Antenna servo system model identification method

Abstract

The invention discloses an antenna servo system model identification method, and belongs to the technical field of antenna servo system control. In the model identification process of the antenna servo system, pseudo-random signals similar to white noise and a low-pass filter are introduced, the input and output response of the antenna servo system at each frequency point is obtained by using the pseudo-random signals with uniformly distributed frequency spectrums, and the low-pass filter is connected in series before the identification model. The method is used for reducing system vibration in a model identification process. According to the method, on the premise of accurately obtaining the frequency domain model characteristics of the antenna servo system, system buffeting in the model identification process can be effectively reduced, and the operation safety of the antenna servo system is ensured.

Description

Technical Field

[0001] The present invention relates to the field of antenna servo systems, and particularly to a method for model identification of an antenna servo system. Background Art

[0002] With the development of communication technologies, people have put forward higher requirements for the servo performance of antennas. In order to improve the dynamic performance of antenna servo systems, various high-performance controller design methods have been applied to the actual engineering of antenna servo systems. An accurate system model is a prerequisite for designing high-performance controllers. With the popularization and application of various advanced controllers, researchers have also proposed various servo system modeling methods.

[0003] For antenna servo systems, the modeling methods mainly include mechanism modeling and system identification. Due to the complex structure of antennas and the large number of internal non-linear factors, it is difficult to conduct mechanism modeling, and only a small number of components such as the driver model, servo motor model, and gear reduction mechanism model can be modeled. At the same time, the parameters of most models are difficult to measure and are usually not applicable to the design of actual antenna servo control systems.

[0004] The system identification method is based on certain prior knowledge. According to the operating characteristics of the identification object, a specific input signal is selected to excite the target system. By analyzing the potential relationship between the input and output data of the system, an accurate mathematical model of the target system is finally established. When conducting model identification, the input excitation signal must fully excite the target system on the premise of ensuring the safe operation of the system. Common excitation signals include swept-frequency signals, white noise signals, and pseudo-random signals, etc.

[0005] For electromechanical systems such as antenna servo systems, using sinusoidal input signals with multiple different frequencies for swept-frequency testing is a commonly used modeling method. The advantage of swept-frequency testing is that it can accurately describe the frequency-domain response characteristics of the servo system at each frequency test point, and the established model is relatively accurate in the middle and low frequency bands. However, the disadvantages of swept-frequency testing are also obvious: the energy of the input signal is too high, and the antenna servo system will vibrate violently at the frequency points of the resonance peaks. At the same time, the energy of the input signal decays very rapidly in the high frequency band, which results in a low effective bandwidth of the system model.

[0006] Pseudo-random signals are another type of identification signal widely used in system modeling. Their spectra are uniformly distributed in the frequency domain. By using the pseudo-random response of the system, the unit impulse response of the system can be obtained, and then accurate modeling can be completed, which can be used for model identification of various systems. When pseudo-random signals are used for antenna servo system modeling, the antenna will still vibrate to a certain extent during the model identification process, and there will also be a certain degree of fluctuation in the frequency-domain response curve of the system, which will lead to a decrease in the model identification accuracy. Summary of the Invention

[0007] In view of the operating characteristics of the antenna servo system, the present invention provides a method for identifying the model of the antenna servo system, which can reduce the vibration of the antenna during the identification process while ensuring the accuracy of the identification model.

[0008] The technical solution adopted by the present invention is as follows:

[0009] A method for identifying the model of an antenna servo system, characterized by comprising the following steps:

[0010] Step 1, according to the structural characteristics and servo parameters of the target antenna object, select the clock period, signal amplitude, signal bit number, repetition period of the pseudo-random signal, and the cut-off frequency of the low-pass filter, and generate the model identification input signal and low-pass filter parameters of the system;

[0011] Step 2, send the model identification signal and low-pass filter parameters to the antenna servo controller. The servo controller applies the identification excitation signal in the antenna speed loop according to the signal parameters, then records the input signal and output signal of the antenna servo system, and uploads them to the upper computer;

[0012] Step 3, process the input signal and output signal, calculate their autocorrelation function and cross-correlation function, obtain the impulse response sequence of the antenna servo system, then construct the Hankel matrix of the system according to the impulse response sequence, and decompose the matrix to obtain the coefficient matrix of the state space equation of the system and the corresponding model;

[0013] Step 4, for the state space equation of the system, calculate the singular values of each order of the system in this state space, select the order of the system model according to the singular value size and antenna type, reduce the order of the system model, and reverse eliminate the low-pass filter part to obtain the state equation of the final identification model.

[0014] Further, in step 1, a low-pass filter is connected in series after the pseudo-random signal.

[0015] Further, in step 4, the order of the system model is reduced based on the singular values of each order of the system and the modal truncation method.

[0016] The beneficial effects of the present invention are:

[0017] 1. Using pseudo-random signals and low-pass filters for model identification can greatly reduce the vibration of the antenna during the model identification process while ensuring the accuracy of modeling;

[0018] 2. Using the Hankel matrix decomposition method to construct the state space equation of the system can effectively ensure the accuracy of the state space equation of the system;

[0019] 3. Model order reduction is carried out based on the system singular value and modal truncation algorithm, which can reduce the model order while ensuring the model accuracy, facilitating subsequent controller design. Description of the Drawings

[0020] Figure 1 It is a parameter diagram of a pseudo-random signal.

[0021] Figure 2 It is a frequency-domain diagram of a low-pass filter.

[0022] Figure 3 It is a diagram showing the composition of the model identification system of an antenna.

[0023] Figure 4 It is a diagram of the input and output data of the model identification of a certain antenna.

[0024] Figure 5 It is a diagram showing the effect of the model identification of a certain antenna. Detailed Implementation Manner

[0025] The present invention will be further described below in conjunction with the drawings and specific embodiments.

[0026] An antenna servo system model identification method includes the following steps:

[0027] First step: According to the structural characteristics and servo parameters of the target antenna object, select parameters such as the clock period, signal amplitude, signal bit number, and repetition period of the pseudo-random signal, as well as filtering coefficients such as the cut-off frequency of the low-pass filter, to generate a system model identification signal and low-pass filter parameters. The pseudo-random signal is connected in series with a low-pass filter, and the model identification input signal is the pseudo-random signal.

[0028] Second step: Send the model identification signal and low-pass filter parameters to the antenna servo controller. The servo controller applies an identification excitation signal to the antenna speed loop according to the signal parameters, then records identification data such as the input signal and output signal of the antenna servo system, and uploads them to the upper computer. A high-performance PLC can be used as the antenna servo controller.

[0029] Third step: Process the input signal and output signal of the identification data, calculate their autocorrelation function and cross-correlation function, obtain the impulse response sequence of the antenna servo system, then construct a Hankel matrix of the system according to this sequence, and decompose the matrix to obtain the coefficient matrix of the state space equation of the system and the corresponding model.

[0030] Fourth step: For the state space equation of the system, calculate the singular values of each order of the system in this state space, select the system model order according to the singular value size and antenna type, reduce the order of the system model based on the modal truncation algorithm, and reverse-eliminate the low-pass filter part to obtain the state equation of the final identification model.

[0031] In the process of model identification of the antenna servo system, a pseudo-random signal similar to white noise and a low-pass filter are introduced. The pseudo-random signal with a uniform spectrum distribution is used to obtain the input-output responses of the antenna servo system at each frequency point. The low-pass filter is connected in series before the identification model to reduce the system vibration during the model identification process. By the above method, this method can effectively reduce the system chattering during the model identification process on the premise of accurately obtaining the frequency-domain model characteristics of the antenna servo system, and ensure the safe operation of the antenna servo system.

[0032] The following is a more specific example:

[0033] A method for identifying the model of an antenna servo system, the specific steps are as follows:

[0034] The first step: Design the signal parameters of the system model identification. When designing the signal parameters, it mainly includes pseudo-random signal parameters and low-pass filter parameters. Among them, the pseudo-random signal is as Figure 1 shown, and the main parameters include clock period, signal amplitude, signal bit number, and repetition period; the low-pass filter parameters are mainly the filtering coefficient. Under different filtering coefficients, the frequency-domain characteristics of the low-pass filter are as Figure 2 shown.

[0035] Among the pseudo-random signal parameters, the clock period Δt is the minimum level change interval. The smaller the clock period, the wider the power spectrum of the pseudo-random signal covers, the autocorrelation function is closer to the τ function, and the higher the recognition rate of the high-frequency part of the system; however, if the clock period is too small, the power spectrum will fall in the high-frequency band too much, reducing the output signal energy, resulting in a decrease in the signal-to-noise ratio, which will affect the final recognition accuracy. According to engineering test experience, the clock frequency is usually selected as 10 to 20 times the estimated cut-off frequency of the system.

[0036] The signal bit number N determines the length of the pseudo-random signal, and the length calculation formula is M = 2 N -1. The signal bit number N determines the power spectrum line intensity and spacing of the signal. According to engineering test experience, if the system adjustment time is T s , then the value of the signal bit number N can be referred to: 2 N -1 = (1.25 to 1.5)(T s / Δt).

[0037] The signal amplitude a represents the amplitude of the pseudo-random positive and negative signals, which determines the intensity of the power spectrum lines of the identification signal. According to engineering test experience, it is usually 5% to 10% of the maximum value of the input signal of the servo system.

[0038] The repetition period q represents the number of times the pseudo-random signal repeats and continuously inputs the test object system. It has no effect on the power spectrum of the identification signal. However, multiple periods help to cancel out the sudden interference caused by measurement noise, which is helpful for reducing the mean square error of the pulse response calculation error. According to engineering test experience, the repetition period q is usually taken as an integer between 1 and 3.

[0039] The low-pass filter is mainly used to suppress the high-frequency components in the pseudo-random signal. Its main parameter is the filtering coefficient. When the filtering coefficient is smaller, the cut-off frequency of the low-pass filter is lower, and the filtering effect is stronger; when the filtering coefficient is larger, the cut-off frequency of the low-pass filter is higher, and the filtering effect is weaker. According to engineering test experience, when the value range of the filtering coefficient is between 0.8 and 0.9, it can not only reduce the system vibration but also fully stimulate the system response.

[0040] Step 2: The antenna servo controller applies the identification excitation signal. Most antennas are multi-axis control systems, which can be divided into types such as AE type mounts, AEC type mounts, and XY type mounts. In the antenna servo control system, the multi-axis control system is usually divided into single-axis subsystems, and each subsystem is responsible for controlling one rotation axis. This method mainly conducts tests on each single-axis subsystem.

[0041] When performing antenna servo system model identification work, speed loop identification is mainly carried out. The speed loop of the antenna includes a servo driver, a drive motor, a gear reducer, an antenna load, and a code disk, etc. The composition of the antenna identification system is as Figure 3 shown, and the identification excitation signal is directly used as the reference signal of the speed loop.

[0042] The identification input is the rotational angular velocity of the antenna, and the identification output is the rotational angular velocity of the antenna fed back by the code disk. The rotational angular velocity of the antenna is sent from the upper computer to the PLC controller, and then the PLC controls the servo driver and the drive motor to complete the rotation of the antenna. The rotational angular velocity of the antenna is fed back to the PLC controller by a high-precision optical encoder through the EtherCAT bus. After the test is completed, the PLC controller packs and sends the identification input and output data to the upper computer for storage.

[0043] According to the above model identification operation, the input and output data effect of a certain antenna servo system is as Figure 4 shown.

[0044] Step 3: Process the identification data and calculate the system state space equation. Assume that the state space coefficient matrix of the model to be identified is {A, B, C, D}, then the state equation of this model can be expressed as G(z) = C(zI - A) -1 B + D, where I is the identity matrix and z is the independent variable of the equation.

[0045] First, perform correlation processing on the identified input and output data to obtain the impulse response sequence of the system. Denote the input and output signal sequences as x(k) and y(k) respectively, and the impulse response as g(k). Define the autocorrelation function R xx (k) and the cross-correlation function R yx (k) as follows:

[0046]

[0047] where K is a positive integer and is the sequence number of the data in the impulse response sequence.

[0048] The autocorrelation function of the input pseudo-random signal sequence can be expressed as:

[0049]

[0050] where a is the amplitude of the impulse response sequence.

[0051] Using the above functions, the impulse response sequence of the system can be calculated as According to this impulse response sequence, an m-dimensional Hankel matrix H1 and H2 can be constructed and are respectively expressed as follows:

[0052]

[0053] Substitute the system impulse response sequence into the Hankel matrices H1 and H2 respectively, and perform matrix decomposition to calculate the coefficient matrices {A, B, C, D} of the system state space equation G(z).

[0054] Fourth step: System model order reduction to obtain the final identified model. For the system model obtained in the third step, calculate the singular values of each mode in this state space to obtain the influence index of each mode on the input and output.

[0055] According to the singular values of each mode, divide the system state x into the retained state x r and the truncated state x t into two parts, that is, x = [x r x t T .

[0056] For the antenna servo system, the coefficient matrix D is usually 0, then the coefficient matrices {A, B, C, D} can be expressed as follows:

[0057]

[0058] Therefore, the reduced-order system model can be described as follows:

[0059] ∑(A r , B r , C r ​, 0)

[0060] According to the above model identification and reduction method, the model identification effect of a certain antenna servo system is as Figure 5 shown.

[0061] In summary, the present invention fully considers the working characteristics of the antenna servo system during operation, uses a pseudo-random signal as the input signal, and connects a low-pass filter in series to reduce the system vibration during the model identification process, and finally obtains the frequency-domain response characteristics of the servo system, improving the accuracy and safety of the antenna servo system model identification.

Claims

1. A method for identifying an antenna servo system model, characterized in that: The following steps are involved: Step 1, according to the structural characteristics and servo parameters of the target antenna object, the clock period, signal amplitude, signal bit number, repetition period, and cutoff frequency of the pseudo-random signal are selected to generate the model identification input signal and low-pass filter parameters of the system; Step 2: Send the model identification signal and low-pass filter parameters to the antenna servo controller, which applies the identification excitation signal to the antenna velocity loop according to the signal parameters, and then records the input and output signals of the antenna servo system and uploads them to the host computer; Step 3, processing the input signal and the output signal, calculating their autocorrelation function and cross-correlation function, obtaining the impulse response sequence of the antenna servo system, then constructing the Hankel matrix of the system according to the impulse response sequence, and decomposing the matrix to obtain the state space equation coefficient matrix of the system and the corresponding model; Step 4: For the state space equation of the system, calculate the singular values ​​of each order of the system in the state space, and select the system model order according to the size of the singular value and the antenna type, reduce the order of the system model, reversely eliminate the low-pass filter part, and obtain the state equation of the final identification model.

2. The antenna servo system model identification method according to claim 1, characterized in that: In step 1, a low-pass filter is connected in series after the pseudo-random signal.

3. The antenna servo system model identification method according to claim 1, characterized in that: In step 4, the system model is reduced based on the singular values ​​of each order of the system and the modal truncation method.

Images