Preset time function observer construction method and separately excited motor state estimation method
By designing a preset time function observer, the problem of the state of the excitation motor cannot be quickly estimated within the preset time in the prior art, and the rapid state estimation within the preset time is realized, which meets the needs of real-time control.
Patent Information
- Application Number
- CN202411019343.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-29
AI Technical Summary
The asymptotic convergence observer designed under the framework of Lyapunov stability theory cannot quickly estimate the rotation speed and current state variables of the excitation motor within the preset time, and cannot meet the needs of real-time control.
A preset time function observer is designed to convert the target vector to be estimated into a new energy-observing linear time-invariant system by constructing the state equation, the output equation and the vector equation to be estimated, and to convert the target vector to be estimated into a new energy-observing linear time-invariant system, and in single output and multiple output situations The time-varying gain is designed to ensure that the observer provides consistent bounded feedback terms within the preset time and achieves fast state estimation.
It realizes the rapid and accurate estimation of system state variables within the preset time, avoids the initial condition dependence of traditional aggressive convergence observers and the discontinuity of the Engel method, and meets the requirements of real-time observation.
Smart Images

Figure CN119051508A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for constructing a preset time function observer and a method for estimating a state of a separately excited motor, and belongs to the field of function observer design. Background Art
[0002] State feedback plays an important role in the field of modern control. However, state feedback requires that the state of the system is measurable. However, in practical applications, the state of the internal variables of the system is often impossible to measure or cannot be fully measured. In order to solve this problem, various observer techniques have been proposed to estimate the state, such as unknown input observers, sliding mode observers, and interval observers. In addition, because in general, only linear functions of the system state are needed in state feedback, in the design of observer-based control systems, function observers are more superior than ordinary state observers, because they do not require the signal required for system control to be given by estimating the entire state of the control system.
[0003] So far, various results on the design of function observers have been reported in the literature. However, most of the literature on the design of function observers focuses on asymptotically convergent observers. Such observers are proposed under the framework of Lyapunov stability theory, so the convergence time cannot be freely specified because it depends on the observer gain and the initial conditions of the original system, both of which are usually unknown.
[0004] In actual feedback control, the state of the system's internal variables is usually impossible to fully measure or even if it can be measured, the cost is higher, such as the speed and current of a separately excited motor. Therefore, various observer techniques have been proposed to estimate the state, such as unknown input observers, sliding mode observers, and interval observers. Among them, function observers are particularly important in the design of observer-based control systems because they do not require a complete estimate of the entire system state to give the signal required for system control. So far, many results on the design of function observers have been reported in the literature. However, most of the research on the design of function observers focuses on asymptotically convergent observers. This type of observer is proposed under the framework of Lyapunov stability theory, so the convergence time cannot be freely set. However, in some control scenarios with high real-time requirements, designers need to obtain the value of state variables in a very short time (preferably within a preset time). For example, in the actual control of a separately excited motor, it is necessary to quickly obtain state variables such as speed and current to quickly achieve the control target. However, to date, there is no effective estimation scheme in this field that can estimate real-time states such as rotor speed and current within a preset time. Summary of the invention
[0005] In order to provide a function observer that can quickly estimate the value of a state variable within a preset time to achieve the purpose of real-time observation, the present invention provides a method for constructing a preset time function observer and a method for estimating the state of a separately excited motor. The technical solution is as follows:
[0006] A method for constructing a preset time function observer of the present invention is used for measuring state variables in a feedback control system, comprising:
[0007] Step 1: Construct the state equation, output equation and vector equation to be estimated, the expressions are:
[0008]
[0009] where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈R r Represents the target vector to be estimated; A∈R n×n , B∈R n×m , C∈R p×n , L∈R r×n is a known constant matrix;
[0010] Step 2: Set the target vector z(t)∈R to be estimated r Transformed into the state vector of a new observable linear time-invariant system, the specific process is as follows:
[0011] Assumption 1:
[0012] Assumption 2:
[0013] If assumptions 1 and 2 are true, then the following conditions hold:
[0014] one,
[0015] 2. The matrix pair (F, G) is observable, that is,
[0016] in L + is the generalized inverse matrix of L, satisfying LL + L = L, when L is full rank, there is LL + =I r ; F=LAL + -Θ1CAL + -Θ2CL + , We get a new observable linear time-invariant system:
[0017]
[0018] in, H=LB-θ1CB,
[0019] Step 3: Considering single-output and multi-output cases respectively, designing a preset time function observer for the new observable linear time-invariant system;
[0020] In the single output case, take the state transition Get the new system:
[0021]
[0022] Where W = [G T (GF) T …(GF r-1 ) T ] T , Then construct the time-varying gain for it:
[0023]
[0024] in, yes Estimation of the time-varying gain It is a function of the preset time T. The specific design method is as follows:
[0025] (a) First, design a time-varying non-singular matrix P(t): ij (t)) r×r :
[0026]
[0027]
[0028]
[0029]
[0030]
[0031] (b) Then design the time-varying gain
[0032]
[0033]
[0034]
[0035] in is a series of constants such that the matrix is the Hurwitz matrix;
[0036] Therefore, based on assumptions 1 and 2, equation (4) determined according to equations (5)-(12) is the preset time observer of equation (3), satisfying Furthermore, In addition, the observer feedback term in equation (4) It remains consistently bounded in the interval [t0, t0+T) and converges to 0 when t→t0+T;
[0037] In the case of multiple outputs, the multiple output system is first converted into a single output system, and then the preset time function observer is designed by the method for the single output case in step 3.
[0038] Optionally, the construction process of the preset time function observer for multiple outputs includes:
[0039] Under the premise of maintaining the observability of the system, we transform it into a single output and first consider an intermediate observer:
[0040]
[0041] In which, by choosing the constant matrix So that the matrix It has r different eigenvalues, which can be obtained by subtracting equation (2) from equation (13):
[0042]
[0043] in
[0044] Next, due to F * There are r different eigenvalues, then there exists a non-singular matrix K∈R r×r Make where s i is the matrix F * The i-th eigenvalue of 1×t , so that G * =ρGK does not contain a column with all elements equal to 0, that is, the matrix pair (F * ,G * ) can be observed, thus obtaining a single-output system:
[0045]
[0046] Maintain observability and then design the observer using the method for the single-output case.
[0047] The state estimation method of the separately excited motor of the present invention adopts a preset time function observer constructed by any of the above methods to measure the state variables of the separately excited motor.
[0048] Optionally, the state variables of the separately excited motor include: the motor shaft angle, the motor rotor rotation rate and the armature current.
[0049] Optionally, the constant matrices of the separately excited motor are respectively:
[0050]
[0051] Among them, C e and C m is the potential constant and electromagnetic torque constant of the separately excited motor, f is the viscosity friction coefficient, L a is the inductance, R a is the resistance, and J is the moment of inertia of the motor.
[0052] The construction system of the preset time function observer of the present invention is used to construct a function observer to realize the measurement of state variables in a feedback control system, and includes: an equation construction module, a conversion module and an observer construction module;
[0053] The equation building module is configured to build a state equation, an output equation and a vector equation to be estimated, and the expressions are:
[0054]
[0055] where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈R r Represents the target vector to be estimated; A∈R n×n , B∈R n×m , C∈R p×n , L∈R r×n is a known constant matrix;
[0056] The conversion module is configured to convert the target vector z(t)∈R r Transformed into the state vector of a new observable linear time-invariant system, the specific process is as follows:
[0057] Assumption 1:
[0058] Assumption 2:
[0059] If assumptions 1 and 2 are true, then the following conditions hold:
[0060] one,
[0061] 2. The matrix pair (F, G) is observable, that is,
[0062] in L + is the generalized inverse matrix of L, satisfying LL + L = L, when L is full rank, there is LL + =I r ; F=LAL + -Θ1CAL + -Θ2CL + , We get a new observable linear time-invariant system:
[0063]
[0064] in, H=LB-θ1CB,
[0065] The observer construction module is configured to consider single-output and multi-output situations respectively, and design a preset time function observer for the new observable linear time-invariant system;
[0066] In the single output case, take the state transition Get the new system:
[0067]
[0068] Where W = [G T (GF) T …(GF r-1 ) T ] T , Then construct the time-varying gain for it:
[0069]
[0070] in, yes Estimation of the time-varying gain It is a function of the preset time T. The specific design method is as follows:
[0071] (a) First, design a time-varying non-singular matrix P(t): ij (t)) r×r :
[0072]
[0073]
[0074]
[0075]
[0076]
[0077]
[0078] (b) Then design the time-varying gain
[0079]
[0080]
[0081]
[0082] in is a series of constants such that the matrix is the Hurwitz matrix;
[0083] Therefore, based on assumptions 1 and 2, equation (4) determined according to equations (5)-(12) is the preset time observer of equation (3), satisfying Furthermore, In addition, the observer feedback term in equation (4) It remains consistently bounded in the interval [t0, t0+T) and converges to 0 when t→t0+T;
[0084] In the case of multiple outputs, the multiple output system is first converted into a single output system, and then the preset time function observer is designed by the method for the single output case in step 3.
[0085] Optionally, the construction process of the preset time function observer for multiple outputs includes:
[0086] Under the premise of maintaining the observability of the system, we transform it into a single output and first consider an intermediate observer:
[0087]
[0088] In which, by choosing the constant matrix So that the matrix It has r different eigenvalues, which can be obtained by subtracting equation (2) from equation (13):
[0089]
[0090] in
[0091] Next, due to F * There are r different eigenvalues, then there exists a non-singular matrix K∈R r×r Make where s i is the matrix F * The i-th eigenvalue of 1×t , so that G * =ρGK does not contain a column with all elements equal to 0, that is, the matrix pair (F * ,G * ) can be observed, thus obtaining a single-output system:
[0092]
[0093] Maintain observability and then design the observer using the method for the single-output case.
[0094] The present invention provides an electronic device, comprising a memory and a processor;
[0095] The memory is used to store computer programs;
[0096] The processor is configured to implement any of the methods described above when executing the computer program.
[0097] The present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any one of the above items is implemented.
[0098] The beneficial effects of the present invention are:
[0099] 1. The present invention studies the problem of designing a preset time function observer for a class of linear time-invariant systems with unknown inputs. Through a series of system deductions, the function observer design problem is converted into a standard Luenberger observer design problem, which provides great convenience for designing a preset time observer to estimate the target vector. Experimental results show that the function observer constructed using the method of the present invention can accurately estimate the state variables of the system.
[0100] 2. Compared with the traditional asymptotically convergent function observer, the present invention can quickly estimate the target vector within a preset time according to the designer's requirements and is not affected by the initial conditions, thus achieving the effect of real-time observation.
[0101] 3. Compared with the preset time observer of Engel's method, the present invention avoids discontinuity at the preset time instant.
[0102] 4. The present invention uses the original matrix to provide the necessary and sufficient conditions for the existence of the preset time function observer, which is convenient for designers to check and verify. BRIEF DESCRIPTION OF THE DRAWINGS
[0103] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0104] Figure 1 It is a flow chart of the method for constructing a preset time function observer of the present invention.
[0105] Figure 2 is the true value of z1, z2 and its corresponding estimated value of the preset time function observer and the observer feedback term γ when T=1 in the second embodiment of the present invention. i (t).
[0106] Figure 3 is the true value of z1, z2 and its corresponding estimated value of the preset time function observer and the observer feedback term γ when T = 0.5 in the second embodiment of the present invention. i (t).
[0107] Figure 4 is the true value of z1, z2 and its corresponding estimated value of the preset time function observer and the observer feedback term γ when T = 0.3 in the second embodiment of the present invention. i (t). DETAILED DESCRIPTION
[0108] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.
[0109] Embodiment 1:
[0110] This embodiment provides a method for designing a preset time function observer for a linear time-invariant system with unknown input, and the steps are as follows:
[0111] Step 1: Construct the state equation, output equation and vector equation to be estimated.
[0112] Step 2: Perform a series of system deductions on the established state space expression to transform the target vector into the state variable of a new observable linear time-invariant system.
[0113] Step 3: Consider the single-output and multi-output cases respectively, and design a preset time function observer for the new observable linear time-invariant system.
[0114] In step 1, the expressions of the state equation, output equation and vector equation to be estimated are:
[0115]
[0116] where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈R r Represents the target vector to be estimated; A∈R n×n ,B∈R n×m ,C∈R p×n ,L∈R r×n is a known constant matrix. Without loss of generality, assume rank(L) = r.
[0117] In step 2, the target vector z(t)∈R r After a series of deductions, it is transformed into a new state vector of an observable linear time-invariant system. The specific process is as follows:
[0118] Assumption 1:
[0119] Assumption 2:
[0120] Where LA and CA represent the multiplication of matrix L and matrix A, and matrix C and matrix A respectively; s represents a complex number, Represents a complex number domain.
[0121] If assumptions 1 and 2 are true, then the following conditions hold:
[0122] one,
[0123] 2. The matrix pair (F, G) is observable, that is,
[0124] in L + is the generalized inverse matrix of L, satisfying LL + L = L, when L is full rank, there is LL + =Ir . F=LAL + -Θ1CAL + -Θ2CL + , I represents the identity matrix, I r represents the r-order identity matrix, Θ represents the linear algebraic equation A solution of , Θ1 and Θ2 represent two block matrices after the matrix Θ is decomposed by column, ∑ represents the matrix and matrix The matrix is divided into rows, ∑ + represents the generalized inverse of the matrix ∑, F and G represent the system matrix and output matrix of the deduced system, respectively.
[0125] Then we can get the deduced observable linear time-invariant system:
[0126]
[0127] in H=LB-θ1CB, is the output vector of the system after deduction, express As a function of y, represents the derivative of y, H represents the control matrix of the system after deduction, and Representation matrix I-∑∑ + A two-block matrix after column decomposition.
[0128] In step 3, the single-output and multi-output cases are considered respectively, and a preset time function observer is designed for the new observable linear time-invariant system.
[0129] In the single output case, take the state transition Get the new system:
[0130]
[0131] Where W = [G T (GF) T …(GF r-1 ) T ] T , express The derivative of express The i-th row element of represents the state vector of the new system after state transition, express The i-th row element of a i express The last row and the i-th column element (i=1,2,…,r), W represents the observability matrix, and Respectively represent the system matrix and output matrix of the new system after state conversion. Then construct the time-varying gain for it:
[0132]
[0133]
[0134] in, express The derivative of express The estimate, express The i-th row element of express The i-th row element (i=1,2,…,r), time-varying gain Represents the function of the preset time T. The specific design method is as follows:
[0135] (a) First, design a time-varying non-singular matrix P(t): ij (t)) r×r :
[0136]
[0137]
[0138]
[0139]
[0140]
[0141] (b) Then design the time-varying gain
[0142]
[0143]
[0144]
[0145] in is a series of constants such that the matrix is the Hurwitz matrix. Therefore, based on assumptions 1 and 2, (4) determined by equations (5)-(12) is the preset time observer of (3), satisfying Furthermore, In addition, the observer feedback term in (4) It remains consistently bounded in the interval [t0, t0+T) and converges to 0 when t→t0+T.
[0146] Multi-output case: Convert to single output while maintaining the observability of the system. First consider an intermediate observer:
[0147]
[0148] in, express The derivative of , the matrix Represents a constant matrix.
[0149] By choosing a constant matrix So that the matrix has r different eigenvalues, and then subtracting (2) and (13) to obtain
[0150]
[0151] in express The derivative of and are the state vector and output vector of the new energy-observable system respectively, and the matrix F * is the system matrix of the new energy observation system.
[0152] Next, due to F * There are r different eigenvalues, then there exists a non-singular matrix K∈R r×r Make where s i is the matrix F * The i-th eigenvalue of 1×t , so that G * =ρGK does not contain a column with all elements equal to 0, that is, the matrix pair (F * ,G * ) can be observed, thus obtaining a single-output system:
[0153]
[0154] And maintain observability, and then use the single-output method to design the observer.
[0155] Embodiment 2
[0156] This embodiment takes the rapid estimation of part of the state of a separately excited motor as an example to introduce the construction method of the preset time function observer of the present invention.
[0157] A separately excited motor is a specific type of DC motor, where "externally excited" means that the excitation current of the motor is provided by an external source rather than generated by the motor itself. This type of motor is usually used in scenarios where precise control of speed and torque is required, such as industrial drives, electric vehicles, and aerospace. For the control of separately excited motors, feedback control is usually required. The goal of feedback control is to make the output of the motor (usually speed or torque) meet the predetermined requirements in order to optimize the stability, response speed, and energy efficiency of the system.
[0158] When implementing feedback control, the state variables that need to be measured and fed back usually include (1) shaft angle (position): the actual angular position of the motor rotor. (2) speed: the rotation rate of the motor rotor. (3) current: the input current of the motor, which is used to excite and control the motor. By measuring these state variables, the control system can adjust the input current and other control signals of the motor in real time to maintain the performance of the motor during operation. Therefore, feedback control of separately excited motors requires measuring and processing these key state variables to achieve the desired control objectives, such as precise speed regulation, torque control, or motion trajectory tracking.
[0159] This embodiment converts the function observer design problem of the separately excited motor into a standard Luenberger observer design problem through system deduction. This observer not only facilitates the estimation of the target vector, but also can achieve good state estimation performance within a preset time when the initial conditions are unknown.
[0160] The function observer construction method for separately excited motor state estimation in this embodiment includes the following steps:
[0161] Step 1: According to the voltage law and the mechanical torque balance law, the state equation, output equation and vector equation to be estimated of the electric couple voltage separately excited motor are established:
[0162]
[0163] y(t)=Cx(t)
[0164] z(t)=Lx(t)
[0165] Where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈Rr Represents the target vector to be estimated. A∈R n×n , B∈R n×m , C∈R p×n , L∈R r×n is a known constant matrix. Without loss of generality, assume rank(L) = r.
[0166] In this embodiment:
[0167]
[0168]
[0169] In the formula, θ, i a are the motor shaft angle, speed and armature current respectively, C e and C m is the potential constant and electromagnetic torque constant of the motor, f is the viscosity friction coefficient, L a is the inductance, R a is the resistance, and J is the moment of inertia of the motor.
[0170] After substituting the above parameters, the constant matrices of this embodiment are expressed as follows:
[0171]
[0172] Step 2: Perform a series of system deductions on the established state space expression to transform the target vector into the state variable of a new observable linear time-invariant system.
[0173] It has been verified that the system matrix satisfies two assumptions, as follows:
[0174] Assumption 1:
[0175] Assumption 2:
[0176] By reconstructing the system, we obtain an observable linear time-invariant system:
[0177]
[0178] Where:
[0179] F=LAL + -Θ1CAL + -Θ2CL +
[0180]
[0181]
[0182]
[0183]
[0184] H=LB-θ1CB
[0185]
[0186]
[0187] The values of the matrices of the new system can be obtained as
[0188]
[0189] Step 3: Consider the single-output and multi-output cases respectively, and design a preset time function observer for the new observable linear time-invariant system.
[0190] Choose ρ = [1 0], then G * =ρG=[1 0], Thus we can obtain a single output system:
[0191]
[0192] In the formula Therefore a1=2, a2=2.
[0193] Design an observer for it:
[0194]
[0195] Among them, choose l1 = 5, l2 = 1 so that the matrix
[0196] is a Hurwitz matrix, and according to:
[0197]
[0198]
[0199]
[0200]
[0201]
[0202] get:
[0203]
[0204] according to:
[0205]
[0206]
[0207] 2≤i≤r-1
[0208]
[0209] get:
[0210]
[0211] Assume u(t) = 1, initial value x(t0) = [1 2 1] T , Respectively Figure 2-4 The cases where the preset times are T=1, T=0.5 and T=0.3 are drawn.
[0212] From the simulation results, it can be concluded that the predicted value achieves an accurate estimate within the preset time and is independent of the initial value.
[0213] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.
[0214] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for constructing a preset time function observer, used for constructing a function observer to realize the measurement of state variables in a feedback control system, characterized in that: The method comprises: Step 1: Construct the state equation, output equation and vector equation to be estimated, the expressions are: where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈R r Represents the target vector to be estimated; A∈R n×n , B∈R n×m , C∈R p×n , L∈R r×n is a known constant matrix; Step 2: Set the target vector z(t)∈R to be estimated r Transformed into the state vector of a new observable linear time-invariant system, the specific process is as follows: Assumption 1: Assumption 2: If assumptions 1 and 2 are true, then the following conditions hold: one, 2. The matrix pair (F, G) is observable, that is, in L + is the generalized inverse matrix of L, satisfying LL + L = L, when L is full rank, there is LL + =I r ; F=LAL + -Θ1CAL + -Θ2CL + , We get a new observable linear time-invariant system: in, H=LB-θ1CB, Step 3: Considering single-output and multi-output cases respectively, designing a preset time function observer for the new observable linear time-invariant system; In the single output case, take the state transition Get the new system: Where W = [G T (GF) T …(GF r-1 ) T ] T , Then construct the time-varying gain for it: in, yes Estimation of the time-varying gain It is a function of the preset time T. The specific design method is as follows: (a) First, design a time-varying non-singular matrix P(t): ij (t)) r×r : (b) Then design the time-varying gain in is a series of constants such that the matrix is the Hurwitz matrix; Therefore, based on assumptions 1 and 2, equation (4) determined according to equations (5)-(12) is the preset time observer of equation (3), satisfying Furthermore, In addition, the observer feedback term in equation (4) It remains consistently bounded in the interval [t0, t0+T) and converges to 0 when t→t0+T; In the case of multiple outputs, the multiple output system is first converted into a single output system, and then the preset time function observer is designed by the method for the single output case in step 3.
2. The method for constructing a preset time function observer according to claim 1, characterized in that: The construction process of the preset time function observer in the case of multiple outputs includes: Under the premise of maintaining the observability of the system, we transform it into a single output and first consider an intermediate observer: In which, by choosing the constant matrix So that the matrix It has r different eigenvalues, which can be obtained by subtracting equation (2) from equation (13): in Next, due to F * There are r different eigenvalues, then there exists a non-singular matrix K∈R r×r Make where s i is the matrix F * The i-th eigenvalue of 1×t , so that G * =ρGK does not contain a column with all elements equal to 0, that is, the matrix pair (F * ,G * ) can be observed, thus obtaining a single-output system: Maintain observability and then design the observer using the method for the single-output case.
3. A state estimation method for a separately excited motor, characterized in that: The method uses a preset time function observer constructed according to the method of any one of claims 1 to 2 to measure the state variables of the separately excited motor.
4. The state estimation method of a separately excited motor according to claim 3, characterized in that: The state variables of the separately excited motor include: the motor shaft angle, the motor rotor rotation speed and the armature current.
5. The state estimation method of a separately excited motor according to claim 3, characterized in that: The constant matrices of the separately excited motor are: Among them, C e and C m is the potential constant and electromagnetic torque constant of the separately excited motor, f is the viscosity friction coefficient, L a is the inductance, R a is the resistance, and J is the moment of inertia of the motor.
6. A system for constructing a preset time function observer, used for constructing a function observer to realize the measurement of state variables in a feedback control system, characterized in that: The system comprises: an equation building module, a conversion module and an observer building module; The equation building module is configured to build a state equation, an output equation and a vector equation to be estimated, and the expressions are: where x(t)∈R n and y(t)∈R p denote the state vector and the measurement output vector respectively, u(t)∈R m represents the known control input vector, z(t)∈R r Represents the target vector to be estimated; A∈R n×n , B∈R n×m , C∈R p×n , L∈R r×n is a known constant matrix; The conversion module is configured to convert the target vector z(t)∈R r Transformed into the state vector of a new observable linear time-invariant system, the specific process is as follows: Assumption 1: Assumption 2: If assumptions 1 and 2 are true, then the following conditions hold: one, 2. The matrix pair (F, G) is observable, that is, in L + is the generalized inverse matrix of L, satisfying LL + L = L, when L is full rank, there is LL + =I r ; F=LAL + -Θ1CAL + -Θ2CL + , We get a new observable linear time-invariant system: in, H=LB-θ1CB, The observer construction module is configured to consider single-output and multi-output situations respectively, and design a preset time function observer for the new observable linear time-invariant system; In the single output case, take the state transition Get the new system: Where W = [G T (GF) T …(GF r-1 ) T ] T , Then construct the time-varying gain for it: in, yes Estimation of the time-varying gain It is a function of the preset time T. The specific design method is as follows: (a) First, design a time-varying non-singular matrix P(t): ij (t)) r×r : (b) Then design the time-varying gain in is a series of constants such that the matrix is the Hurwitz matrix; Therefore, based on assumptions 1 and 2, equation (4) determined according to equations (5)-(12) is the preset time observer of equation (3), satisfying Furthermore, In addition, the observer feedback term in equation (4) It remains consistently bounded in the interval [t0, t0+T) and converges to 0 when t→t0+T; In the case of multiple outputs, the multiple output system is first converted into a single output system, and then the preset time function observer is designed by the method for the single output case in step 3.
7. The system for constructing a preset time function observer according to claim 6, characterized in that: The construction process of the preset time function observer in the case of multiple outputs includes: Under the premise of maintaining the observability of the system, we transform it into a single output and first consider an intermediate observer: In which, by choosing the constant matrix So that the matrix It has r different eigenvalues, which can be obtained by subtracting equation (2) from equation (13): in Next, due to F * There are r different eigenvalues, then there exists a non-singular matrix K∈R r×r Make where s i is the matrix F * The i-th eigenvalue of 1×t , so that G * =ρGK does not contain a column with all elements equal to 0, that is, the matrix pair (F * ,G * ) can be observed, thus obtaining a single-output system: Maintain observability and then design the observer using the method for the single-output case.
8. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is configured to implement the method according to any one of claims 1 to 5 when executing the computer program.
9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.
