Control parameter optimization method applied to zero-stiffness microgravity simulation device

Abstract

A control parameter optimizing method applied to a zero-stiffness microgravity simulation device comprises the steps that firstly, a pseudo-random signal is used for generating an M sequence, and the M sequence serves as an excitation signal of the zero-stiffness microgravity simulation device to obtain an output signal of the device, namely position information; introducing the obtained input and output data into an identification algorithm, and introducing an auxiliary variable model into a recursive least square method; and finally, performing control parameter optimization on the obtained zero-stiffness microgravity simulation device identification model by using a genetic algorithm. According to the method, a pseudo-random signal and an auxiliary variable method are combined to carry out model identification on the zero-stiffness microgravity system, on the basis, a genetic algorithm is used for optimizing control parameters, the number setting time is shortened, and the control precision is improved.

Description

Technical field:

[0001] The invention relates to a control parameter optimization method applied to a zero-rigidity microgravity simulation device, and belongs to the field of zero-rigidity microgravity simulation device model identification and control parameter optimization. Background technology:

[0002] With the continuous development of aerospace technology, major aerospace powers have launched explorations of small celestial bodies (asteroids, comets, meteoroids, etc.). Unlike the gravity of the moon and Mars, the gravity of small celestial bodies is closer to microgravity, which can easily lead to landing failures. In order to ensure the completion of the exploration mission, microgravity simulation of the landing environment of small celestial bodies on the ground has become a focus of research at home and abroad.

[0003] The zero-stiffness microgravity simulation device is generally composed of a magnetoelectric coupling system, a motor control system, and a stiffness coupling system. The device has a complex structure and a high degree of nonlinearity, which makes it difficult to identify the model parameters of the zero-stiffness microgravity simulation device, and thus increases the complexity of the control parameter setting. Therefore, for the precise control of complex devices, accurate identification of the model and setting of the control parameters are very important. The control parameter optimization results of the microgravity simulation device based on model identification will not only affect the reliability of the microgravity simulation, but also the accuracy of the microgravity simulation. Summary of the invention:

[0004] Based on the problem that the zero-stiffness microgravity simulation device has a complex structure and a high degree of nonlinearity, which leads to difficulties in setting the control parameters, the purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a control parameter optimization method for a zero-stiffness microgravity simulation device. The method adopts a combination of pseudo-random signals and auxiliary variable methods to identify the model of the zero-stiffness microgravity system, and on this basis, uses a genetic algorithm to optimize the control parameters, thereby shortening the numerical setting time and improving the control accuracy.

[0005] As conceived above, the technical solution of the present invention is as follows:

[0006] A control parameter optimization method for a zero-stiffness microgravity simulation device is characterized by: firstly, using a pseudo-random signal to generate an M sequence, and using the M sequence as an excitation signal of the zero-stiffness microgravity simulation device to obtain the output signal of the device, that is, position information; then, introducing the obtained input and output data into an identification algorithm, and introducing an auxiliary variable model into the recursive least squares method; finally, using a genetic algorithm to optimize the control parameters of the obtained zero-stiffness microgravity simulation device identification model.

[0007] Preferably, the specific process of the method is as follows:

[0008] (1) Use a pseudo-random signal to generate an M sequence. The formula for the M sequence is as follows:

[0009]

[0010] In the formula, x i is the state of the i-th register; C i The value is 1 or 0, indicating that the line is connected (participating in feedback) or disconnected, respectively. i The value of determines the feedback link and sequence structure of the shift register;

[0011] (2) The zero-stiffness low microgravity simulation device is represented by the following model:

[0012] A(z -1 )g(k)=B(z -1 )x(k)+ε(k)

[0013] in:

[0014] A(z -1 )=1+a1z -1 +a2z -2 +a3z -3 +a4z -4

[0015] B(z -1 )=b0+b1z -1 +b2z -2 +b3z -3 +b4z -4

[0016] In the formula, x(k) is the input signal of the M sequence; g(k) output signal is the position information of the zero stiffness device; ε(k) is the noise;

[0017] (3) Rewrite the above model into a least squares model:

[0018] g(k)=h(k)θ+ε(k)

[0019] Where h(k) = [-g(k-1) ... -g(k-4)x(k-1) ... x(k-4)] T is the observable data; θ=[a1 a2a3 a4b0 b1 b2 b3 b4] is the system variable to be identified;

[0020] Let k = 1, 2, ... m, then G m =[g(1)g(2)...g(m)] T , where G m is a vector composed of m continuous system outputs;

[0021]

[0022] In the formula, H mis a matrix consisting of m consecutive observable data; ε(m) = [ε(1)ε(2)…ε(m)] T ε(m) is a vector composed of m continuous noises, and the matrix form is as follows:

[0023] G m =H m θ+ε m

[0024] (4) Introducing auxiliary variable h * (k), and construct the auxiliary variable matrix as

[0025]

[0026] The auxiliary variable H * m Introduced into the recursive least squares method, the estimated value of the model θ:

[0027]

[0028] In the auxiliary variable method, y(k) is uncorrelated with ε(k), but is strongly correlated with g(k) and x(kd), so an auxiliary variable h can be constructed. * (k)

[0029]

[0030] Make P(0) = αI. Where λ is a zero vector or a sufficiently small positive zero vector, α is a sufficiently large positive real number, and the recursive auxiliary variable least squares method is expressed as

[0031]

[0032] (5) Using genetic algorithm on the identified model, the fitness function in the genetic algorithm is:

[0033]

[0034] Where, e(t) is the system error; p is the overshoot; t r is the rise time; t s is the adjustment time; ω1, ω2, ω3, ω4 are weights.

[0035] The present invention adopts a pseudo-random signal as a model identification input signal, and introduces a recursive auxiliary variable least square method, and then uses the above method to obtain a zero-stiffness microgravity simulation device system model, and designs a genetic algorithm based on the model to optimize the control parameters. This method can obtain an unbiased estimate of the model parameters in the presence of noise interference in the identification process, and also ensures the accuracy of parameter setting and improves the parameter setting efficiency. Description of the drawings:

[0036] Figure 1 It is a flow chart of the present invention. Specific implementation method:

[0038] The specific implementation of the present invention will be further described in detail below in conjunction with the drawings and examples.

[0039] Referring to the accompanying drawings, the present invention provides a control parameter optimization method for a zero-stiffness microgravity simulation device, a control parameter optimization method for a zero-stiffness microgravity simulation device, characterized in that: first, a pseudo-random signal is used to generate an M sequence, and the M sequence is used as an excitation signal of the zero-stiffness microgravity simulation device to obtain the output signal of the device, that is, position information; then the obtained input and output data are introduced into the identification algorithm, due to the influence of external noise and measurement errors, the output data will inevitably have noise interference, and an auxiliary variable model is introduced into the recursive least squares method to offset the influence of noise interference; then a genetic algorithm is used to optimize the control parameters of the obtained zero-stiffness microgravity simulation device identification model.

[0040] The specific implementation process and conditions of the method of the present invention are as follows:

[0041] (1) Use a pseudo-random signal to generate an M sequence. The formula for the M sequence is as follows:

[0042]

[0043] In the formula, x i is the state of the i-th register; C i The value is 1 or 0, indicating that the line is connected (participating in feedback) or disconnected, respectively. i The value of determines the feedback link and sequence structure of the shift register.

[0044] (2) The zero-stiffness low-microgravity simulation device can be represented by the following model:

[0045] A(z -1 )g(k)=B(z -1 )x(k)+ε(k)

[0046] in:

[0047] A(z -1 )=1+a1z -1 +a2z -2 +a3z -3 +a4z -4

[0048] B(z -1 )=b0+b1z -1 +b2z -2+b3z -3 +b4z -4

[0049] Where x(k) is the M sequence input signal; g(k) is the output signal for the zero stiffness device position information; and ε(k) is the noise.

[0050] (3) Rewrite the above model into a least squares model:

[0051] g(k)=h(k)θ+ε(k)

[0052] Where h(k) = [-g(k-1) ... -g(k-4)x(k-1) ... x(k-4)] T is the observable data; θ=[a1 a2a3 a4b0 b1 b2 b3 b4] is the system variable to be identified.

[0053] Let k = 1, 2, ... m, then G m =[g(1)g(2)...g(m)] T Among them, G m is a vector consisting of m consecutive system outputs.

[0054]

[0055] In the formula, H m is a matrix consisting of m consecutive observable data; ε(m) = [ε(1)ε(2)…ε(m)] T ε(m) is a vector composed of m continuous noises. The matrix form is as follows:

[0056] G m =H m θ+ε m

[0057] (4) To ensure that the estimated values ​​of the model parameters are unbiased and consistent, an auxiliary variable h is introduced. * (k), and construct the auxiliary variable matrix as

[0058]

[0059] The auxiliary variable H * m Introduced into the recursive least squares method, the estimated value of the model θ:

[0060]

[0061] In the auxiliary variable method, y(k) is uncorrelated with ε(k), but is strongly correlated with g(k) and x(kd), so an auxiliary variable h can be constructed. * (k)

[0062]

[0063] Make P(0) = αI. Where λ is a zero vector or a sufficiently small positive zero vector, and α is a sufficiently large positive real number. The recursive auxiliary variable least squares method can be expressed as

[0064]

[0065] (5) Using genetic algorithm on the identified model, the fitness function in the genetic algorithm is:

[0066]

[0067] Where, e(t) is the system error; p is the overshoot; t r is the rise time; t s is the adjustment time; ω1, ω2, ω3, ω4 are weights.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A control parameter optimization method for a zero-stiffness microgravity simulation device, characterized in that: Firstly, a pseudo-random signal is used to generate an M sequence, and the M sequence is used as the excitation signal of the zero-stiffness microgravity simulation device to obtain the output signal of the device, namely the position information. Then, the obtained input and output data are introduced into the identification algorithm, and an auxiliary variable model is introduced into the recursive least squares method. Finally, a genetic algorithm is used to optimize the control parameters of the obtained identification model of the zero-stiffness microgravity simulation device.

2. The control parameter optimization method for a zero-stiffness microgravity simulation device according to claim 1, characterized in that: The specific process of the method described is as follows: (1) Use a pseudo-random signal to generate an M sequence. The formula for the M sequence is as follows: In the formula, x i is the state of the i-th register; C i The value is 1 or 0, indicating that the line is connected (participating in feedback) or disconnected, respectively. i The value of determines the feedback link and sequence structure of the shift register; (2) The zero-stiffness low microgravity simulation device is represented by the following model: A(z -1 )g(k)=B(z -1 )x(k)+ε(k) in: <h2 style=";text-align:left;direction:ltr">A(z<h2 style=";text-align:left;direction:ltr"> -1 <h2 style=";text-align:left;direction:ltr"> )=1+a1z<h2 style=";text-align:left;direction:ltr"> -1 <h2 style=";text-align:left;direction:ltr"> +a2z<h2 style=";text-align:left;direction:ltr"> -2 <h2 style=";text-align:left;direction:ltr"> +a3z<h2 style=";text-align:left;direction:ltr"> -3 <h2 style=";text-align:left;direction:ltr"> +a4z<h2 style=";text-align:left;direction:ltr"> -4 B(z -1 )=b0+b1z -1 +b2z -2 +b3z -3 +b4z -4 In the formula, x(k) is the input signal of the M sequence; g(k) output signal is the position information of the zero stiffness device; ε(k) is the noise; (3) Rewrite the above model into a least squares model: g(k)=h(k)θ+ε(k) Where h(k) = [-g(k-1) ... -g(k-4)x(k-1) ... x(k-4)] T is the observable data; θ=[a1 a2 a3 a4 b0b1b2 b3 b4] is the system variable to be identified; Let k = 1, 2, ... m, then G m =[g(1)g(2)...g(m)] T , where G m is a vector composed of m continuous system outputs (; In the formula, H m is a matrix consisting of m consecutive observable data; ε(m) = [ε(1)ε(2)…ε(m)] T ε(m) is a vector composed of m continuous noises, and the matrix form is as follows: G m =H m the+e m (4) Introducing auxiliary variable h * (k), and construct the auxiliary variable matrix as The auxiliary variable H * m Introduced into the recursive least squares method, the estimated value of the model θ: In the auxiliary variable method, y(k) is uncorrelated with ε(k), but is strongly correlated with g(k) and x(kd), so an auxiliary variable h can be constructed. * (k) Make P(0) = αI. Where λ is a zero vector or a sufficiently small positive zero vector, α is a sufficiently large positive real number, and the recursive auxiliary variable least squares method is expressed as (5) Using genetic algorithm on the identified model, the fitness function in the genetic algorithm is: Where, e(t) is the system error; p is the overshoot; t r is the rise time; t s is the adjustment time; ω1, ω2, ω3, ω4 are weights.

Images