Ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method and system for launch vehicles

Abstract

This invention relates to a method and system for ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control of launch vehicles, belonging to the fields of launch vehicle overall design, launch vehicle dynamics and control. This invention proposes for the first time a dual-center frequency adaptive filter that can simultaneously identify two elastic mode frequencies, completely avoiding frequency crossover problems. It has good application effects in handling the adaptive suppression of second and third-order elastic modes throughout the rocket. Furthermore, unlike previous series notch filtering approaches, this invention proposes an elastic suppression strategy based on additive compensation, adding a compensation signal to the original control loop. This compensation signal can be used to control the impact of this loop on the original control system scheme through amplitude limiting and other means. It has a good application foundation in terms of safety and stability, and has good engineering application prospects. Moreover, it has good fault-tolerant adaptability to changes in rocket dynamic characteristics under differences in space and ground design and fault conditions.

Description

Technical Field

[0001] This invention relates to a method and system for ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control of launch vehicles, belonging to the fields of overall launch vehicle design, launch vehicle dynamics and control. Background Technology

[0002] Large liquid-fueled clustered launch vehicles are subject to propellant sloshing, elastic vibrations of the rocket body, and other strong external disturbances, including aerodynamic forces, during flight. Their attitude control faces challenges such as multimodal coupling and severe channel cross-linking. As launch vehicle size increases and slenderness ratios improve, the first-order elastic mode frequency of the rocket body decreases sharply. Taking a series of clustered rocket configurations as an example, the elastic frequency has reached the lowest level among large launch vehicles worldwide, making the stable control of mid-to-low frequency rigid-sloshing elastic modes a design challenge.

[0003] Due to the differences between space and ground and limitations of ground testing conditions, model parameters often cannot be accurately predicted. Ground design deviations are conservatively chosen, and using a single controller to adapt to various combinations of deviations throughout the flight phase often increases controller complexity and reduces controller performance. Furthermore, with technological advancements, major engineering launch missions place higher demands on the reliability and safety of launch vehicles, requiring rocket systems to have fault tolerance capabilities that allow normal operation even in the event of a single failure. In summary, the attitude control system design of fourth-generation launch vehicles must, while meeting the performance requirements of traditional systems, possess stronger deviation adaptability and fault tolerance capabilities, and retain sufficient stability margins to accommodate uncertainties such as nonlinearity, channel crosstalk, and unmodeled disturbances.

[0004] Currently, adaptive notch filters used in engineering can only handle adaptive filtering at the first-order center frequency. Suppression of multi-order elastic signals is achieved through filter cascading, but this method inevitably involves center frequency crossovers. During these crossover periods, the notch filter's center frequency becomes unstable, potentially affecting the filtering effect and even system stability. Therefore, this invention proposes an adaptive attitude control method suitable for ultra-low frequency elastic control of launch vehicles, which, compared to traditional control methods, exhibits stronger adaptability to dynamic inconsistencies and typical power system faults. Summary of the Invention

[0005] The purpose of this invention is to overcome the above-mentioned shortcomings of the prior art and provide a method and system for ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control of launch vehicles, solving the problem of undesignable ultra-low frequency elastic control of launch vehicles. This method uses a dual-center frequency notch filter to achieve online identification of the elastic frequency and real-time compensation of the elastic information in the control loop to realize the adaptive adjustment of the controller, thereby improving the robustness of the system.

[0006] The above-mentioned objectives of the present invention are mainly achieved through the following technical solutions:

[0007] A method for ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control of a launch vehicle includes:

[0008] Remove rigid body motion information from the identification signal;

[0009] The identification signal, after removing rigid body motion information, is filtered to remove low-frequency and high-frequency components, while retaining information within the required frequency band.

[0010] For the filtered identification signal, the parameters of the dual-center frequency adaptive notch filter are updated using a recursive least squares algorithm with a forgetting factor to obtain the two highest-energy mode frequencies in the identification signal.

[0011] The two-order modal frequencies are updated or maintained in the following way: when the vibration magnitude is greater than the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the identification frequency of the current beat; otherwise, the notch filter uses the identification frequency of the previous beat.

[0012] A reference signal is generated using the updated or maintained two-order mode frequencies, and an elastic compensation signal is generated based on the reference signal using a minimum mean square error adaptive filtering algorithm.

[0013] The elastic compensation signal is incorporated into the control swing angle command to achieve adaptive fault-tolerant attitude stability control of the rocket body.

[0014] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, the removal of rigid body motion information from the identification signal includes:

[0015] If multiple rate gyroscopes are installed throughout the launch vehicle, the angular velocity information measured by any two rate gyroscopes installed at different locations is subtracted to remove the rigid body motion signal.

[0016] If the launch vehicle is equipped with only one rate gyroscope, the angular velocity calculated by the inertial navigation system is subtracted from the angular velocity information measured by the rate gyroscope to remove the rigid body motion signal;

[0017] If the launch vehicle is not equipped with a rate gyroscope, the angular velocity is calculated using an inertial navigation system and the rigid body motion signal is filtered out using a high-pass filter.

[0018] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, a bandpass filter is used to filter the identification signal after removing rigid body motion information. The form of the bandpass filter is as follows:

[0019]

[0020] in, B=ω high-ω low ω low To preserve the lower boundary frequency of the bandpass filter band, ω high The upper boundary frequency of the bandpass filter's reserved frequency band is s, where s is a Laplace variation complex frequency domain variable.

[0021] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, a recursive least squares algorithm with a forgetting factor is used to update the parameters of the dual-center frequency adaptive notch filter to obtain the two highest-energy mode frequencies in the identified signal, including:

[0022] (1) Initialize the cross-correlation vector by letting z(0) = 0; initialize the correlation matrix by letting Φ(0) = δI, where δ is the regularization parameter and I is the unit diagonal matrix;

[0023] (2) Based on the filtered identification signal, update the weight matrix input correlation matrix Φ(n). The recursive formula for the correlation matrix Φ(n) is:

[0024]

[0025] Where λ is the exponentially weighted forgetting factor, and λ is a number close to 1 but less than 1; where x(n), x(n-1), x(n-2), and x(n-3) are the outputs of the filter poles at the nth, (n-1), (n-2), and (n-3)th beats, respectively.

[0026] (3) Based on the filtered identification signal, update the cross-correlation vector z(n) between the weight input and the desired response. The recursive update formula for the cross-correlation vector z(n) is as follows:

[0027]

[0028] Where x(n-4) is the output of the filter pole part in the (n-4)th step;

[0029] (4) Update the recursive least squares weight vector based on the correlation matrix Φ(n) and the cross-correlation vector z(n). The updated formula is:

[0030]

[0031] (5) Based on the recursive least squares weight vector The characteristic parameters a1(n) and a2(n) of the notch filter are calculated using the following formulas:

[0032]

[0033] The two-mode frequencies identified by the filter are then obtained, as expressed below:

[0034]

[0035] Where ΔT is the sampling time.

[0036] In the above-mentioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, at least one of the following conditions must be met:

[0037] The regularization parameter δ is related to the signal-to-noise ratio (SNR), taking a smaller value for high SNR and a larger value for low SNR. δ is typically set to 1 × 10⁻⁶. -4 ;

[0038] The smaller the value of the exponentially weighted forgetting factor λ, the faster the forgetting occurs and the smoother the recognition frequency. The value of λ ranges from 0.95 to 0.99.

[0039] The sampling time ΔT is set to the control period of 0.02s.

[0040] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stability control method for launch vehicles, the vibration magnitude is determined using the output value y of the spectral damper. s Alternatively, the amplitude of the input signal, |y(n)|, can be identified by frequency.

[0041] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, the reference signal is generated using the updated or maintained second-order modal frequencies according to the following formula:

[0042]

[0043] Where d(n) is the reference signal at the nth time step, and A is the amplitude of the reference signal. The two-mode frequencies are updated or maintained, where t is time, ΔT is the sampling time, and δ is the sampling pulse function.

[0044] In the aforementioned ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles, an elastic compensation signal is generated based on the reference signal using a minimum mean square error adaptive filtering algorithm, including:

[0045] (1) Define the weight vector

[0046]

[0047] Where M is the filter order, let the initial weight vector be...

[0048] (2) Calculate the estimation error e(n) between the input signal and the output signal:

[0049] e(n) = u(n) - δ ek (n)

[0050] Where e(n) is the error signal at the nth time step, δ ek (n) represents the frequency of the nth time step. The k-th order elastic compensation signal, u(n) is the residual elastic signal after deducting the information filtered by the dual center frequency filter from the identification signal after filtering, and is the M×1 input vector at the n-th time step;

[0051] (3) Based on the weight vector Given the error signal e(n), calculate the recursive weight vector under stochastic gradient descent. The recursive formula for the weight vector is:

[0052] w(n+1)=w(n)+μu(n)e(n)

[0053] Where μ is the calculation step size;

[0054] (4) Calculate the filter output at time step n+1 based on the reference signal d(n):

[0055] δ ek (n+1)=w T (n)d(n)

[0056] Where, δ ek (n+1) represents the frequency of the (n+1)th time step. The compensation signal for the k-th order elasticity; T is the transpose.

[0057] A launch vehicle ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control system includes:

[0058] The information removal module removes rigid body motion information from the identification signal;

[0059] The filtering module filters the identification signal after removing rigid body motion information, filtering out low-frequency and high-frequency components while retaining information within the required frequency band.

[0060] The modal frequency acquisition and adaptive filtering module updates the parameters of the dual-center frequency adaptive notch filter using a recursive least squares algorithm with a forgetting factor for the filtered identification signal, thereby obtaining the two highest-energy modal frequencies in the identification signal.

[0061] The frequency update or hold module updates or holds the two-order modal frequencies in the following way: when the vibration magnitude is greater than the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the identification frequency of the current beat; otherwise, the notch filter uses the identification frequency of the previous beat.

[0062] The compensation signal generation module generates a reference signal using the updated or maintained two-order mode frequencies, and generates an elastic compensation signal based on the reference signal using a minimum mean square error adaptive filtering algorithm.

[0063] The signal compensation module compensates the elastic compensation signal into the control swing angle command to achieve adaptive fault-tolerant attitude stability control of the rocket body.

[0064] A computer device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described above.

[0065] Compared with the prior art, the present invention has at least the following beneficial effects:

[0066] (1) This invention proposes a dual-center frequency adaptive filter for the first time, which can simultaneously identify the frequencies of two elastic modes in parallel, completely avoiding the frequency crossover problem. It has a good application effect on the adaptive suppression of the second and third elastic modes of the whole rocket. In addition, unlike the previous series notch filtering approach, this invention proposes an elastic suppression strategy based on the idea of ​​additive compensation. It adds a compensation signal to the original control loop, and can control the influence of the loop on the original control system scheme by means of amplitude limiting, etc. It has a good application foundation in terms of safety and stability, and has good engineering application prospects.

[0067] (2) This invention relates to an adaptive attitude control method applicable to ultra-low frequency elastic control of launch vehicles, which has good fault tolerance and adaptability to changes in the dynamic characteristics of the rocket body under the differences between the ground and space design and faults. Compared with traditional control methods, it has stronger adaptability to the inconsistency between dynamics and space and typical power system faults. Attached Figure Description

[0068] Figure 1 This is a control scheme architecture diagram in an embodiment of the present invention;

[0069] Figure 2 This is a schematic diagram of the spectrum damper in step four of this embodiment of the invention;

[0070] Figure 3 This is a schematic diagram of the frequency update and maintenance module in step four of this embodiment of the invention;

[0071] Figure 4 This is an example diagram of algorithm simulation verification in an embodiment of the present invention;

[0072] Figure 5 This is an example diagram of algorithm simulation verification in an embodiment of the present invention. Detailed Implementation

[0073] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments:

[0074] like Figure 1 As shown in the figure, the ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles in this embodiment of the invention specifically includes the following steps:

[0075] Step 1: Remove rigid body trend terms;

[0076] To ensure the accuracy of frequency estimation, it is necessary to eliminate rigid body motion information in the identification signal and retain only elastic information.

[0077] If multiple rate gyroscopes are installed throughout the launch vehicle, the angular velocity information measured by any two rate gyroscopes installed at different locations is subtracted to remove the rigid body motion signal.

[0078] If the launch vehicle is equipped with only one rate gyroscope, the angular velocity calculated by the inertial navigation system is subtracted from the angular velocity information measured by the rate gyroscope to remove the rigid body motion signal;

[0079] If the launch vehicle is not equipped with a rate gyroscope, the angular velocity is calculated using an inertial navigation system and the rigid body motion signal is filtered out using a high-pass filter.

[0080] Step 2: Filter the identification signal after removing rigid body motion information based on the design prior information.

[0081] Preprocessing the input data using a bandpass filter to remove low-frequency and high-frequency components and retain only information within the required frequency band can effectively improve the accuracy and continuity of frequency identification. The bandpass filter takes the following form:

[0082]

[0083] In the formula, B=ω high -ω low ω low To preserve the lower boundary frequency of the bandpass filter band, ω high The upper boundary frequency of the bandpass filter's reserved frequency band is s, where s is a Laplace variation complex frequency domain variable.

[0084] Step 3: For the filtered identification signal, update the parameters of the dual-center-frequency adaptive notch filter using the recursive least squares algorithm (RLS) with a forgetting factor. Furthermore, the two highest-energy mode frequencies in the signal were identified;

[0085] The recursive least squares algorithm calculation process is as follows:

[0086] (1) Algorithm initialization: Initialize the cross-correlation vector z(0) = 0; initialize the correlation matrix, Φ(0) = δI, where δ is a regularization parameter, set to be related to the signal-to-noise ratio (SNR), taking a small value for high SNR and a larger value for low SNR. In this embodiment, the value is 1 × 10. -4 I is a unit diagonal matrix.

[0087] (2) Based on the filtered identification signal, update the weight matrix input correlation matrix Φ(n). The correlation matrix recursive formula is as follows:

[0088]

[0089] In the formula, λ is the exponentially weighted forgetting factor. λ is a number close to 1 but less than 1. The smaller the value of λ, the faster the forgetting and the smoother the recognition frequency. In this embodiment, the value is taken as 0.95 to 0.99. x(n), x(n-1), x(n-2), and x(n-3) are the outputs of the filter poles of the nth, n-1th, n-2th, and n-3th beats, respectively.

[0090] (3) Based on the filtered identification signal, update the cross-correlation vector z(n) between the weight input and the desired response. The recursive update formula for z(n) is:

[0091]

[0092] In the formula, x(n-4) is the output of the filter pole part at the (n-4)th step.

[0093] (4) Update the recursive least squares weight vector based on the correlation matrix Φ(n) and the cross-correlation vector z(n). The updated formula is:

[0094]

[0095] (5) Based on the recursive least squares weight vector The characteristic parameters a1(n) and a2(n) of the notch filter are calculated using the following formulas:

[0096]

[0097] The two-mode frequencies identified by the filter are then obtained, as expressed below:

[0098]

[0099] In the formula, ΔT is the sampling time, which is taken as 0.02s in this embodiment. The identification frequency results are output in ascending order.

[0100] Step 4: Update or maintain the frequencies of the two modal frequencies.

[0101] When the vibration magnitude of the rocket is small, the frequency identification module will identify inelastic signals; however, the signal components in the rocket attitude control loop are complex, and the vibration frequency identification value is unstable. Therefore, the following improvement strategy is adopted for adaptive notch filter control:

[0102] When the vibration magnitude exceeds the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the current modal frequency; otherwise, the notch filter uses the modal frequency of the previous cycle. The rate of change of the modal frequency is the difference between the modal frequency of the current cycle and the modal frequency of the previous cycle divided by the time period. See the schematic diagram of the frequency update and hold module. Figure 3 .

[0103] The magnitude of vibration can be determined using the output value y of the spectral damper. s Spectrum damper forms such as Figure 2 As shown, the amplitude |y(n)| of the input signal can also be identified by frequency. The criterion for judging whether the frequency estimate is stable is to judge the magnitude of the frequency increment.

[0104] Step 5: Generate a reference signal using the updated or maintained second-order mode frequencies, as shown in the following formula:

[0105]

[0106] Where d(n) is the reference signal at the nth time step, and A is the amplitude of the reference signal. The two-mode frequencies are updated or maintained, where t is time, ΔT is the sampling time, and δ is the sampling pulse function.

[0107] The frequency is generated using the minimum mean square error (LMS) adaptive filtering algorithm. The compensation signal δ of the k-th order elasticity ek ;

[0108] The calculation process of the LMS algorithm is as follows:

[0109] (1) Define the weight vector

[0110]

[0111] In the formula, M represents the filter order, and in this embodiment, M = 5; the weight vector is initialized.

[0112] (2) Calculate the estimation error e(n) between the input signal and the output signal:

[0113] e(n) = u(n) - δ ek (n)

[0114] In the formula, e(n) is the error signal at the nth time step, δ ek (n) represents the frequency of the nth time step. The k-th order elastic compensation signal; u(n) is the residual elastic signal after deducting the information filtered by the dual center frequency filter from the identification signal after filtering; and is the M×1 input vector at the n-th time step.

[0115] (3) Based on the weight vector Given the error signal e(n), calculate the recursive weight vector under stochastic gradient descent. The recursive formula for the weight vector is:

[0116] w(n+1)=w(n)+μu(n)e(n)

[0117] In the formula, μ is the calculation step size;

[0118] (4) Calculate the filter output at time step (n+1) based on the reference signal d(n):

[0119] δ ek (n+1)=w T (n)d(n)

[0120] Where, δ ek (n+1) represents the frequency of the (n+1)th time step. The compensation signal for the k-th order elasticity; T is the transpose.

[0121] Step Six: Calculate the swing angle compensation command δ for both the two-order characteristic frequencies identified in Steps Three and Four using the algorithm from Step Five. e1 and δ e2 The elastic compensation signal δ e1 and δ e2 The compensation is incorporated into the control angle command, thereby achieving adaptive fault-tolerant attitude stability control of the rocket body's elastic vibration. Simulation verification of the control method proposed in this embodiment is shown below. Figure 4 , Figure 5 Adaptive control strategies can effectively improve system robustness.

[0122] This invention also provides an ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control system for launch vehicles, comprising:

[0123] The information removal module removes rigid body motion information from the identification signal;

[0124] The filtering module filters the identification signal after removing rigid body motion information, filtering out low-frequency and high-frequency components while retaining information within the required frequency band.

[0125] The modal frequency acquisition and adaptive filtering module updates the parameters of the dual-center frequency adaptive notch filter using a recursive least squares algorithm with a forgetting factor for the filtered identification signal to obtain the two highest-energy modal frequencies in the identification signal; and performs adaptive notch filtering on the filtered identification signal to extract the elastic signal.

[0126] The frequency update or hold module updates or holds the two-order modal frequencies in the following way: when the vibration magnitude is greater than the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the identification frequency of the current beat; otherwise, the notch filter uses the identification frequency of the previous beat.

[0127] The compensation signal generation module generates a reference signal using the updated or maintained two-order mode frequencies, and generates an elastic compensation signal based on the reference signal using a minimum mean square error adaptive filtering algorithm.

[0128] The signal compensation module compensates the elastic compensation signal into the control swing angle command to achieve adaptive fault-tolerant attitude stability control of the rocket body.

[0129] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0130] The above description is only the best specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the protection scope of the present invention.

[0131] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A method for ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control of a launch vehicle, characterized in that, include: Remove rigid body motion information from the identification signal; The identification signal, after removing rigid body motion information, is filtered to remove low-frequency and high-frequency components, while retaining information within the required frequency band. For the filtered identification signal, the parameters of the dual-center frequency adaptive notch filter are updated using a recursive least squares algorithm with a forgetting factor to obtain the two highest-energy mode frequencies in the identification signal. The two-order modal frequencies are updated or maintained in the following way: when the vibration magnitude is greater than the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the identification frequency of the current beat; otherwise, the notch filter uses the identification frequency of the previous beat. A reference signal is generated using the updated or maintained two-order mode frequencies, and an elastic compensation signal is generated based on the reference signal using a minimum mean square error adaptive filtering algorithm. The elastic compensation signal is incorporated into the control swing angle command to achieve adaptive fault-tolerant attitude stabilization control of the rocket body. The parameters of the dual-center frequency adaptive notch filter are updated using a recursive least squares algorithm with a forgetting factor to obtain the two highest-energy mode frequencies in the identified signal, including: (1) Initialize the cross-correlation vector, let Initialize the correlation matrix, let ,in For regularization parameters, It is a unit diagonal matrix; (2) Update the weight matrix and input correlation matrix based on the filtered identification signal. Correlation matrix The recursive formula is: in, For the exponentially weighted forgetting factor, Take numbers that are close to 1 but less than 1; among them , , , These are the outputs of the filter poles for the nth, (n-1)th, (n-2)th, and (n-3)th time steps, respectively. (3) Update the cross-correlation vector between the weight input and the desired response based on the filtered identification signal. Cross-correlation vector The recursive update formula is: in, This is the output of the filter pole portion at the (n-4)th pulse. (4) Based on the correlation matrix and cross-correlation vector Update the recursive least squares weight vector The updated formula is: (5) Based on the recursive least squares weight vector Calculate the characteristic parameters of the notch filter. The calculation formula is: The two-mode frequencies identified by the filter are then obtained, as expressed below: in, Sampling time.

2. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 1, characterized in that, The removal of rigid body motion information from the identification signal includes: If multiple rate gyroscopes are installed throughout the launch vehicle, the angular velocity information measured by any two rate gyroscopes installed at different locations is subtracted to remove the rigid body motion signal. If the launch vehicle is equipped with only one rate gyroscope, the angular velocity calculated by the inertial navigation system is subtracted from the angular velocity information measured by the rate gyroscope to remove the rigid body motion signal; If the launch vehicle is not equipped with a rate gyroscope, the angular velocity is calculated using an inertial navigation system and the rigid body motion signal is filtered out using a high-pass filter.

3. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 1, characterized in that, A bandpass filter is used to filter the identification signal after removing rigid body motion information. The form of the bandpass filter is as follows: in, , , To preserve the lower boundary frequency of the bandpass filter band, To preserve the upper boundary frequency of the bandpass filter band, For Lagrange transformations, the complex frequency domain variable is used.

4. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 1, characterized in that, At least one of the following must be met: Regularization parameters It is related to the signal-to-noise ratio (SNR), taking a smaller value for high SNR and a larger value for low SNR. The value is 1×10 -4 ; Exponentially weighted forgetting factor Smaller values ​​indicate faster forgetting and a smoother recognition frequency. The value ranges from 0.95 to 0.99; Sampling time The control cycle is set to 0.02s.

5. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 1, characterized in that, The vibration magnitude is determined using the output value of the spectral damper. Alternatively, the amplitude of the input signal can be identified by frequency. .

6. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 1, characterized in that, The reference signal is generated using the updated or maintained two-mode frequencies according to the following formula: in, For the first Reference signal for time step, Reference signal amplitude, To update or maintain the two-order modal frequencies, For time, For the time of adoption; This is the sampling pulse function.

7. The ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control method for launch vehicles according to claim 6, characterized in that, Based on the reference signal, an elastic compensation signal is generated using a minimum mean square error adaptive filtering algorithm, including: (1) Define the weight vector : in, Let the filter order be denoted by and the initial weight vector be denoted by . ; (2) Calculate the estimation error between the input signal and the output signal. : in, For the first Error signal of time step, For the first The frequency of time steps is The k-th order elastic compensation signal, To subtract the residual elastic signal from the information filtered by the dual-center frequency filter from the filtered identification signal, the result is the first... time step Input vector; (3) Based on the weight vector Sum of error signals Calculate the recursive weight vector under stochastic gradient descent. The recursive formula for the weight vector is: in, To calculate the step size; (4) Based on the reference signal Calculate the first Time step filter output: in, For the first The frequency of time steps is The compensation signal for the k-th order elasticity; This is a transpose.

8. A launch vehicle ultra-low frequency elastic adaptive fault-tolerant attitude stabilization control system, characterized in that, include: The information removal module removes rigid body motion information from the identification signal; The filtering module filters the identification signal after removing rigid body motion information, filtering out low-frequency and high-frequency components while retaining information within the required frequency band. The modal frequency acquisition and adaptive filtering module updates the parameters of the dual-center frequency adaptive notch filter using a recursive least squares algorithm with a forgetting factor for the filtered identification signal, thereby obtaining the two highest-energy modal frequencies in the identification signal. The frequency update or hold module updates or holds the two-order modal frequencies in the following way: when the vibration magnitude is greater than the required value and the rate of change of the modal frequency is less than the required value, the notch filter uses the identification frequency of the current beat; otherwise, the notch filter uses the identification frequency of the previous beat. The compensation signal generation module generates a reference signal using the updated or maintained two-order mode frequencies, and generates an elastic compensation signal based on the reference signal using a minimum mean square error adaptive filtering algorithm. The signal compensation module compensates the elastic compensation signal into the control swing angle command to realize the adaptive fault-tolerant attitude stability control of the rocket body; The parameters of the dual-center frequency adaptive notch filter are updated using a recursive least squares algorithm with a forgetting factor to obtain the two highest-energy mode frequencies in the identified signal, including: (1) Initialize the cross-correlation vector, let Initialize the correlation matrix, let ,in For regularization parameters, It is a unit diagonal matrix; (2) Update the weight matrix and input correlation matrix based on the filtered identification signal. Correlation matrix The recursive formula is: in, For the exponentially weighted forgetting factor, Take numbers that are close to 1 but less than 1; among them , , , These are the outputs of the filter poles for the nth, (n-1)th, (n-2)th, and (n-3)th time steps, respectively. (3) Update the cross-correlation vector between the weight input and the desired response based on the filtered identification signal. Cross-correlation vector The recursive update formula is: in, This is the output of the filter pole portion at the (n-4)th pulse. (4) Based on the correlation matrix and cross-correlation vector Update the recursive least squares weight vector The updated formula is: (5) Based on the recursive least squares weight vector Calculate the characteristic parameters of the notch filter. The calculation formula is: The two-mode frequencies identified by the filter are then obtained, as expressed below: in, Sampling time.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method of claim 1.

Images