Dynamic thrust reconstruction method independent of parameter model and based on dynamic matrix control

Through the dynamic matrix control method that does not rely on the parameter model, the problem of dynamic parameter calibration error and cumbersome operation in dynamic thrust measurement is solved, and more accurate and efficient dynamic thrust reconstruction is achieved, which is suitable for thrust replacement and low sampling frequency conditions. measurement.

CN119004671APending Publication Date: 2024-11-22SHANGHAI JIAOTONG UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411098801.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing dynamic thrust measurement methods rely heavily on the dynamic parameter model, resulting in large parameter calibration errors and recalibration of the dynamic equations required to be recalibrated. The operation is cumbersome and the reconstruction results are distorted under the influence of signal filtering and other factors.

Method used

Using a closed-loop measurement method based on dynamic matrix control that does not rely on parameter models, by obtaining the unit step response sequence and output displacement sequence of the measurement system, the output of the dynamic matrix control algorithm is used to control the system to track the actual displacement, and solve the control amount as the dynamic thrust. Measured value.

Benefits of technology

The problem of inaccurate calibration of dynamic parameters is avoided, the operation process after the thrust is replaced is simplified, the testing efficiency is improved, and more accurate measurement results are obtained at low sampling frequency, reducing the reconstruction mean square error.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004671A_ABST
    Figure CN119004671A_ABST
Patent Text Reader

Abstract

The invention relates to a dynamic thrust reconstruction method based on dynamic matrix control independent of a parameter model, which is characterized in that a tracking control problem is constructed based on a closed-loop measurement principle, and the output of a measurement system model is controlled by using a dynamic matrix control algorithm to track actually measured displacement. The method takes a controlled quantity solved in a control process as a measured value of a reconstructed dynamic thrust, and comprises the following steps: acquiring a unit step response sequence a and an output displacement sequence of an actual thrust measurement system; acquiring control parameters determined offline, wherein the control parameters comprise a predicted length M of dynamic matrix control and a cost function J (k) of the kth period in a control process; in the control process, based on a set control target and the control parameters, the control quantity of each period is solved, and the control quantity is the thrust obtained through reconstruction. Compared with the prior art, the method has the advantages of high accuracy, effective improvement of efficiency and the like.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of aerospace technology, and relates to a micro-thrust measurement method of a micro-thruster, and in particular to a dynamic thrust reconstruction method based on dynamic matrix control that is independent of a parameter model. Background Art

[0002] Space gravitational wave detection is one of the key research directions in the current field of space science. Space gravitational wave detection requires maintaining an ultra-quiet and ultra-stable satellite platform through drag-free control. In drag-free control, micro-thrusters with excellent performance are used as actuators to offset the influence of non-conservative forces. In order to accurately calibrate and evaluate the performance of micro-thrusters, it is crucial to develop a high-performance micro-thrust measurement system.

[0003] A widely used measurement method in micro-thrust measurement is to fix the thruster on a thrust frame that can rotate around an elastic pivot or elastic torsion wire, and measure the thrust according to the angular displacement or linear displacement generated by the thrust frame when the thruster is working. Measuring the steady-state thrust or pulse impulse generated by the thruster by measuring the steady-state displacement or maximum displacement of the thrust frame can meet the thrust measurement needs in most cases, but this measurement method cannot obtain the dynamic response process of the thruster, and it is also impossible to measure the performance parameters such as the response time of the thruster. If you want to further understand the physical process of the thruster's operation, conduct a more comprehensive calibration and evaluation of the thruster, and develop a higher-performance thruster, it is necessary to measure the dynamic change process of the thrust.

[0004] At present, several existing dynamic thrust measurement methods can be mainly divided into two types according to the measurement method: open-loop reconstruction and closed-loop compensation. Among them, the open-loop reconstruction method usually calculates the dynamic force based on the dynamic model such as the transfer function of the measurement system and the measured displacement; while the closed-loop compensation method introduces a standard force of known magnitude to balance the force to be measured during the measurement process, so that the pendulum always remains at zero position, thereby reflecting the magnitude of the force to be measured through the magnitude of the standard force.

[0005] In the open-loop reconstruction method, the method of decomposing the dynamic input force into three items, elastic force (stiffness multiplied by displacement), damping force (damping coefficient multiplied by velocity) and inertial force (moment of inertia multiplied by acceleration), based on the dynamic equation and then synthesizing and calculating them is simple and easy to use, with clear physical meaning. It is an algorithm commonly used in open-loop measurement. This method can generally be called the inversion of dynamic equation (IODE) algorithm. For example, the document "Dynamic-force extraction for micro-propulsion testing: Theory and experimental validation" (.REVIEW OF SCIENTIFIC INSTRUMENTS89,115110 (2018)) discloses a micro-propulsion power extraction method based on the least squares method. After modeling the micro-Newtonian torsion pendulum displacement vibration, if the station constant is properly calibrated, the time evolution of the dynamic force can be calculated. According to the linear characteristics of the motion equation, the reconstruction of the dynamic thrust is simplified to solving the linear equation. However, although the inversion of the dynamic equation is sufficient to meet the requirements of dynamic thrust reconstruction most of the time, the algorithm still has some shortcomings:

[0006] 1. This algorithm is heavily dependent on the dynamic equation model. Before use, the measurement system's rotational inertia, damping coefficient, stiffness coefficient and other parameters must be accurately calibrated. In actual measurement, due to signal filtering, signal transmission delay and other reasons, the dynamic equation model may not match the actual system model, and the parameter calibration error may be large, resulting in distorted reconstruction results.

[0007] 2. A measurement system is usually used to measure thrusters of various masses. In a measurement system based on a thrust frame, the thruster is directly installed on the thrust frame. Replacing the thruster will cause changes in parameters such as the system's moment of inertia. If the dynamic equation inversion is used, the dynamic equation needs to be recalibrated each time the thruster is replaced, which is a cumbersome operation. Summary of the invention

[0008] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a dynamic thrust reconstruction method based on dynamic matrix control which has high accuracy and effectively improves efficiency and is independent of parameter models.

[0009] The purpose of the present invention can be achieved by the following technical solutions:

[0010] A dynamic thrust reconstruction method based on dynamic matrix control that does not rely on parameter models is proposed. The method constructs a tracking control problem based on the principle of closed-loop measurement, and uses a dynamic matrix control algorithm to control the output of the measurement system model to track the actual measured displacement. The control variable solved in the control process is used as the measurement value of the reconstructed dynamic thrust.

[0011] Further, the method comprises the following steps:

[0012] Obtain the unit step response sequence a and output displacement sequence of the actual thrust measurement system;

[0013] Obtaining control parameters determined offline, including the prediction length M of the dynamic matrix control and the cost function J(k) of the kth cycle in the control process;

[0014] During the control process, based on the set control target and the control parameters, the control amount of each cycle is solved, and the control amount is the reconstructed thrust.

[0015] Furthermore, the control objective is set to minimize the cost function.

[0016] Furthermore, the unit step response sequence is represented as a={a1, a2, ..., a N}, the length N of the unit step response sequence satisfies: when i>N, a i ≈a N .

[0017] Furthermore, the prediction length M of the dynamic matrix control satisfies M≤N, where N is the length of the unit step response sequence.

[0018] Furthermore, the cost function J(k) of the kth cycle in the control process is:

[0019]

[0020] in, is the prediction of the system model output for the next M cycles when the control quantity for the next M cycles is implemented in the kth control cycle; X M (k) is the reference trajectory of the k+1th to k+Mth periods, i.e., a sequence consisting of the k+1th to k+Mth elements in the output displacement sequence; U M (k) is the control increment for the next M cycles; Q is the error weight matrix, and R is the control weight matrix.

[0021] Furthermore, the Calculated by the following formula:

[0022]

[0023] in, It is the initial prediction of the system model output for the next M periods when no control is implemented in the future; A is the dynamic matrix containing the dynamic information of the step response.

[0024] Furthermore, the control quantity for solving each cycle is specifically:

[0025] In the 0th to LMth cycles, the control amount of each cycle is calculated by iterative solution, where L is the length of the output displacement sequence;

[0026] In the L-M+1th to L-1th cycles, the control increment U obtained directly based on the LMth cycle M (LM) calculates the control amount of each cycle, and the calculation formula is:

[0027] u(L-M+j)=u(L-M+j-1)+Δu(L-M+j)

[0028] Among them, u(L-M+M) and u(L-M+j-1) are the control quantities at the time of L-M+j and L-M+j-1 respectively, and Δu(L-M+j) is the control increment U M The j+1th element in (LM), j=1,…,M-1.

[0029] Furthermore, the calculation of the control amount of each cycle by iterative solution method specifically includes:

[0030] a) Initialize variables: cycle count k = 0, initial prediction vector Initial control amount u(-1)=0;

[0031] b) Construct a k-cycle reference trajectory X M (k)=[x(k+1),x(k+2),…,x(k+M)] T , the initial prediction sequence Square brackets represent vectors The 1st to the Mth elements;

[0032] c) Calculate the control increment vector U of k cycles according to the set control target M (k), take out the first element Δu(k) and calculate the control amount u(k)=u(k-1)+Δu(k) for the k-cycle;

[0033] d) Update the initial prediction value at time k+1 Where S is the shift matrix;

[0034] e) k=k+1, jump to step b) until k>LM.

[0035] Furthermore, the shift matrix S is:

[0036]

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] First, when the method of the present invention is used to perform dynamic thrust reconstruction, only a step response sequence of the measurement system needs to be obtained, and there is no need to calibrate the dynamic parameters of the measurement system, thereby avoiding the problem of inaccurate calibration of the dynamic parameters.

[0039] Second, each time the thruster is replaced, the present invention does not need to carry out a cumbersome calibration process, and only needs to measure a step response sequence of the measurement system, which simplifies the operation process and improves the thruster test efficiency.

[0040] Third, when the sampling frequency of the measurement process is low, the method of the present invention has a lower reconstruction mean square error than the traditional method and can obtain more accurate measurement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the principle of the present invention.

[0042] Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0043] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0044] This embodiment provides a dynamic matrix control based dynamic thrust reconstruction method (DMC-DTR) that does not rely on a parameter model and is used for a measurement system with good linearity. When used, only a unit step response sequence of the measurement system needs to be obtained without calibrating the dynamic equations.

[0045] The above method is based on the principle of closed-loop measurement (when the displacement of the thrust frame caused by the standard force and the force to be measured is equal, the standard force is equal to the force to be measured), and transforms the dynamic thrust reconstruction problem into a tracking control problem. The DMC algorithm is used to control the output of the measurement system model to track the measured displacement. The control quantity solved in the control process is the measured value of the dynamic thrust. Figure 2 As shown, the specific process is as follows:

[0046] Step S1: Obtain the unit step response sequence a={a1, a2, ..., aN}, and the output displacement sequence {x(1), x(2), ..., x(L)}, where the length N of the unit step response sequence satisfies when i>N, a i ≈a N , L is the length of the output displacement sequence.

[0047] Step S2, obtaining control parameters determined offline, including the prediction length M of the dynamic matrix control and the cost function J(k) of the kth cycle in the control process.

[0048] The prediction length M of the dynamic matrix control satisfies M≤N. The larger M is, the smaller the reconstruction error is and the longer the calculation time of each cycle is. M reaches 2 times the sampling frequency f s After the length of , the reconstruction error remains basically unchanged.

[0049] The cost function J(k) of the kth cycle in the control process is:

[0050]

[0051] in, is the prediction of the system model output for the next M cycles when the control quantity for the next M cycles is implemented in the kth control cycle, which can be calculated by the following formula:

[0052]

[0053] in, is the initial prediction of the system model output for the next M cycles when no control is implemented in the future; A is the dynamic matrix containing the dynamic information of the step response, expressed as:

[0054]

[0055] X M (k) is the reference trajectory of the k+1th to k+Mth periods, that is, a sequence consisting of the k+1th to k+Mth elements in the output displacement sequence, which can be expressed as:

[0056] X M (k) = [x(k+1)…x(k+M)] T ;

[0057] U M (k) is the control increment for the next M cycles, which can be expressed as:

[0058] U M (k) = [Δu(k)…Δu(k+M-1)] T ;

[0059] Q is the error weight coefficient q1, q2, ..., q Mis the error weight matrix with diagonal elements, R is the control weight coefficient r1, r2, ..., r M is the control weight matrix of the diagonal elements, which can be expressed as:

[0060] Q = diag(q1, q2, ..., q M )

[0061] R = diag(r1, r2, ..., r M ).

[0062] Step S3: During the control process, based on the set control target and the control parameters, the control amount of each cycle is solved, and the control amount is the reconstructed thrust.

[0063] In this embodiment, the control target is set to solve U M (k) minimizes the cost function. Specifically, we can minimize the cost function by letting J(k) M The partial derivative of (k) is equal to 0, and the solution is U that minimizes J(k) M (k), i.e.

[0064]

[0065] Where D = (A T QA+R) -1 A T Q is a vector that can be calculated offline after A, Q, and R are determined, and does not need to be updated online. In each control cycle, a control increment vector U is calculated. M (k), but only the first element is taken to calculate the control amount of the current cycle, which is the reconstructed thrust, unless the number of remaining displacement data is less than M and is insufficient to form a reference trajectory.

[0066] Therefore, the control quantity for each cycle is:

[0067] In the 0th to LMth cycles, the control quantity of each cycle is calculated by iterative solution, where;

[0068] In the L-M+1th to L-1th cycles, the control increment U obtained directly based on the LMth cycle M (LM) Calculate the control amount for each cycle.

[0069] Solving the control quantity from the 0th to the LMth cycle by iterative solution specifically includes:

[0070] a) Initialize variables: cycle count k = 0, initial prediction vector Initial control amount u(-1)=0;

[0071] b) Construct a k-cycle reference trajectory XM (k)=[x(k+1),x(k+2),...,x(k+M)] T , the initial prediction sequence Square brackets represent vectors The 1st to the Mth elements, and similar square brackets in other places also indicate this meaning;

[0072] c) Calculate the control increment vector U of k cycles according to the set control target M (k), take out the first element Δu(k) and calculate the control amount u(k)=u(k-1)+Δu(k) for the k-cycle;

[0073] d) Update the initial prediction value at time k+1 Where S is the shift matrix, expressed as:

[0074]

[0075] e) k=k+1, jump to step b), and repeat the above process until k>LM.

[0076] The last cycle of the above iterative solution process will solve the control increment vector U of the LMth cycle. M (LM), the remaining displacement data is not enough to support the above iterative solution operation, so the subsequent calculation is directly based on U M The control increment in (LM) calculates the control amount of the remaining cycles. The specific steps are as follows:

[0077] a) Initialize j = 1;

[0078] b) The control increment Δu(L-M+j) at the time L-M+j is the vector U M The j+1th element of (LM), and the control amount u(L-M+j)=u(L-M+j-1)+Δu(L-M+j);

[0079] c) j=j+1, jump to step b), and repeat the process until j>M-1.

[0080] After the above steps are completed, all control variables {u(0), u(1), ..., u(L-1)} have been calculated, which is the reconstruction result of the measured dynamic thrust.

[0081] In addition to reconstructing the thrust offline after all displacement data are collected, the above method can also reconstruct the thrust in real time while measuring. When applied to real-time reconstruction, this method should be started simultaneously with the measurement process to ensure that the initial conditions are correct, and M displacement data need to be cached. After measuring the displacement data x(k+M) of k+M cycles, the thrust u(k) of k cycles is reconstructed using step S3. At this time, only the step response of the measurement system in step S1 and step S2 need to be executed offline, and step S3 needs to be executed online in real time.

[0082] In this embodiment, a micro-thrust measurement system is used to conduct an actual dynamic thrust measurement experiment to verify the effectiveness and performance of the above method. Specifically, the thrust of a sine wave, sawtooth wave and / or square wave is generated by the Ampere force between the coil and the permanent magnet to simulate the dynamic thrust to be measured. Figure 1 As shown in the figure, the output displacement of the measurement system is first collected during the experiment, and then Figure 2 The process shown reconstructs the thrust.

[0083] like Figure 1 As shown, when the output displacement acquisition experiment is performed, the thruster is fixed on the torsion pendulum thrust frame. When the thruster is set to work with the thrust and disturbance action, the output displacement generated by the torsion pendulum thrust frame is collected through the sensor. Then, based on the output displacement, the model output of the torsion pendulum model is controlled by a closed-loop tracking control method to achieve thrust reconstruction.

[0084] When the dynamic equation parameters are calibrated accurately, the reconstruction mean square error of the algorithm of the present invention is compared with that of the traditional dynamic equation inversion algorithm. Then, by turning on the smoothing filter of the displacement sensor and simulating the situation of dynamic equation model mismatch, the reconstruction mean square error of the algorithm of the present invention is compared with that of the traditional dynamic equation inversion algorithm.

[0085] Among them, the reconstruction mean square error is defined as:

[0086]

[0087] Where T M is the measurement time, F(t) is the true value of thrust, For the reconstruction results, the unit of RMSE is μN.

[0088] Table 1 shows the experimental results. In the case of model adaptation, the reconstruction mean square error of the algorithm of the present invention is basically consistent with that of the traditional algorithm, while in the case of model mismatch, the reconstruction mean square error of the algorithm of the present invention is significantly smaller than that of the traditional algorithm.

[0089] Table 1 Results of the implementation cases of the present invention

[0090]

[0091]

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. The present invention can be used in most linear measurement systems and dynamic thrusts of different waveforms.

[0093] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A dynamic thrust reconstruction method based on dynamic matrix control that does not rely on parameter models, characterized in that: This method constructs a tracking control problem based on the principle of closed-loop measurement, and uses a dynamic matrix control algorithm to control the output of the measurement system model to track the actual measured displacement. The control variable solved during the control process is used as the measurement value of the reconstructed dynamic thrust.

2. The dynamic thrust reconstruction method based on dynamic matrix control that does not rely on a parameter model according to claim 1 is characterized in that: The method comprises the following steps: Obtain the unit step response sequence a and output displacement sequence of the actual thrust measurement system; Obtaining control parameters determined offline, including the prediction length M of the dynamic matrix control and the cost function J(k) of the kth cycle in the control process; During the control process, based on the set control target and the control parameters, the control amount of each cycle is solved, and the control amount is the reconstructed thrust.

3. The dynamic thrust reconstruction method based on dynamic matrix control that does not rely on parameter models according to claim 2 is characterized in that: The control objective is set to minimize the cost function.

4. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 2 is characterized in that: The unit step response sequence is represented by a={a1, a2, ..., a N }, the length N of the unit step response sequence satisfies: when i>N, a i ≈a N .

5. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 2 is characterized in that: The prediction length M of the dynamic matrix control satisfies M≤N, where N is the length of the unit step response sequence.

6. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 2 is characterized in that: The cost function J(k) of the kth cycle in the control process is: in, is the prediction of the system model output for the next M cycles when the control quantity for the next M cycles is implemented in the kth control cycle; X M (k) is the reference trajectory of the k+1th to k+Mth periods, i.e., a sequence consisting of the k+1th to k+Mth elements in the output displacement sequence; U M (k) is the control increment for the next M cycles; Q is the error weight matrix, and R is the control weight matrix.

7. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 6 is characterized in that: Said Calculated by the following formula: in, It is the initial prediction of the system model output for the next M periods when no control is implemented in the future; A is the dynamic matrix containing the dynamic information of the step response.

8. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 6 is characterized in that: The control quantity for solving each cycle is specifically: In the 0th to LMth cycles, the control amount of each cycle is calculated by iterative solution, where L is the length of the output displacement sequence; In the L-M+1th to L-1th cycles, the control increment U obtained directly based on the LMth cycle M (LM) calculates the control amount of each cycle, and the calculation formula is: u(L-M+j)=u(L-M+j-1)+Δu(L-M+j) Among them, u(L-M+M) and u(L-M+j-1) are the control quantities at the time of L-M+j and L-M+j-1 respectively, and Δu(L-M+j) is the control increment U M The j+1th element in (LM), j=1,…,M-1.

9. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 8 is characterized in that: The control amount of each cycle calculated by iterative solution specifically includes: a) Initialize variables: cycle count k = 0, initial prediction vector Initial control amount u(-1)=0; b) Construct a k-cycle reference trajectory X M (k)=[x(k+1),x(k+2),…,x(k+M)] T , the initial prediction sequence Square brackets represent vectors The 1st to the Mth elements; c) Calculate the control increment vector U of k cycles according to the set control target M (k), take out the first element Δu(k) and calculate the control amount u(k)=u(k-1)+Δu(k) for the k-cycle; d) Update the initial prediction value at time k+1 Where S is the shift matrix; e) k=k+1, jump to step b) until k>LM.

10. The dynamic thrust reconstruction method based on dynamic matrix control and independent of parameter model according to claim 9, characterized in that: The shift matrix S is:

Citation Information

Cited By

  • Micro thrust dynamic identification method and system

    CN120296367A

  • A method and system for dynamic identification of small thrust

    CN120296367B