Large-scale controlled spin-stabilized projectile

Abstract

The application provides a large-scale difference spin tail control missile pellet, which comprises a main body, a stern fixedly connected with the main body, and a plurality of spin reduction wing pieces arranged on the stern, wherein the plurality of spin reduction wing pieces are configured to rapidly pop out in a direction away from an axis of the pellet under the action of centrifugal force when the pellet is in a self-rotation state at a speed greater than a preset speed threshold, but do not leave the main body. The application solves the technical problem that the obtained motion parameters of the pellet are unstable due to the effect of air resistance on the spin reduction wing pieces when the pellet rotates at a high speed.

Description

Technical Field

[0001] This application relates to the field of navigation technology, and more specifically, to a large-scale differential spin tail control missile projectile. Background Technology

[0002] With the continuous development of modern military technology, the precision strike of projectiles against targets has become a major research and development direction. Understanding the position, velocity and direction of projectiles in space at various times is crucial for their precision guidance.

[0003] However, high-speed projectiles are subjected to impacts of up to 20,000G upon launch, resulting in high speed and rotational speed. This immense impact force damages the sensors mounted on the projectile, reducing their measurement accuracy and making them more susceptible to noise interference, thus increasing navigation errors. Simultaneously, the projectile's high speed and high vibration during flight make accurate measurement of its state data even more difficult, necessitating the use of mathematical methods (navigation methods) to fuse state data with measurement data to obtain optimal state estimates and compensate for sensor errors. Therefore, researching projectile navigation methods is crucial.

[0004] Since the actual motion of a projectile during flight is nonlinear, albeit weakly so under certain conditions, the Extended Kalman Filter (EKF) algorithm is typically used to estimate its state. The core of the EKF algorithm is based on first-order linearization of the nonlinear system, followed by Kalman filtering to estimate the system's state. Traditional EKF algorithms set both the measurement noise covariance matrix and the system noise covariance matrix to constant values. However, the noise during projectile flight is random, leading to errors in the filtering results. Furthermore, linearization causes the error to increase with the Kalman gain, compromising the convergence of the extended Kalman filter. To mitigate the impact of noise randomness on the system, an adaptive extended Kalman filter (AEKF) algorithm has been proposed. This algorithm estimates measurement noise during iteration, thereby reducing system filtering errors. However, the projectile's flight is divided into four phases: launch, ascent, apex crossing, and acceleration descent. The motion characteristics and random noise variations differ in each phase, causing changes in the measurement noise characteristics. The noise estimator parameters of the general AEKF algorithm are not immediately adjusted according to the different flight stages of the projectile during the algorithm iteration. This makes the traditional AEKF algorithm not very effective in mitigating the impact of noise randomness on the system.

[0005] There is currently no effective solution to the above problems. Summary of the Invention

[0006] This invention provides a large-scale differential spin tail control missile projectile to at least solve the technical problem of unstable obtained projectile motion parameters caused by the effect of air resistance on the decoy vanes when the projectile rotates at high speed.

[0007] According to one aspect of the present invention, a large-scale differential spin-tail control missile projectile is provided, comprising: a main body; a stern fixedly connected to the main body, wherein a plurality of spin-damping vanes are provided on the stern, the plurality of spin-damping vanes being configured such that when the projectile is spinning at a speed greater than a preset speed threshold, it is rapidly ejected away from the axis of the projectile under the action of centrifugal force, but does not leave the projectile body.

[0008] In this embodiment of the invention, the large-scale differential spin-tail control missile projectile includes a main body and a stern, wherein the stern is fixedly connected to the main body, and a plurality of anti-spin vanes are provided on the stern. These anti-spin vanes are configured such that when the projectile is spinning at a speed greater than a preset speed threshold, it is rapidly ejected away from the projectile's axis under the action of centrifugal force, but does not leave the projectile body. This application solves the technical problem that the obtained motion parameters of the projectile are unstable due to the effect of air resistance on the anti-spin vanes when the projectile is rotating at high speed. Attached Figure Description

[0009] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0010] Figure 1 This is a diagram illustrating the composition of a combined navigation system according to an embodiment of this application;

[0011] Figure 2 This is a schematic diagram of the navigation coordinate system and the projectile coordinate system according to an embodiment of this application;

[0012] Figure 3 This is a schematic diagram of the translated coordinate system according to an embodiment of this application;

[0013] Figure 4 This is a structural diagram of the measurement unit hardware according to an embodiment of this application;

[0014] Figure 5A This is a structural diagram of a large-scale differential spin tail controlled missile projectile according to an embodiment of this application;

[0015] Figure 5B This is a diagram of the stern structure of a large-scale differential spin tail control missile projectile according to an embodiment of this application;

[0016] Figure 6This is a flowchart of a segmented AEKF projectile integrated navigation method based on CNN according to an embodiment of this application;

[0017] Figure 7 This is a schematic diagram of a CNN network structure according to an embodiment of this application;

[0018] Figure 8 This is a schematic diagram of parameter estimation for a CNN-based noise estimator according to an embodiment of this application;

[0019] Figure 9 This is a flight phase identification and analysis diagram according to an embodiment of this application;

[0020] Figure 10 This is a confusion matrix diagram according to an embodiment of this application;

[0021] Figure 11 It is a pre-modified integrated navigation flight trajectory diagram based on existing technology;

[0022] Figure 12 It is a position error diagram based on existing technology before improvement;

[0023] Figure 13 It is a speed error diagram based on existing technology before improvement;

[0024] Figure 14 This is a diagram illustrating the projectile flight phases according to an embodiment of this application;

[0025] Figure 15 This is a comparison diagram of speed error before and after the improvement according to the embodiments of this application;

[0026] Figure 16 This is a comparison diagram of the positional error before and after the improvement according to the embodiments of this application;

[0027] Figure 17 This is a modified integrated navigation flight trajectory diagram according to an embodiment of this application. Detailed Implementation

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0030] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail. Any specific values ​​in all examples shown and discussed herein should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0031] Overview

[0032] In high-speed projectile integrated navigation systems, the statistical characteristics of measurement noise cannot be accurately obtained due to the complex aerial environment and the projectile's own high-speed spin and vibration, leading to a decrease in the accuracy of the integrated navigation system. To address this issue, this application proposes a piecewise adaptive extended Kalman filter (AEKF) high-speed projectile integrated navigation method based on a convolutional neural network (CNN). Within a loosely integrated GNSS / SINS framework, this method trains a CNN model capable of real-time identification of projectile flight stages (launch phase, ascent phase, apex crossing phase, and acceleration descent phase) based on the velocity change rate and pitch angle characteristics of each flight stage of the high-speed projectile. Specific flight stages are associated with the noise estimator parameters in the AEKF algorithm, allowing the projectile to adaptively adjust the measurement noise covariance matrix of the AEKF filtering algorithm according to the flight stage during high-speed flight, thereby improving the accuracy of projectile integrated navigation. The method provided in this application embodiment was tested and compared with the general AEKF projectile combined navigation method under the same dataset. The test results show that the average velocity error and position error of this method decreased by 56.33% and 42.87% respectively, which has good reference and application value.

[0033] Example 1

[0034] This application provides a segmented AEKF high-speed projectile integrated navigation system based on CNN, such as... Figure 1 As shown, the system includes a satellite navigation module, an AEKF algorithm module, a navigation solution module, and a convolutional neural network model. The integrated navigation system provided in this embodiment is designed in a loosely coupled manner, which can reduce the computational load of the system and has a certain degree of redundancy.

[0035] During the training phase of the CNN model, flight trajectory data of the missile projectile controlled by a 155mm large-scale differential spin tail is used for model training until a model capable of accurately identifying the projectile's flight phase is obtained. In the navigation phase, by inputting data from the accelerometer and gyroscope, the CNN model identifies the projectile's flight phase and outputs the noise estimator parameter λ corresponding to the flight phase, updating the noise estimator of the AEKF algorithm. Since the measurement accuracy of inertial devices gradually decreases during projectile flight, affecting navigation, satellite (GNSS) observation data is used to correct the SINS data. GNSS and SINS operate in parallel; the difference between the data output from the GNSS receiver and the SINS data is calculated to obtain the observations for the projectile's integrated navigation. This difference is input into the AEKF filtering algorithm to output the optimal error correction value, which is then fed back to the SINS to correct navigation parameters, resulting in accurate navigation information.

[0036] 1) Selection of coordinate system for navigation system.

[0037] The local navigation coordinate system (n) selected in this embodiment is the North-East-Ground coordinate system. The origin is o. n It is placed at a fixed point on the Earth's surface; the north, east, and ground coordinate axes are x, y, and x respectively. n y n and z n It is tangent to the Earth's surface. The carrier coordinate system (b) is right-front-bottom. The origin o is... b It is placed at the center of mass of the projectile; x in the carrier coordinate system b The axis is defined along the radial axis of symmetry of the projectile, pointing towards the front of the projectile; y b Defined as perpendicular to the projectile's axis and pointing to the right of the projectile; z b The axis points downwards according to the right-hand rule, such as... Figure 2 As shown. Because satellite observations were used to correct the SINS data, it is necessary to transform the satellite observation data from the geodetic coordinate system to the intermediate geocentric coordinate system, and then to the N-E-Geometric coordinate system.

[0038] Since the output data of the inertial measurement unit is in the carrier coordinate system, while the navigation parameters are solved in the navigation coordinate system, a rotation matrix is ​​required. Establish a connection between two coordinate systems. Specifically, this can be represented as:

[0039]

[0040] In the formula, since the transformation between the navigation coordinate system and the vehicle coordinate system is a three-dimensional transformation, the rotation matrix is... It is obtained by multiplying the two-dimensional rotation matrices corresponding to rotations around the three coordinate axes:

[0041]

[0042] For easier observation, Figure 2 o in b Translate to o n Place, such as Figure 3 As shown. Figure 3 θ, ψ represents the pitch angle, roll angle, and yaw angle of the carrier, respectively, and x represents the x-axis. c y c z c This is the intermediate coordinate system during rotational transformations.

[0043] 2) Construction of Mathematical Model for Navigation System

[0044] In a projectile-based integrated navigation system, the system's state equations are generally composed of state errors, and the expressions for the state equations are as follows:

[0045]

[0046] In the formula, X k F represents the 15-dimensional error state variable at time k; k+1 / k D is the 15×15 invertible state transition matrix of the system from time k to time k+1; k This is a 15-dimensional noise matrix. δρ represents the attitude error, where ρ is the attitude (yaw, pitch, and yaw) defined by Euler angles, ρ=[φ θ ψ] T δv represents the 3D velocity error; δP represents the 3D position error; δb a For accelerometer zero bias error (3×1 vector); δb ω This represents the gyroscope's zero bias error (3×1 vector).

[0047] The error state equation is:

[0048]

[0049] In the formula: δρ k+1 The attitude error at time k+1; δv k+1 The velocity error at time k+1; δP k+1 The position error at time k+1; δb a(k+1) δb represents the accelerometer zero bias error at time k+1. ω(k+1) The gyroscope bias error at time k+1; β1 is the direction cosine matrix; β2 is the accelerometer zero-bias error proportionality coefficient; D is the gyroscope zero-bias error proportionality coefficient; ak and D ωk This represents the noise of the accelerometer and gyroscope at time k.

[0050] The measurement equation is:

[0051] Z k =H k X k +V k (5)

[0052] Wherein: H k For the measurement matrix, V k For measuring noise arrays.

[0053] Assume the noise matrix D in the state equation k Measurement noise V of the observation equation k All are zero-mean white noise and are uncorrelated, that is:

[0054]

[0055] Where γ k,i Defined as:

[0056]

[0057] This embodiment uses a loosely combined GNSS / SINS model and a modified AEKF algorithm to achieve projectile navigation. Simulation results show that the modified AEKF algorithm has smaller velocity and position errors than the traditional AEKF algorithm, making the system more stable.

[0058] Therefore, this embodiment is based on a convolutional neural network (CNN) to estimate the flight stage of the projectile by inputting accelerometer and gyroscope data, and dynamically changes the parameters in the noise estimator to improve the AEKF algorithm.

[0059] Example 2

[0060] This application provides a projectile.

[0061] Considering the high overload, high rotational speed, and high roll of the projectile during flight, as well as the limited space for installing integrated navigation devices on the projectile, the hardware structure of the projectile's measurement unit is specifically designed as follows: Figure 4 As shown.

[0062] The projectile is equipped with military-grade sensors, including a GPS receiver, a three-axis MEMS gyroscope, a three-axis MEMS accelerometer, a three-axis magnetometer, and a telemetry transceiver. Furthermore, the projectile utilizes an ARM Cortex-M7 microprocessor, coupled with an FPGA, to meet the data acquisition and processing requirements during flight.

[0063] In addition, considering that the projectile is in a high-speed spin state during flight, it poses a great difficulty in obtaining the projectile's navigation parameters, affecting navigation accuracy. Furthermore, existing projectile despin reduction technology is not suitable for high-dynamic, high-speed environments. Therefore, a large-scale differential spin tail control projectile was developed, the structure of which is as follows: Figure 5A As shown.

[0064] The large-scale differential spin-tail controlled missile projectile consists of two parts: the main body and the stern. Unlike traditional guided missile projectiles, the stern is equipped with anti-spin vanes. When the projectile is in a high-speed spinning state, the anti-spin vanes are rapidly ejected away from the projectile's axis due to centrifugal force, but they do not leave the projectile body. The air resistance during the projectile's high-speed rotation acts on the anti-spin vanes, reducing the projectile's tail rotation speed. This allows for stable acquisition of the projectile's motion parameters, improving navigation accuracy. The anti-spin effect only affects the aerodynamic forces at the projectile's tail, having a relatively small impact on the overall trajectory.

[0065] The structure of the stern of a projectile ship is as follows Figure 5B As shown, the stern of the projectile is equipped with four anti-spin blades, with equal spacing between each blade, so that the tail of the projectile can be evenly subjected to air resistance during the anti-spin process, ensuring the stability of the projectile flight.

[0066] Example 3

[0067] This application provides a segmented AEKF projectile navigation method based on CNN, such as... Figure 6 As shown, the method includes the following steps:

[0068] Step S602: Based on a convolutional neural network, the flight phase of the projectile is identified using data from the accelerometer and gyroscope.

[0069] Step S604: Associate the identified flight phase with the parameters of the noise estimator in the AEKF algorithm to obtain the improved AEKF algorithm, and adaptively adjust the parameters of the noise estimator according to the identified flight phase.

[0070] Step S606: Navigate the projectile based on the improved AEKF algorithm.

[0071] Traditional Extended Kalman Filter (AEKF) algorithms typically require accurate knowledge of the mathematical characteristics of system noise and measurement noise. However, system noise is random in real-world scenarios, making accurate estimation impossible. Therefore, a noise estimator is introduced into the AEKF algorithm, forming a new filtering algorithm. The introduction of the noise estimator allows the measurement noise covariance matrix to be updated continuously during algorithm iterations, enabling the system to adapt to changing noise environments and improving the accuracy of system estimation.

[0072] For nonlinear systems, it is necessary to establish their corresponding linear discretized system equations and measurement equations:

[0073]

[0074] In the formula: X k Z is the s-dimensional state estimate of the system at time k; k Let F be the u-dimensional observation sequence of the system at time k; k / k-1 It is the system state transition matrix; H k D is the system's observation matrix; k-1 For system noise sequence; V k To measure the noise sequence, D k-1 and V k They follow a normal distribution and are independent of each other.

[0075] The specific implementation steps of the AEKF algorithm are as follows:

[0076] State prediction:

[0077]

[0078] Measurement Update:

[0079]

[0080] Where: ε k This represents the residual information of the system at time k. K is the mean of the measured noise. k The Kalman gain at time k; P k Let be the system covariance matrix at time k; I is the identity matrix.

[0081] The introduced noise estimator can be set as follows:

[0082]

[0083] In the formula: The system measurement noise covariance matrix; Here is the system noise covariance matrix; d is the system noise mean; k This is the forgetting factor, which reduces the error of the filtering algorithm. Its specific form is:

[0084] d k = (1-λ) / (1-λ) k+1 (0<λ<1) (12)

[0085] The improved AEKF algorithm will be described in detail below.

[0086] As can be seen from formula (11), traditional noise estimators estimate both system noise and measurement noise. However, estimating both the measurement noise covariance matrix and the system noise covariance matrix simultaneously may cause the filtering algorithm results to diverge, thus compromising system stability. Therefore, existing technologies propose estimating only the measurement noise. However, considering that the noise characteristics of each flight stage of the projectile are different, a fixed noise estimator parameter λ cannot quickly respond to changes in noise characteristics. Therefore, this application uses a convolutional neural network (CNN) to identify the flight stage using accelerometer and gyroscope data. Each flight stage corresponds to a noise estimator parameter λ, and the estimation of measurement noise is dynamically adjusted.

[0087] The specific improvements to the AEKF algorithm are as follows:

[0088] λ=[λ1,λ2,λ3,λ4] (13)

[0089]

[0090] The following will describe CNN.

[0091] CNN is a deep learning model specifically designed for processing data with network structures. Its model structure is as follows: Figure 7 As shown. Convolutional layers are the most important part of a CNN network structure. Through convolutional computation, each data point can be associated with neighboring regions, capturing the features of the data, and are commonly used for processing signals and images. Pooling layers typically include max pooling and average pooling. Max pooling and average pooling operations replace the original values ​​by calculating the maximum or average value of the pooling window, achieving feature compression and reducing the computational cost of the model. Finally, the extracted feature maps are transformed into the final output through fully connected layers.

[0092] like Figure 8 As shown, during the projectile's flight, at each discrete time point, the CNN reads data from the accelerometers and gyroscopes deployed on the projectile, identifies the projectile's flight stage, assigns a noise estimator parameter λ to update the measurement noise, and outputs estimated projectile attitude, velocity, and position deviations through a filtering algorithm, along with accelerometer, gyroscope, and satellite measurement data. These deviation data also aid in model training.

[0093] The CNN was trained and used for prediction on a newly developed dataset of 155mm large-scale differential spin-tail control missile projectile flight trajectories, totaling data from 10 projectiles. The CNN network structure is shown in Table 1.

[0094] Table 1 CNN Network Structure

[0095]

[0096] The input to the convolutional neural network is a 100×6×1 three-dimensional time series sliding window with 100 time steps and 6 features per time step. Furthermore, non-linear activation functions—Rectified Linear Unit (ReLU) and Softmax—are used for all layers. Dropout layers are used to prevent overfitting, and the results from the convolutional and activation layers are then passed to fully connected layers to output projectile flight stage indicators.

[0097] The proposed method identifies the projectile's flight phase based on the velocity change rate dv / dt and pitch angle θ characteristics obtained from accelerometer and gyroscope data in the navigation system.

[0098] In order to reduce the amount of computation and ensure system stability when solving attitude information, the attitude matrix quaternion form is used to solve the attitude.

[0099] The expression for constructing a quaternion is:

[0100] Q = q0 + q1i + q2j + q3k (15)

[0101] The differential equation is:

[0102]

[0103] In the formula, The attitude angular rate of the b-frame relative to the n-frame Expanding equation (16) yields:

[0104]

[0105] After orthogonalization, the quaternion form of the attitude matrix is:

[0106]

[0107]

[0108] The attitude angle θ can then be represented by the elements in the attitude matrix. ψ:

[0109]

[0110] The velocity calculation is based on the specific force information f output by the accelerometer. b =[f x f y f z ] T The proportional differential equation in the navigation system is:

[0111]

[0112] Where: f n=[f N f E f D ] T The accelerometer measurement value f b In the projection of the n-system, g n Let be the projection of gravitational acceleration onto the n-frame. Let be the projection of the projectile's velocity in the n-frame.

[0113]

[0114]

[0115] Therefore, the velocity component representation can be obtained:

[0116]

[0117] The system's calculated rate of change of velocity, velocity, and pitch angle correspond to the projectile's flight stages as follows: Figure 9 As shown.

[0118] When the rate of change of projectile velocity dv / dt < 0 and the pitch angle θ > 0, the projectile is in the rising phase, and its velocity continuously decreases; when dv / dt < 0, θ = θ s When the velocity is 0, the projectile is at the apex, but the velocity at the apex is v. s =318.15m / s has not reached its minimum value and will continue to decrease; when dv / dt = 0 and θ < 0, the projectile is at its velocity minimum point and is preparing to enter the accelerated descent phase; when dv / dt > 0 and θ < 0, the projectile accelerates its descent. Based on the correspondence between velocity, rate of change of velocity, pitch angle, and flight phase, the correspondence between flight phase and the raw data from the accelerometer and gyroscope can be obtained, thus facilitating the training of the CNN model.

[0119] like Figure 10 As shown, this application evaluates the CNN network model using a confusion matrix. It can be seen that the CNN model may make misjudgments, but overall the model can be used for the identification of projectiles during flight.

[0120] This application proposes an improved AEKF algorithm based on CNN-based accelerometer and gyroscope data to identify the projectile's flight phases and dynamically adjust noise estimator parameters for navigation during high-speed projectile flight. According to test results from this application, the algorithm can provide accurate navigation for the projectile. Other filtering algorithms can also be used to effectively improve ground velocity and position errors. Furthermore, to enhance the algorithm's versatility, it can learn to use more complex projectile dynamics models, such as gravity and aerodynamic models, thus extending this navigation method to other types of projectiles.

[0121] Validation and Analysis

[0122] To verify the feasibility of the improved AEKF algorithm, this application compares the following two methods and presents the projectile navigation estimation results:

[0123] (1) In the general AEKF algorithm, the noise estimator parameter λ = 0.9 is set and not changed.

[0124] (2) The improved AEKF algorithm has its noise estimator parameter λ obtained by the CNN model after identifying the projectile flight stage and is dynamically adjusted.

[0125] The data used is flight data of a 155mm howitzer projectile. Projectile flight time: 92.8s, flight distance: approximately 30,000m, sampling rate: 1kHz, system initialization parameters are shown in Tables 2 and 3.

[0126] Table 2 Initial parameters of gyroscope and accelerometer

[0127]

[0128] Table 3 Initializing heading angle and position

[0129]

[0130] 1) AEKF Algorithm Test

[0131] This application sets the forgetting factor λ to 0.9 for the general AEKF algorithm. The projectile's combined navigation flight trajectory, position error, and velocity error are as follows: Figures 11 to 13 As shown.

[0132] Depend on Figure 11 It can be seen that there is a large discrepancy between the flight trajectory estimated by the integrated navigation system and the position updated by satellite measurements.

[0133] Depend on Figure 12 Figure 13 It can also be seen that from the start of the projectile launch, the northward and ground-oriented positional errors increased progressively, reaching maximum values ​​of 25.328m and 26m respectively. These errors only began to decrease after more than half the flight time, but remained relatively high. While the northward and eastward velocity errors initially decreased, they showed signs of rebounding after 50 seconds. The ground-oriented velocity error reached zero at 41 seconds and formed a clear turning point. Figure 9 The elevation curve shows that the projectile is at the apex of its trajectory at this point, with a relatively slow velocity and less interference. It then enters the acceleration descent phase, where the ground error begins to diverge. Therefore, it can be seen that the performance of the unmodified AEKF algorithm is poor.

[0134] 2) Testing the improved AEKF algorithm

[0135] As can be seen from the output of the AEKF algorithm before the improvement, when the error starts to increase, the system cannot suppress its growth in a short period of time, and even allows it to diverge.

[0136] The improved AEKF algorithm is a filtering method that dynamically adjusts the noise estimator parameter λ by using CNN to read accelerometer and gyroscope data to identify the projectile's flight phase. Other parameter settings remain the same as before the improvement. Therefore, the noise estimator parameter λ corresponding to the flight phase needs to be pre-defined. The specific correspondence is shown in Table 4. The projectile's flight phase identification is as follows: Figure 14 As shown.

[0137] Table 4. Dynamic Adjustment Table for Noise Estimator Parameter λ

[0138]

[0139] Under the above dynamic adjustment strategy, the velocity and position errors output by the improved AEKF algorithm are compared with the errors before the improvement. Figure 15 and 16 As shown in the figure. The errors of the two methods were statistically analyzed, and the average value of all error results is shown in Table 5.

[0140] Table 5. Statistics of Velocity Error and Position Error

[0141]

[0142] from Figure 15 and 16 As shown in Table 5, the improved AEKF algorithm reduces both velocity and position errors. Specifically, the northward velocity error stabilizes near 0 after 25 seconds, while the eastward velocity error rapidly converges to near 0 after projectile launch and remains there until the projectile lands. The maximum northward position error decreases from 25.328m to 15.8m.

[0143] In the final moments, the northward position error was optimized from 10m to 3m, the eastward position error from 0.5m to 0.2m, and the groundward position error from 13m to 5m; the northward velocity error was optimized from 5m / s to 2m / s, and the eastward velocity error from 5m / s to 0.3m / s. Table 5 also shows that compared to the general AEKF algorithm, the CNN-based segmented AEKF algorithm reduced the average velocity and position errors by 56.33% and 42.87%, respectively.

[0144] In this case, the projectile's combined navigation flight trajectory is as follows: Figure 17 As shown.

[0145] Depend on Figure 17As can be seen from the enlarged portion of the trajectory, the deviation between the improved integrated navigation trajectory and the satellite measurement update trajectory is much smaller than the deviation before the improvement.

[0146] Therefore, it can be concluded that the segmented AEKF algorithm based on CNN can enable accurate and stable navigation of high-speed projectiles, and has good reference and application value.

[0147] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0148] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0149] In the several embodiments provided in this application, it should be understood that the disclosed terminal device can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, or indirect coupling or communication connection between units or modules, and may be electrical or other forms.

[0150] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0151] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0152] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A large-scale differential spin-tail control missile projectile, characterized in that, Includes a projectile body, said projectile body comprising: main missile body; The stern is fixedly connected to the main projectile body. Multiple anti-rotation blades are provided on the stern. The multiple anti-rotation blades are configured to, when the projectile is spinning at a speed greater than a preset speed threshold, be rapidly ejected away from the axis of the projectile under the action of centrifugal force, but without leaving the projectile body. The projectile is also equipped with: sensors, including a GPS receiver for receiving satellite observation data, a gyroscope and accelerometer for sensing SINS data, and a triaxial magnetometer for measuring magnetic field strength; and a microprocessor for navigating the projectile based on the satellite observation data, the SINS data, and the magnetic field strength. The microprocessor is further configured to: identify the flight phases of the projectile based on SINS data from the accelerometer and the gyroscope using a convolutional neural network; each flight phase corresponds to a noise estimator parameter used to update the measurement noise; obtain the attitude, velocity, and position deviation of the projectile using the AEKF algorithm based on the satellite observation data, the SINS data, and the noise estimator parameters; and navigate the projectile based on the attitude, velocity, and position deviation.

2. The projectile according to claim 1, characterized in that, The plurality of rotor blades are also configured to reduce the rotational speed of the stern by means of the air resistance acting on the plurality of rotor blades when the projectile rotates at a speed greater than the preset speed threshold.

3. The projectile according to claim 2, characterized in that, The bottom surface of the stern facing away from the main body is provided with multiple blade slots corresponding to the multiple anti-rotor blades, and the multiple anti-rotor blades are fixed to the stern through the multiple blade slots.

4. The projectile according to claim 3, characterized in that, There are four wing slots, and the spacing between the wing slots is equal. The straight lines along the grooves of the wing slots converge at the center of the bottom surface of the stern.

5. The projectile according to claim 1, characterized in that, The stern of the projectile also includes a miniature antenna mounting slot for mounting the GPS receiver; The microprocessor is also configured to use the satellite observation data to correct the SINS data, obtain the observations of the projectile integrated navigation, input the observations into the AEKF algorithm to obtain the optimal error correction value, and then feed the optimal error correction value back to the inertial navigation system to correct the navigation parameters.

6. The projectile according to claim 1, characterized in that, The microprocessor is also configured to: Using data from the accelerometer and the gyroscope, the velocity change rate and pitch angle characteristics of the projectile in each flight phase are calculated in the navigation system; Based on the rate of change of velocity and the pitch angle characteristics, the flight stage of the projectile is identified in real time.

Images