076] Among them, formula (5) is the formula for determining the predicted state. Indicates the previous moment; Indicates the current moment; This indicates the predicted state of the aircraft's attitude at the current moment; This represents the prediction matrix, the parameters of which can be determined based on the aircraft's dynamic model; This indicates the aircraft's attitude as determined at the previous moment. This represents Gaussian noise.

[0077] Formula (6) is the formula for determining the error covariance matrix. Let represent the prior error covariance matrix at the current moment, and represent the predicted state of the aircraft's attitude at the current moment. Estimation of the difference between the actual aircraft attitude and the actual aircraft attitude; This represents the process noise covariance matrix corresponding to the previous moment. The parameters of this process noise covariance matrix can be set empirically, representing the influence of the predicted unmodeled factors (such as wind direction and other unknown disturbances) on the aircraft attitude. This is the state transition matrix from the previous moment, which can also be determined based on the aircraft's dynamics model; for Transpose of; Let be the posterior error covariance matrix of the previous time step, the specific meaning of which will be explained below.

[0078] The update process can refer to the following formula:

[0079] (7)

[0080] (8)

[0081] (9)

[0082] Among them, formula (7) is the formula for determining the Kalman gain. This represents the Kalman gain at the current moment, indicating the proportion of the predicted state and the combined measurement value to be trusted when determining the aircraft's attitude at the current moment. The measurement matrix corresponding to the current moment is used to convert the comprehensive measurement value and the predicted state to the same parameter space (taking this embodiment as an example, the comprehensive measurement value is called the comprehensive measurement value of the aircraft attitude change rate, including the three-axis angular velocity and the three-axis acceleration, and the aircraft attitude is represented by attitude quaternions, and the corresponding predicted state is also represented by attitude quaternions. Therefore, if it is necessary to calculate the comprehensive measurement value and the predicted state, the parameter space of the two must be unified). for Transpose of; The measurement noise covariance at the current moment represents the error caused by the sensor's systematic error, which can be determined through a prior calibration test of the sensor; other parameters can be found in the previous text.

[0083] Formula (8) is the formula for determining the attitude of the aircraft at the current moment. This indicates the current attitude of the aircraft; This represents the local measurement value of the i-th target location at the current moment; This is the comprehensive measurement value; other parameters can be found in the previous text. As mentioned above, This indicates the percentage by which the predicted state and the overall measurement value should be considered. According to formula (4), when... The larger the value, the higher the proportion of the overall measurement value that can be trusted.

[0084] Formula (9) is the formula for determining the posterior error covariance matrix at the current time. Let be the posterior error covariance matrix at the current time, representing the aircraft attitude at the current time. Estimation of the difference between the actual aircraft attitude and the actual aircraft attitude; This is the identity matrix; other parameters can be found in the previous text. In the previous text, the posterior error covariance matrix of the previous time step... It can also be determined using formula (9).

[0085] It should be noted that, to further improve the accuracy of the aircraft attitude determined at the current moment, the aircraft attitude can include sensor error estimates in addition to the attitude quaternions. This allows for continuous correction of sensor errors during the prediction of the aircraft attitude at each moment using the Kalman filtering method. Therefore, the aircraft attitude can be expressed as:

[0086]

[0087] in, For attitude quaternions; For three-axis gyroscope zero bias, The zero bias of the triaxial accelerometer is represented by both, which indicate the estimation of sensor error.

[0088] It should be noted that to obtain the yaw, roll, and pitch angles of the aircraft, further calculations of the attitude quaternions are required.

[0089]

[0090]

[0091]

[0092] in, Indicates the yaw angle; Indicates the roll angle; This represents the pitch angle. In some implementations, since yaw information can be obtained through other channels (such as dual GPS or magnetometers in RTK), and the phase lag of the fuselage in the yaw direction is generally small, it is not necessary to determine the yaw angle of the aircraft based on quaternion calculations. This improves the efficiency of determining the aircraft's attitude.

[0093] In addition to the relevant content of the comprehensive measurement values, the parameters in the above formulas can also be determined by referring to the usage of Kalman filtering methods in various common aircraft attitude determination schemes, such as the extended Kalman filter formula in the open-source PX4 aircraft control system, which will not be elaborated here.

[0094] In this embodiment, the current attitude of the aircraft can be determined by combining the comprehensive measurement value with the aircraft attitude at the previous moment.

[0095] like Figure 2 The method for determining the attitude of an aircraft is shown, which proposes a deployment scheme of deploying sensors at each target location to form an observation network. Based on this, a method for fusing the measurement values ​​of each target location is further proposed. The resulting comprehensive measurement value can reduce the interference of local deformation and obtain a more accurate aircraft attitude.

[0096] In addition, in such Figure 2 In step S201 shown, the output data of the sensor can also be preprocessed, and then the preprocessed output data can be converted and fused to obtain the local measurement values ​​of each target part.

[0097] In some implementations, for any output data from several sensors deployed on the fuselage at the current moment, the output data is input into a pre-determined fuselage rotor transfer function to obtain the hysteresis compensation result corresponding to the output data. The parameters in the fuselage rotor transfer function are determined through vibration observation experiments on the aircraft. The hysteresis compensation results are fused to obtain the local measurement value of the attitude change rate corresponding to the fuselage.

[0098] The fuselage is the most stable part among all target components, but this also means that the fuselage's sensing of aircraft deformation is relatively lagging. The process of an aircraft changing its flight state usually begins with the rotor changing its speed and direction. However, because the rotor and fuselage are not rigidly connected, the transmission of changes in flight state to the fuselage is always delayed, and the output data of the sensors deployed on the fuselage cannot directly reflect these changes in flight state. Therefore, lag compensation can be applied to the output data of the sensors deployed on the fuselage to offset the phase lag of the corresponding output data. Specifically, the output data can be input into the fuselage rotor transfer function to obtain the corresponding compensation result. This fuselage rotor transfer function can be expressed as the following formula:

[0099]

[0100] in, That is, the fuselage rotor transfer function. This is the steady-state gain, representing the ratio of the system's output to its input in steady state. The damping ratio describes the degree of damping of the system. It is the undamped natural frequency. This is a Laplace variable, derived from the Laplace transform, representing the result of transforming the output data from the time domain to the frequency domain. , , These parameters can all be determined through observational experiments on the aircraft.

[0101] In other implementations, for any output data from several sensors deployed on the wing at the current moment, the output data is input into a predetermined deformation compensation function to obtain the deformation compensation result corresponding to the output data. The parameters in the deformation compensation function are determined by establishing a finite element model for the aircraft. The deformation compensation results are then fused to obtain a local measurement value of the attitude change rate corresponding to the wing.

[0102] Due to limitations in materials and shape, the wing is the most easily deformed and undergoes the greatest deformation among all target components. This also means that the output data of sensors deployed on the wing may contain acceleration deviations caused by airframe deformation. Therefore, the output data can be processed using a pre-determined deformation compensation function to obtain the corresponding deformation compensation result.

[0103] Specifically, the parameters in the deformation compensation matrix can be determined based on the finite element model of the aircraft. This involves discretizing the aircraft structure into a finite number of elements, pre-determining the material of each element and the connections between them, and then correcting parameters such as the mass of each element using experimental data obtained from observational experiments of the aircraft. Based on this finite element model, the deformation and stress state of various parts of the aircraft under different stress conditions can be determined. Furthermore, the sensor output data can also reflect the stress and partial states of corresponding parts, thus allowing the deformation compensation function for the wing section to be determined based on the finite element model.

[0104] In other implementations, for any sensor deployed on the rotor platform, based on the sensor's output data at the current moment and several historical moments, the corresponding output signal sequence of the sensor is determined. The output signal sequence is then decomposed into multiple sub-bands corresponding to multiple frequency intervals. Each sub-band corresponds to different wavelet packet coefficients. Several sub-bands corresponding to preset noise intervals are determined as several target sub-bands. For any target sub-band, the wavelet packet coefficients corresponding to the target sub-band are determined as the target coefficients of the target sub-band. The output data at the current moment is denoised based on each target coefficient to determine the denoising result corresponding to the sensor. The denoising results are then fused to obtain the local measurement value of the attitude change rate corresponding to the rotor platform.

[0105] The rotor platform is the closest part to the vibration source (rotor) among all target parts, and correspondingly, the phase delay of the output data of the sensors deployed on the rotor platform is the lowest. However, on the other hand, the vibration of the rotor introduces too much noise into the output data corresponding to the rotor platform. Therefore, in this embodiment, the output data of the sensors deployed on the rotor platform is denoised to remove the noise introduced by the rotor vibration from the output data.

[0106] Specifically, this embodiment utilizes wavelet packet decomposition to determine the corresponding output signal sequence of a sensor by sequentially sorting the sensor's output data at the current moment and several consecutive historical moments prior to the current moment. Wavelet packet decomposition is then applied to the output signal sequence, dividing it into multiple sub-bands corresponding to multiple frequency ranges.

[0107] Figure 4 A schematic diagram of the wavelet packet decomposition process is shown. For example... Figure 4As shown, the complete process of wavelet packet decomposition can include multiple decompositions. In the first decomposition, the original output signal sequence is decomposed into two nodes: a high-frequency part and a low-frequency part. Taking the frequency distribution of the original signal sequence in the frequency range of [0, 1000] as an example, the nodes of the high-frequency part and the low-frequency part correspond to the frequency ranges of [0, 500] and (500, 1000], respectively. In the second decomposition, the nodes obtained from the first decomposition are further decomposed, and each node is decomposed into two sub-nodes: a high-frequency part and a low-frequency part. Four child nodes are obtained, corresponding to the four frequency intervals [0, 250), (250, 500), (500, 750), and (750, 1000). This process is repeated in subsequent decomposition layers until the preset number of decomposition layers is reached. The child nodes of the last layer represent the sub-bands. Each sub-band corresponds to a wavelet packet coefficient, which represents the projection of the original output signal sequence onto the corresponding frequency. The larger the value of the wavelet packet coefficient, the more feature information the original signal has at the frequency corresponding to that sub-band.

[0108] Typically, vibration noise has a higher frequency than the real signal representing the rate of change of aircraft attitude. Therefore, the higher frequency range can be preset as the noise range, and the sub-band corresponding to the noise range can be determined as the target sub-band. In some implementations, the noise range can be set to [200Hz, 1000Hz].

[0109] Furthermore, the target coefficients of the target sub-bands are processed to remove sub-bands containing less feature information. Then, the wavelet packet coefficients corresponding to each sub-band are reconstructed using the reconstruction method corresponding to wavelet packet decomposition, and the denoising result corresponding to the sensor can be obtained.

[0110] Specifically, the target coefficients of the target subband can be processed by setting a noise threshold. Target coefficients below this noise threshold can be set to zero, thereby removing high-frequency, low-intensity noise signals from the output signal sequence.

[0111] Next, the wavelet packet coefficients (which may include the target coefficients set to zero) of all sub-bands (not just the target sub-band, but all sub-bands corresponding to all frequencies obtained from wavelet packet decomposition) are reconstructed and aggregated to determine the denoised output signal sequence. The element corresponding to the current time in this denoised output signal sequence is the denoising result of the sensor.

[0112] In addition, in some implementations, the noise threshold can be determined based on the rotor speed of the target aircraft at the current moment, wherein the higher the rotor speed, the greater the noise threshold.

[0113] Specifically, based on pre-conducted observation experiments of the aircraft, the rotor speed of the aircraft at each moment can be obtained as sample data, and the noise intensity of the aircraft at each moment (this noise intensity can be determined by processing the sensor output data of the aircraft at each moment using various common filtering methods) can be obtained as the label corresponding to each sample data. A noise threshold prediction model is trained using this sample data and the label.

[0114] Furthermore, the rotor speed of the target aircraft at the current moment can be input into the noise threshold prediction model to determine the noise threshold at the current moment, and the wavelet packet coefficients can be processed based on the dynamically determined noise threshold.

[0115] Therefore, by suppressing high-frequency vibrations while preserving as much effective signal as possible, the problem of excessive signal attenuation by traditional filters can be avoided.

[0116] In addition, the observation experiments mentioned above can be simulation experiments using digital models, iron bird test bench experiments, or actual flight experiments, etc. This manual does not impose any restrictions on these.

[0117] It should be understood that the descriptions such as "first" and "second" in this article are merely for the sake of simplicity in description and to distinguish similar concepts, and do not have any other limiting function.

[0118] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0119] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0120] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. The software modules can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

[0121] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for determining the attitude of an aircraft, characterized in that, The method involves multiple sensors deployed at various target locations on an aircraft, including wings, fuselage, and rotor platform. The method includes: Based on the output data of the multiple sensors at the current moment, a local measurement value of the attitude change rate for each target part is determined. The attitude change rate includes three-axis angular velocity and three-axis acceleration. The three axes include pitch axis, roll axis and yaw axis. For any target location, the triaxial vibration acceleration of the target location is determined based on the local measurement value of the triaxial acceleration of the target location; based on the triaxial vibration acceleration of the target location at the current time and at each historical time, the average vibration acceleration of the target location is determined as the vibration state of the target location. The weight of the target part is determined based on the average vibration acceleration of the target part. The greater the average vibration acceleration of the target part, the smaller the weight of the target part. Based on the weight corresponding to each target part, the local measurement values ​​are weighted and fused to determine the comprehensive measurement value of the attitude change rate at the current moment; Based on the comprehensive measurement values ​​and the predicted state of the aircraft attitude at the current moment, the aircraft attitude at the current moment is determined using the Kalman filter method. The predicted state at the current moment is calculated based on the aircraft attitude determined at the previous moment. The aircraft attitude includes at least attitude quaternions.

2. The method as described in claim 1, characterized in that, Based on the output data of the multiple sensors at the current moment, determine the local measurement value of the attitude change rate for each target location, specifically including: For any target location, the output data of several sensors deployed at that target location at the current moment are fused to obtain a local measurement value of the attitude change rate of that target location.

3. The method as described in claim 1, characterized in that, Based on the triaxial vibration acceleration of the target location at the current moment and at various historical moments, the average vibration acceleration of the target location is determined, specifically including: The length of the time window corresponding to the current moment is determined based on the flight status of the aircraft, and the flight status includes at least takeoff and landing and cruise. Based on the length of the time window, several target times are determined from each historical time. Based on the triaxial vibration acceleration of the target location at the current time and at several target times, determine the average vibration acceleration of the target location.

4. The method as described in claim 2, characterized in that, For any target location, the output data of several sensors deployed at that target location at the current moment are fused to obtain a local measurement of the attitude change rate of that target location, specifically including: For any output data from several sensors deployed on the fuselage at the current moment, the output data is input into a predetermined fuselage rotor transfer function to obtain the hysteresis compensation result corresponding to the output data. The parameters in the fuselage rotor transfer function are determined by vibration observation experiments on the aircraft. By fusing the results of each hysteresis compensation, a local measurement of the attitude change rate corresponding to the fuselage is obtained.

5. The method as described in claim 2, characterized in that, For any target location, the output data of several sensors deployed at that target location at the current moment are fused to obtain a local measurement of the attitude change rate of that target location, specifically including: For any output data from several sensors deployed on the wing at the current moment, the output data is input into a predetermined deformation compensation function to obtain the deformation compensation result corresponding to the output data. The parameters in the deformation compensation function are determined by establishing a finite element model for the aircraft. By fusing the results of various deformation compensations, a local measurement value of the attitude change rate corresponding to the wing is obtained.

6. The method as described in claim 2, characterized in that, For any target location, the output data of several sensors deployed at that target location at the current moment are fused to obtain a local measurement of the attitude change rate of that target location, specifically including: For any sensor deployed on the rotor platform, determine the corresponding output signal sequence of the sensor based on the output data of the sensor at the current time and several historical times; The output signal sequence is decomposed into multiple sub-bands corresponding to multiple frequency intervals, and each sub-band corresponds to different wavelet packet coefficients. Several sub-bands corresponding to a preset noise range are identified as several target sub-bands; For any target sub-band, determine the wavelet packet coefficients corresponding to the target sub-band, and use them as the target coefficients for that target sub-band; The output data at the current moment is denoised based on each target coefficient to determine the denoising result corresponding to the sensor. The denoising results are fused together to obtain the local measurement value of the attitude change rate of the rotor platform.

7. The method as described in claim 6, characterized in that, The output data at the current moment is denoised based on each target coefficient to determine the denoising result corresponding to the sensor, specifically including: The noise threshold is determined based on the rotor speed of the target aircraft at the current moment, wherein the higher the rotor speed, the greater the noise threshold. For any target coefficient, if the target coefficient is less than the noise threshold, the wavelet packet coefficient corresponding to the target coefficient is set to zero; The wavelet packet coefficients in the multiple sub-bands are aggregated to determine the denoised output signal sequence. Based on the denoised output signal sequence, determine the denoising result corresponding to the sensor.

8. A computing device comprising a memory and a processor, wherein the memory stores executable code, characterized in that, When the processor executes the executable code, it implements the method of any one of claims 1-7.

Images