A method for active control of broadband micro-vibration containing unknown harmonics

Abstract

The application discloses a kind of active control methods for containing unknown harmonic broadband microvibration, it is related to spacecraft microvibration control technical field, including the following steps: using acceleration sensor obtains the residual acceleration response at spacecraft load;Based on residual acceleration response, using PID controller obtains first control force;Using reference signal generator, harmonic frequency is roughly estimated, refined estimation and fine adjustment are carried out, and reference signal is generated;Based on reference signal, using parallel filter-based least mean square algorithm obtains second control force;According to first control force and second control force, total control force is obtained, and is output by executing mechanism, realizes the efficient suppression to containing multiple unknown harmonic broadband microvibration.The application solves the problem that traditional vibration control method is insufficient for harmonic disturbance control ability, and filter-based least mean square algorithm is difficult to obtain the reference signal meeting the requirements under the condition that disturbance characteristics are unknown.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of spacecraft micro-vibration control, and particularly relates to a wide-frequency micro-vibration active control method for unknown harmonics. BACKGROUND

[0002] In recent years, the payloads carried on the spacecraft are more and more sensitive to the micro-vibration environment, and various moving parts need to be equipped on the spacecraft to ensure the normal on-orbit operation of the spacecraft, such as a refrigerator, a reaction flywheel, a flexible deployable mechanism and the like. These parts will inevitably generate random vibration and harmonic vibration when working, which will affect the normal work of the load after being transmitted to the load. In order to meet the stringent requirements of the sensitive load on the micro-vibration environment, a vibration isolation device is installed between the spacecraft and the load to reduce the influence of the spacecraft disturbance on the performance of the load. The vibration control algorithm is the key to efficient isolation of the vibration isolation device and is the prerequisite for ensuring the ultra-static and ultra-stable working environment of the load.

[0003] Considering the complexity of the micro-vibration environment on the spacecraft, the active vibration isolation method has been widely applied. According to the structure of the control algorithm, the active vibration isolation algorithm can be generally divided into feedback control and feedforward control. The feedback control can realize the conventional isolation of the random vibration and has the wide-band vibration control capability, such as the commonly used PID control and its derivative method, but the control gain of the feedback control at the harmonic disturbance is limited, especially when the resonant frequency is close to the bandwidth boundary of the feedback control, the control performance is limited. The feedforward control has good control effect on single-frequency resonance through modeling the dynamics of the forward channel and measuring the vibration of the vibration source. The filter-based least mean square algorithm can realize efficient suppression of the harmonic disturbance and does not affect the signals at other frequencies, and is a commonly used narrow-band disturbance control method, but this algorithm needs a reference signal related to the disturbance, and it is difficult to obtain a reference signal meeting the requirements in the case of unknown harmonic disturbance characteristics. In the face of the wide-frequency micro-vibration environment containing unknown harmonics on the spacecraft, an efficient active control algorithm is needed to realize the wide-frequency vibration isolation and accurately estimate the harmonic disturbance characteristics to further efficiently suppress the harmonic disturbance. SUMMARY

[0004] In view of the above problems in the prior art, the wide-frequency micro-vibration active control method for unknown harmonics provided by the present application solves the problems of insufficient harmonic disturbance control capability of the traditional vibration control method and the difficulty of obtaining a reference signal meeting the requirements in the case of unknown disturbance characteristics based on the filter-based least mean square algorithm.

[0005] In order to achieve the above application purpose, the technical scheme adopted by the present application is as follows: a wide-frequency micro-vibration active control method for unknown harmonics, comprising the following steps:

[0006] S1: obtaining residual acceleration response at spacecraft load by using acceleration sensor;

[0007] S2: obtaining first control force by using PID controller based on residual acceleration response;

[0008] S3: performing coarse estimation, refined estimation and fine adjustment on harmonic frequency by using reference signal generator based on residual acceleration response, and generating reference signal;

[0009] S4: obtaining second control force by using parallel filter-based least mean square algorithm based on reference signal;

[0010] S5: obtaining total control force according to first control force and second control force, and outputting by executing mechanism, realizing efficient suppression on wideband micro-vibration containing multiple unknown harmonics, and completing active control on wideband micro-vibration containing unknown harmonics.

[0011] The beneficial effects of the above scheme are: the application proposes an active control method for wideband micro-vibration containing unknown harmonics, which enhances the suppression ability of harmonic disturbance on the basis of realizing wideband vibration control by comprehensively designing PID controller and parallel filter-based least mean square algorithm. In order to realize the estimation of harmonic disturbance frequency characteristics, a reference signal generator based on triangular cascade structure adaptive notch filter (TANF), Kalman filter and autoregressive model is designed, which can accurately estimate the harmonic disturbance frequency according to the residual micro-vibration information at the load in the case of unknown harmonic disturbance, and generate relevant reference signal for parallel filter-based least mean square algorithm. It can solve the problems of insufficient control ability of traditional vibration control method on harmonic disturbance and generation of reference signal of filter-based least mean square algorithm in the case of unknown disturbance characteristics.

[0012] Further, the residual acceleration response at the spacecraft load in S1 is The calculation formula is:

[0013]

[0014] Wherein, r(k) is the acceleration response of control force, d(k) is the acceleration response of disturbance, and n(k) is the measurement noise.

[0015] Further, the first control force F PID (k) in S2 is:

[0016]

[0017] Wherein, k p , k i and k d are proportional coefficient, integral coefficient and differential coefficient respectively, and Nd s is a Laplace operator, is a residual acceleration response.

[0018] The above further scheme has the beneficial effect that the first stage of control is to achieve wide-band vibration suppression through the PID controller, and according to the measured acceleration information, the control force is output through the PID controller to effectively suppress the disturbance in the control bandwidth, but in this process, multiple harmonic disturbances in the disturbance cannot be fully suppressed.

[0019] Further, the reference signal generator in S3 includes a triangular cascade structure adaptive notch filter, a Kalman filter and an autoregressive model, and the S3 includes the following steps:

[0020] S3-1: Based on the residual acceleration response, the triangular cascade structure adaptive notch filter is used to estimate the harmonic frequency in real time;

[0021] S3-2: The Kalman filter is used to refine the estimated harmonic frequency to ensure that the frequency error is within the allowable range of the initial value error of the autoregressive model;

[0022] S3-3: The frequency value output by the Kalman filter at the stop time is taken as the initial value of the autoregressive model, and based on the initial value and the residual acceleration response, the autoregressive model is used to iteratively and finely adjust the estimated frequency to generate a reference signal.

[0023] The above further scheme has the beneficial effect that the reference signal generation module is composed of a triangular cascade structure adaptive notch filter, a Kalman filter and an autoregressive model through the above technical scheme. The triangular cascade structure adaptive notch filter estimates the coarse estimate value of the harmonic disturbance frequency at k time according to the residual disturbance signal after PID control; the Kalman filter is used to reduce the fluctuation of the estimated value caused by noise, and the obtained result is taken as the initial value of the autoregressive model to avoid system divergence; the autoregressive model adjusts the frequency finely according to the initial value provided by the Kalman filter and the error signal of the control system to obtain a fine estimate value of the frequency, and generates a related reference signal for the parallel filter-based least mean square algorithm in the controller to achieve suppression of the harmonic disturbance.

[0024] Further, the triangular cascade structure adaptive notch filter H L (z) in S3-1 has the following calculation formula:

[0025]

[0026] ρ = -cos(ω)

[0027]

[0028] where z -1and z -2 are delay operators, p is the cosine value of the notch angle frequency of the notch filter, a is the filter factor, w is the notch angle frequency of the notch filter, f is the target frequency, f s is the sampling frequency,

[0029] The square of the output value of the triangular cascade structure adaptive notch filter is taken as the optimization target, and the gradient descent method is used to obtain the coarse estimated harmonic frequency and parameter update of the triangular cascade structure adaptive notch filter, and the formula is:

[0030] w anf (k) = arccos[-p(k)]

[0031] p(k+1) = p(k) - 2m anf |e anf (k) = y(k) - h(k-1)

[0032] y(k) = h(k-1) - (1+a)e anf (k-1)

[0033] where w anf (·) is the coarse estimated harmonic frequency of the notch filter, k is the time, p(·) is the cosine function of the notch angle frequency of the notch filter, m anf is the step size of the notch filter to be updated, e anf (·) is the output of the notch filter, y(·) is the derivative of the output of the notch filter with respect to p, h(·) is the input of a single adaptive notch filter.

[0034] The beneficial effects of the above further scheme are: the residual disturbance at the load after PID control is mainly composed of harmonic signals that are not fully suppressed, and the above steps provide refined initial values for the Kalman filter through fast frequency coarse identification of the signal by the triangular cascade structure adaptive notch filter.

[0035] Further, the calculation formula of the Kalman filter in S3-2 is:

[0036]

[0037] P - (k) = P(k-1) + Q

[0038]

[0039] P(k) = P - (k) - K(k)·P - (k)

[0040] where, is the prior estimated frequency vector of the Kalman filter, w Kal(k-1|k-1) is the calculation value of the Kalman filter at k-1 time, P - (·) is the predicted error covariance matrix, P(·) is the error covariance matrix, Q and R are the measurement and process noise covariance matrices, respectively, K(·) is the Kalman gain matrix, ω Kal (k|k) is the calculation value of the Kalman filter at k time.

[0041] The beneficial effect of the above further scheme is that due to the influence of noise in the signal, the frequency fluctuation estimated by the triangular cascade structure adaptive notch filter is large, and directly introducing the autoregressive model may lead to divergence of the autoregressive model, the rough identification value of the frequency is refined by the Kalman filter, the fluctuation is reduced, the anti-interference ability of the system is improved, and the reliability of the method under low signal-to-noise ratio is ensured.

[0042] Further, the reference signal generated in S3-3 is:

[0043]

[0044] Wherein, x i (·) is the reference signal generated by the autoregressive model, is the parameter related to the frequency in the ith autoregressive model, N anf is the time when the triangular cascade structure adaptive notch filter and the Kalman filter stop working, is the initial value of the autoregressive model, is the step size of the ith autoregressive model, w i0 (·) is the first coefficient of the ith adaptive filter in the parallel filter-based least mean square algorithm, is the first coefficient of the finite impulse response filter model.

[0045] The beneficial effect of the above further scheme is that the autoregressive model takes the frequency refined by the Kalman filter as the initial value of iteration, generates the corresponding reference signal to fine-tune the frequency in combination with the residual disturbance information, and generates the reference signal for the parallel filter-based least mean square algorithm and continuously iterates and updates.

[0046] Further, the second control force F LMS (k) in S4 is:

[0047]

[0048] Wherein, n is the number of parallel filter-based least mean square algorithms, W i (·) is the ith adaptive filter, is the finite impulse response filter model of the transfer function from force to load plane acceleration, W i (·) is the coefficient vector of the ith adaptive filter, μi Step size of the i-th module of the parallel filter-based least mean square algorithm, i-th filtered reference signal.

[0049] Further, the total control force F c (k) is:

[0050] F c (k) = F PID (k) + F LMS (k).

[0051] The above further scheme has the beneficial effect that the reference signal is generated after the autoregressive model is started, and the parallel filter-based least mean square algorithm receives the reference signal to calculate the corresponding control force, and the control force generated by the PID controller is output by the actuator together to achieve efficient suppression of harmonics in the residual disturbance. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 A flow chart of a wideband micro-vibration active control method for containing unknown harmonics.

[0053] Figure 2 A system block diagram of a wideband micro-vibration active control method for containing unknown harmonics.

[0054] Figure 3 A reference signal generator block diagram.

[0055] Figure 4 A triangular cascade structure adaptive notch filter block diagram.

[0056] Figure 5 A single adaptive notch filter block diagram.

[0057] Figure 6 A vibration control effect diagram in the time domain of the embodiment of the present application.

[0058] Figure 7 A vibration control effect diagram in the frequency domain of the embodiment of the present application.

[0059] Figure 8 A frequency estimation result diagram of the embodiment of the present application. DETAILED DESCRIPTION

[0060] The present application will be further described below in conjunction with the drawings and specific embodiments.

[0061] As shown in Figure 1 and Figure 2 , a wideband micro-vibration active control method for containing unknown harmonics includes the following steps:

[0062] S1: Obtain the residual acceleration response at the spacecraft payload using an accelerometer;

[0063] S2: Based on the residual acceleration response, the first control force is obtained using a PID controller;

[0064] S3: Based on the residual acceleration response, the harmonic frequency is coarsely estimated, refined, and finely adjusted using a reference signal generator, and a reference signal is generated.

[0065] S4: Based on the reference signal, the second control force is obtained using a parallel filter-based minimum mean square algorithm;

[0066] S5: Based on the first control force and the second control force, the total control force is obtained and output by the actuator to achieve efficient suppression of broadband micro-vibrations containing multiple unknown harmonics, thus completing the active control of broadband micro-vibrations containing unknown harmonics.

[0067] This invention proposes an active control algorithm for broadband micro-vibrations containing unknown harmonics. By comprehensively designing a PID controller and a parallel, filter-based least mean square (LMS) algorithm, it enhances the suppression of harmonic disturbances while achieving broadband vibration control. To estimate the frequency characteristics of harmonic disturbances, a reference signal generator based on a triangular cascaded adaptive notch filter (TANF), a Kalman filter, and an autoregressive model is designed, such as... Figure 2 The diagram shown is a system control block diagram of the present invention, wherein: a0(k) is the bottom acceleration of the vibration isolation device; f(k) is the disturbance force acting on the load plane; P(z) is the transfer function from the bottom acceleration of the vibration isolation device to the acceleration of the load plane; S(z) is the transfer function from the force to the acceleration of the load plane. S(z) is the finite impulse response filter model; d(k) is the acceleration response generated by the disturbance in the load plane; F LMS (k) represents the control force (second control force) generated by the parallel filter-based minimum mean square algorithm; F PID (k) represents the control force (first control force) generated by the PID controller; F c (k) represents the total control force; r(k) represents the response of the load plane caused by the control force; and n(k) represents the measurement noise.

[0068] Residual acceleration response at the spacecraft payload in S1 The calculation formula is:

[0069]

[0070] Where r(k) is the acceleration response generated by the control force, d(k) is the acceleration response generated by the disturbance, and n(k) is the measurement noise.

[0071] The first control force F in S2 PID (k) is:

[0072]

[0073] Where, k p k i and k d These are the proportional coefficient, integral coefficient, and differential coefficient, respectively, N. d Here are the filter coefficients, and s is the Laplace operator. This is the residual acceleration response.

[0074] like Figure 3 As shown, the reference signal generator in S3 includes a triangular cascaded adaptive notch filter, a Kalman filter, and an autoregressive model. S3 includes the following sub-steps:

[0075] S3-1: Based on the residual acceleration response, the harmonic frequency is coarsely estimated in real time using a triangular cascaded adaptive notch filter.

[0076] S3-2: Use a Kalman filter to refine the coarsely estimated harmonic frequencies, ensuring that the frequency error is within the allowable range of the autoregressive model for the initial frequency value error.

[0077] S3-3: The frequency value output at the stopping moment of the Kalman filter is used as the initial value of the autoregressive model. Based on the initial value and the residual acceleration response, the estimated frequency is estimated by iterating and fine-tuning the autoregressive model to generate a reference signal.

[0078] S3-1 Triangular Cascaded Adaptive Notch Filter H L The formula for calculating (z) is:

[0079]

[0080] ρ=-cos(ω)

[0081]

[0082] Among them, z -1 and z -2 All are delay operators. ρ is the cosine of the notch filter's angular frequency, α is the filter factor (a positive value less than 1), and it is related to the width of the notch filter; the closer α is to 1, the narrower the notch filter, and the more significant the attenuation of the target frequency signal. ω is the notch filter's angular frequency, and f is the target frequency. s Sampling frequency,

[0083] With the optimization objective of minimizing the square of the output value of the triangular cascaded adaptive notch filter, the gradient descent method is used to obtain a coarse estimate of the harmonic frequency and parameter update of the triangular cascaded adaptive notch filter. The formula is as follows:

[0084] ω anf (k) = arccos[-p(k)]

[0085] p(k+1) = p(k) - 2μ anf |e anf (k) = y(k) - h(k-1)

[0086] y(k) = h(k-1) - (1+ a)e anf (k-1)

[0087] where ω anf (·) is the coarse estimate of harmonic frequency of the notch filter, k is the time instant, p(·) is the cosine function of the notch angle frequency of the notch filter, μ anf is the step size of the notch filter to be updated, e anf (·) is the output of the notch filter, y(·) is the derivative of the output of the notch filter with respect to p, h(·) is the input of a single adaptive notch filter.

[0088] After the action of the PID controller, the disturbance near the fundamental frequency is effectively suppressed, and the residual acceleration response at the load mainly contains the harmonic disturbance which is not sufficiently suppressed. The residual acceleration response enters the triangular cascade structure adaptive notch filter, and the harmonic frequency is estimated in real time.

[0089] In this embodiment, the triangular cascade structure adaptive notch filter for estimating three frequencies is taken as an example, and three groups of adaptive notch filters are needed at this time, as shown in FIG. 1. Figure 4 The parameters of the three adaptive notch filters H 11 (z), H 12 (z) and H 13 (z) in the first group are independently iteratively adjusted; in order to reduce the calculation amount, two adaptive notch filters are used in the second group, the input is the output of H 11 (z) in the first group of adaptive notch filters, and the parameters of H 13 (z) are directly copied to the notch filter H 22 (z), the parameters of H 23 (z) are independently adjusted; in the third group of adaptive notch filters, the parameters of H 31 (z) are copied from H 13 (z), the parameters of H 32 (z) are copied from H 23 (z), and the parameters of H 33 (z) are independently adjusted. According to the structure of the entire triangular cascade structure adaptive notch filter, the input signal of the first group of notch filters contains multiple harmonics, and the parameters of H 11 (z) and H 12For the center frequency of (z), its input is a low signal-to-noise ratio signal, thus an accurate frequency estimate cannot be obtained. However, as long as the input signal frequency is within the neighborhood of the notch filter's center frequency, the notch filter's response output will decrease significantly. Therefore, the input signal, after passing through H... 11 (z) and H 12 (z) after which, for the notch filter H 13 (z), whose input signal has a high signal-to-noise ratio, is passed through H 13 (z) A relatively accurate frequency estimate can be obtained. For the second set of notch filters, since the notch filter H... 22 (z) directly copied H 13 The parameters of (z), therefore when H 13 (z) When an accurate frequency estimate is obtained, the signal passes through H 22 The signal corresponding to (z) will be filtered out from the input signal, thus making the notch filter H... 23 The input signal (z) has a relatively high signal-to-noise ratio, enabling accurate frequency estimation. Similarly, for the third set of notch filters, H 33 (z) can also yield the corresponding third accurate frequency. By increasing the number of cascaded notch filters, a triangular cascaded adaptive notch filter structure can be obtained to estimate more frequencies. When the number of harmonics whose frequencies need to be estimated is n, n sets of adaptive notch filters are required, and the total number of notch filters is (n... 2 +3n-2) / 2, while the number of independently adjustable notch filters is only 2n-1.

[0090] The estimation principle of a single adaptive notch filter is as follows: Figure 5 As shown, the notch filter parameters are adjusted using an adaptive algorithm to minimize the notch filter output e. anf (k) When the notch filter output is minimized, the center frequency of the notch filter is the frequency to be estimated.

[0091] After the triangular cascaded adaptive notch filter estimates the harmonic frequencies in real time, the estimation results are then refined by a Kalman filter. According to... Figure 3 Thus, the n frequency values ​​estimated by the received triangular cascaded adaptive notch filter are obtained. At that time, the Kalman filter operates synchronously, refining the estimated frequency to obtain...

[0092] The formula for calculating the Kalman filter in S3-2 is:

[0093]

[0094] P - (k)=P(k-1)+Q

[0095]

[0096] P(k) = P - (k) - K(k) - P - (k)

[0097] wherein, is the prior estimation frequency vector of Kalman filter, ω Kal (k-1|k-1) is the calculated value of Kalman filter at k-1 time, P - (·) is the predicted error covariance matrix, P(·) is the error covariance matrix, Q and R are the measurement and process noise covariance matrices respectively, K(·) is the Kalman gain matrix, ω Kal (k|k) is the calculated value of Kalman filter at k time.

[0098] The matrices in the calculation formula of the Kalman filter are square matrices, and the dimension of the matrix and the length of the vector are the same as the number of estimated frequencies.

[0099] The triangular cascade structure adaptive notch filter and the Kalman filter stop working; after the triangular cascade structure adaptive notch filter performs rough estimation on the harmonic frequency, the Kalman filter performs frequency refinement to ensure that the error cannot exceed the allowable range of the initial value error of the autoregressive model. Therefore, when the triangular cascade structure adaptive notch filter and the Kalman filter converge, the frequency estimation value should meet the convergence requirements of the autoregressive model, and the triangular cascade structure adaptive notch filter and the Kalman filter stop working after the functions are completed, which can reduce the calculation amount, and the stopping process can be realized by the unit step function in Figure 3 .

[0100] The autoregressive model is started, and the estimated frequency is finely adjusted and a reference signal is generated: the frequency value output by the Kalman filter at the stopping time is taken as the initial value , and the fine estimation value of the frequency is obtained by iteration The reference signals x1(k), x2(k), …, x n (k) are generated by the parallel filter-based least mean square algorithm.

[0101] The reference signal generated in S3-3 is:

[0102]

[0103] wherein, x i (·) is the reference signal generated by the autoregressive model, is the frequency-related parameter in the i-th autoregressive model, N anf is the time when the triangular cascade structure adaptive notch filter and the Kalman filter stop working, is the initial value of the autoregressive model, is the step size of the i-th autoregressive model, w i0(·) is the first coefficient of the i-th adaptive filter in the parallel filter-based least mean square algorithm, (·) is the first coefficient of the finite impulse response filter model of the transfer function from force to load plane acceleration,

[0104] The parallel filter-based least mean square algorithm starts to calculate the control force after receiving the reference signal.

[0105] The second control force F in S4 LMS (k) is:

[0106]

[0107] Where n is the number of the parallel filter-based least mean square algorithm, W i (·) is the i-th adaptive filter, (·) is the finite impulse response filter model of the transfer function from force to load plane acceleration, W i (·) is the coefficient vector of the i-th adaptive filter, μ i (·) is the step size of the i-th module in the parallel filter-based least mean square algorithm, (·) is the i-th filtered reference signal.

[0108] The total control force F in S5 c (k) is:

[0109] F c (k) = F PID (k) + F LMS (k).

[0110] In the above steps S1-S5, the acceleration measurement in step S1 and the PID controller calculation in step S2 are executed throughout the control process, the frequency coarse identification and refinement in steps S3-1 and S3-2 are executed at the beginning of the algorithm operation, and stop working after convergence, and steps S3-3 to S5 are executed after steps S3-1 and S3-2 stop, in addition, the finite impulse response model of the transfer function S(z) from force to load plane acceleration needs to be identified offline.

[0111] In an embodiment of the present application, the control performance of the control method designed by simulation analysis on the control performance of three harmonic disturbances and random vibration is verified, and the advancement and superiority of the present application in dealing with the wideband micro-vibration control problem containing unknown harmonics are verified.

[0112] The system parameters in this simulation case are as follows: the base frequency of the vibration isolation system is 5 Hz, the sampling frequency is 5 kHz, the load is installed on the upper plane of the vibration isolation system, and the lower plane is connected to the spacecraft; the frequencies of the harmonic disturbances are set to 0.8 Hz, 30 Hz and 60 Hz, respectively, among which 0.8 Hz is used to simulate the disturbance acceleration generated by flexible accessories with a peak value of 0.0005 mg, 30 Hz is used to simulate the disturbance acceleration generated by reaction flywheels with a peak value of 0.1 mg, and 60 Hz represents the disturbance force generated by a refrigerator with a peak value of 0.1 N, which directly acts on the load plane; the random vibration transmitted from the spacecraft to the bottom of the vibration isolation system is simulated by a Gaussian white noise with a mean value of 0 and a variance of 5×10 -6 mg; p The parameters of the PID controller are k i = 1, k d = 20, and k d = N; i The modules contained in the parallel filter-based least mean square algorithm have three steps, which are 0.003, 0.005 and 0.08, respectively, among which the order of the adaptive filter W 11 is 50; the stopping time of the triangular cascade structure adaptive notch filter and the Kalman filter is 25 s; the step size of H 33 and H -4 in the triangular cascade structure adaptive notch filter is 1×10 12 , and the filter factor is 0.999; the filter factor of H 23 and H 23 is 0.99, and the step size of H -3 is 1×10 13 ; the filter factor of H 22 and H 13 is 0.99, and the step size of H -3 is 1×10 -4 ; the covariance matrix Q in the Kalman filter is diag(0.001 0.001 0.001), and R is diag(0.01 0.01 0.01); the step sizes of the three autoregressive models are 0.0003, 0.001 and 0.008, respectively.

[0113] Figure 6 and Figure 7 demonstrate the control ability of the proposed control algorithm for such wideband micro-vibration containing harmonic disturbances, and the time and frequency domain results show that the proposed control algorithm can not only achieve disturbance suppression in a wide frequency band, but also efficiently suppress harmonic disturbances; Figure 8 demonstrates the final frequency estimation results of the designed reference signal generator, and the estimation accuracy of the three frequencies is better than 0.001 Hz, 5×10 -4 Hz and 1×10 -4Hz; from these simulation results, it can be seen that the control method for wideband micro-vibration containing unknown harmonics proposed in the application can accurately estimate the harmonic frequency under the condition that the harmonic disturbance characteristics are unknown, efficiently suppress the harmonic disturbance, and has a wideband disturbance suppression capability compared with the narrowband filter-based least mean square algorithm.

[0114] The application comprehensively designs a PID controller and a parallel filter-based least mean square algorithm, realizes wideband disturbance suppression through the PID controller, and then further efficiently suppresses multiple harmonics in the disturbance by using the parallel filter-based least mean square algorithm; the application designs a reference signal generator, which can estimate the harmonic frequency according to the system residual disturbance under the condition that the harmonic disturbance characteristics are unknown, and generate relevant reference signals for the parallel filter-based least mean square algorithm to ensure the convergence of the controller; the triangular cascade structure adaptive notch filter in the reference signal generator realizes rapid coarse estimation of the frequency, and the Kalman filter refines the initial value of the frequency estimation, reduces the fluctuation of the frequency estimation value, can provide an accurate initial value for the autoregressive model under the condition of low signal-to-noise ratio, and ensure the convergence of the autoregressive model.

[0115] Those skilled in the art will realize that the embodiments described herein are for the purpose of helping the reader understand the principles of the application, and should be understood as not limiting the scope of protection of the application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations according to the technical inspiration disclosed in the application without departing from the essence of the application, and these modifications and combinations are still within the scope of protection of the application.

Claims

1. A method for active control of broadband micro-vibration containing unknown harmonics, characterized in that, The method comprises the following steps: S1: obtaining residual acceleration response at a spacecraft load by using an acceleration sensor; S2: obtaining a first control force by using a PID controller based on the residual acceleration response; The first control force in S2 is: wherein, , and are a proportional coefficient, an integral coefficient and a derivative coefficient, respectively, is a filter coefficient, is a Laplace operator, is a residual acceleration response; S3: performing coarse estimation, refined estimation and fine adjustment on harmonic frequencies by using a reference signal generator based on the residual acceleration response, and generating a reference signal; The reference signal generator in S3 comprises a triangular cascade structure adaptive notch filter, a Kalman filter and an autoregressive model, and S3 comprises the following steps: S3-1: performing real-time coarse estimation on harmonic frequencies by using the triangular cascade structure adaptive notch filter based on the residual acceleration response; S3-2: performing refined estimation on the coarse estimated harmonic frequencies by using the Kalman filter, so that the frequency error is within the allowable range of the initial value error of the autoregressive model; S3-3: taking the frequency value output at the stopping moment of the Kalman filter as the initial value of the autoregressive model, and performing iterative and fine adjustment on the estimated frequency by using the autoregressive model based on the initial value and the residual acceleration response, to generate the reference signal; S4: obtaining a second control force by using a parallel filter-based least mean square algorithm based on the reference signal; S5: obtaining a total control force according to the first control force and the second control force, and outputting the total control force by an actuator, so as to realize efficient suppression of wideband micro-vibration containing multiple unknown harmonics, and complete active control on the wideband micro-vibration containing unknown harmonics.

2. The method of claim 1, wherein the method is characterized by, The residual acceleration response at the spacecraft payload in S1 The calculation formula is: wherein, is the acceleration response for control force generation, is the acceleration response for disturbance generation, is the measurement noise.

3. The method of claim 2, wherein the method is characterized by, The S3-1 triangular cascade structure adaptive wave trap The calculation formula is: wherein and are delay operators, is the cosine of the notch angle frequency of the notch filter, is the filter factor, is the notch angle frequency of the notch filter, is the target frequency, is the sampling frequency, The gradient descent method is adopted to obtain the coarse estimated harmonic frequencies and parameter updates of the triangular cascade structure adaptive notch filter, with the square of the output value of the triangular cascade structure adaptive notch filter as the optimization target, and the formula is as follows: in, For a rough estimate of the harmonic frequency of the notch filter, For a moment, Let be the cosine function of the notch filter's notch angular frequency. For the step size of the notch filter that needs to be updated, For notch filter output, For notch filter output relative to The derivative of This is the input to a single adaptive notch filter.

4. The method of claim 3, wherein the method is characterized by, The calculation formula of the Kalman filter in S3-2 is as follows: wherein is the a priori estimate frequency vector of the Kalman filter, is the computed value of the Kalman filter at time is the computed value of the Kalman filter at time is the predicted error covariance matrix, is the error covariance matrix, and are the measured and process noise covariance matrices, respectively, is the Kalman gain matrix, is the computed value of the Kalman filter at time is the computed value of the Kalman filter at time 5. The method of claim 4, wherein the method is characterized by, The reference signal generated in S3-3 is as follows: wherein, is a reference signal generated for an autoregressive model, is a frequency-dependent parameter in the autoregressive model, is the time at which the Kalman filter and the triangular cascaded structure adaptive notch filter are stopped, is an initial value for the autoregressive model, is a step size for the autoregressive model, is a first coefficient of the autoregressive filter in the parallel filter-based least mean square algorithm, is a first coefficient of the finite impulse response filter model.

6. The method for active control of broadband micro-vibration containing unknown harmonics according to claim 5, wherein, The second control force in S4 is: in, This represents the number of parallel filter-based minimum mean square algorithms. For the first An adaptive filter, The finite impact response filter model is the transfer function of the force-to-load plane acceleration. For the first The coefficient vector of an adaptive filter, For the parallel filter-based least mean square root algorithm, the first... The step size of each module, For the first A filtered reference signal.

7. The method of claim 6, wherein the method is characterized by, The total control force in S5 is: 。

Images