CN116186464A - Nonlinear input/output system parameter identification method based on high-order least square method - Google Patents
Nonlinear input/output system parameter identification method based on high-order least square method Download PDFInfo
- Publication number
- CN116186464A CN116186464A CN202310464689.2A CN202310464689A CN116186464A CN 116186464 A CN116186464 A CN 116186464A CN 202310464689 A CN202310464689 A CN 202310464689A CN 116186464 A CN116186464 A CN 116186464A
- Authority
- CN
- China
- Prior art keywords
- identified
- order
- nonlinear
- square method
- output system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 238000005259 measurement Methods 0.000 claims description 33
- 239000011159 matrix material Substances 0.000 claims description 10
- 230000036961 partial effect Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 229940088594 vitamin Drugs 0.000 claims description 3
- 229930003231 vitamin Natural products 0.000 claims description 3
- 235000013343 vitamin Nutrition 0.000 claims description 3
- 239000011782 vitamin Substances 0.000 claims description 3
- 150000003722 vitamin derivatives Chemical class 0.000 claims description 3
- 238000004088 simulation Methods 0.000 description 30
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
- Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)
Abstract
The invention discloses a nonlinear input/output system parameter identification method based on a high-order least square method, which belongs to the technical field of system identification and comprises the following specific steps of: determining parameters of a nonlinear input/output system and a system to be identified; step S2: converting the system parameters to be identified into minimum values of residual functions to be identified; step S3: performing high-order Taylor expansion on the residual function to be identified; step S4: according to the type of the nonlinear input and output system, adopting a least square method to solve the update value of the parameter to be identified of the system; step S5: and solving the parameters to be identified of the system by using an iteration formula. By adopting the nonlinear input/output system parameter identification method based on the high-order least square method, the accuracy of the parameters to be identified of the system model is improved by adopting high-order expansion of the objective function, so that the introduced error is reduced.
Description
Technical Field
The invention relates to the technical field of system identification, in particular to a nonlinear input/output system parameter identification method based on a high-order least square method.
Background
The least square method is widely applied to various fields such as error estimation of a radar system, curve fitting, parameter identification of a time-varying system, orbit prediction and the like, the identification is to discuss the parameter estimation problem of a model under the condition that the state is measurable, the parameter identification of the traditional least square method is biased estimation, and the linear model parameter can be well identified. When using the traditional least square method for parameter identification, each observation point in the obtained sample data is treated equally. In this case, the dependent variable changes with its own sample value, and thus the condition of homogeneity of the residual variance is not satisfied. In the prior art, when the least square method is adopted to identify the model parameters, a first-order linear solution of the model is generally adopted. However, both linear and nonlinear models introduce errors in the process of solving, resulting in divergence of the system model parameters to be solved. For most of the existing system models, a large amount of errors are introduced when the system parameters are identified through linearization models, namely, through Taylor expansion. When the nonlinearity of the system model is stronger, the model parameters to be solved are more seriously diverged, so that the model parameters to be solved are inaccurate.
Disclosure of Invention
The invention aims to provide a nonlinear input/output system parameter identification method based on a high-order least square method, which reduces introduced errors and improves the accuracy of parameters to be identified of a model.
In order to achieve the above purpose, the present invention provides a nonlinear input/output system parameter identification method based on a high-order least square method, which comprises the following specific steps:
step S1: determining parameters of a nonlinear input/output system and a system to be identified;
step S2: converting the system parameters to be identified into minimum values of residual functions to be identified;
step S3: performing high-order Taylor expansion on the residual function to be identified;
step S4: according to the type of the nonlinear input and output system, adopting a least square method to solve the update value of the parameter to be identified of the system;
step S5: and solving the parameters to be identified of the system by using an iteration formula.
Preferably, in step S1, the nonlinear input-output system is a strong nonlinear system of multidimensional input and multidimensional output, and the expression is as follows:
wherein ,is a known input and output value, < >>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Dimension of input value and output value, respectively, +.>Is a time series; />Is about->Is a continuously differentiable multidimensional nonlinear function, < >> and />Respectively a linear parameter and a non-linear parameter, +.> and />For the system parameters to be identified, < > for>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Parameters-> and />Is a dimension of (2); />Is Gaussian white noise, obeys the expectation that 0 variance is +.>Has statistical properties->;
And (3) making:
wherein ,is about->Is a continuously differentiable multidimensional nonlinear function; />Representing a transpose of the matrix;
combining equations (1) and (2) yields:
preferably, in step S2,
the residual function to be identified is:
preferably, in step S3, the residual function to be identified is subjected to a higher-order taylor expansion,
assuming that the first is obtainedSolving value +.>Then is provided with->Second to->The next iteration formula is as follows:
will beAt->The high-order Taylor expansion is carried out at the position, and the expansion process is as follows:
wherein ,,/>in the form of a jacobian matrix,is the partial derivative; />Is the order of expansion; vector function->Is about->Vitamin variable->Jacobian matrix of (a); />Is->Kronnecker product of (a); />For the variables->Is used for the estimation of the estimated value of (a).
Preferably, in step S4,
the types of the nonlinear input-output system comprise an over-measurement system and an under-measurement system, and the over-measurement system and the under-measurement system respectively correspond to an over-measurement equation solution and an under-measurement equation solution.
Preferably, in step S4, the solving of the residual function to be identified is converted into the solving of the identification parameter to be updatedI.e.:
the equivalent of equation (8) is rewritten as:
and (3) recording:
when (when)When the equation (9) is an overdetering equation, the overdetering equation solution can be obtained by adopting a least square method to solve the overdetering equation:
when (when)When the equation (9) is an under-measurement equation, the solution of the under-measurement equation is obtained by adopting a least square method to solve the solution of the under-measurement equation:
therefore, the nonlinear input/output system parameter identification method based on the high-order least square method is adopted, and when the objective function is linearized, the introduction of errors is reduced by utilizing the high-order terms expanded by Taylor, so that the precision of the parameter to be estimated is improved.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
FIG. 1 is a flow chart of a nonlinear input/output system parameter identification method based on a high-order least square method;
Detailed Description
Examples
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
Referring to fig. 1, a nonlinear input/output system parameter identification method based on a high-order least square method specifically comprises the following steps:
step S1: and determining parameters of the nonlinear input/output system and the system to be identified.
The nonlinear input-output system is a strong nonlinear system with multidimensional input and multidimensional output, and the expression is as follows:
wherein ,is a known input and output value, < >>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Dimension of input value and output value, respectively, +.>Is a time series; />Is about->Is a continuously differentiable multidimensional nonlinear function, < >> and />Respectively a linear parameter and a non-linear parameter, +.> and />For the system parameters to be identified, < > for>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Parameters->Andis a dimension of (2); />Is Gaussian white noise, obeys the expectation that 0 variance is +.>Has statistical properties->;
And (3) making:
wherein ,is about->Is a continuously differentiable multidimensional nonlinear function; />Representing a transpose of the matrix;
combining equations (1) and (2) yields:
step S2: and converting the system parameters to be identified into minimum values of residual functions to be identified.
the residual function to be identified is:
step S3: and performing high-order Taylor expansion on the residual function to be identified.
Assuming that the first is obtainedSolving value +.>Then is provided with->Second to->The next iteration formula is as follows:
will beAt->The high-order Taylor expansion is carried out at the position, and the expansion process is as follows:
wherein ,,/>is Jacobian matrix->Is the partial derivative; />Is the order of expansion. Vector function->Is about->Vitamin variable->Jacobian matrix of (a); />Is->Kronnecker product of (a); />For the variables->Is used for the estimation of the estimated value of (a).
Step S4: and adopting a least square method to solve the parameter updating value to be identified of the system according to the type of the nonlinear input and output system.
The types of the nonlinear input-output system comprise an over-measurement system and an under-measurement system, and the over-measurement system and the under-measurement system respectively correspond to an over-measurement equation solution and an under-measurement equation solution.
A simple linear model is expressed as follows:
wherein ,called state value->In which case the system is an overdetering system. When->Called state value->In which case the system is an under-measurement system.Is a measurement matrix; />The normal distribution is met; />To model errors conforming to Gaussian distribution, obeying the expectation that 0 variance is +.>And has statistical properties->。
Least squares method for over-measurement system state estimation.
When the system model is an overdetermined model, the linear model can be equivalently deformed into:
mathematical expectations for the above equation are:
the estimation error covariance matrix is as follows:
least squares method for under-measured system state estimation.
In a time-dependent linear modelWhen the system is called an under-measurement system. Introducing intermediate variablesThe method comprises the following steps:
substituting the above into a linear model to obtain:
the equivalent deformation is carried out to obtain the following components:
for the mathematical expectation calculation, there are
For a pair ofAfter mathematical expectation operation is carried out on two sides, the formula is substituted into the formula to obtain:
the corresponding estimation error covariance matrix is:
in step S4, the residual function to be identified is solved and converted into the identification parameter to be updatedI.e.:
the equivalent of equation (8) is rewritten as:
and (3) recording:
when (when)When the equation (9) is an overdetering equation, the overdetering equation solution can be obtained by adopting a least square method to solve the overdetering equation:
when (when)When the equation (9) is an under-measurement equation, the solution of the under-measurement equation is obtained by adopting a least square method to solve the solution of the under-measurement equation:
step S5: and solving the parameters to be identified of the system by using an iteration formula.
To verify the effectiveness of the present invention, simulation experiments were performed and analyzed.
Simulation 1:
for a strong nonlinear multiple input multiple output system of the type:
wherein ,the model accords with normal distribution, and is an under-measurement equation. />Is white noise with a mean value of zero and a variance of 0.2; />The linear parameter and the nonlinear parameter, respectively. Wherein->,. Initial data of the experiment are->,/>. Simulation results of the 20 Monte Carlo simulation tests are shown in FIGS. 2-9. The following table shows a comparison of the performance of the different algorithms.
Table 1 the performance of the different algorithms is compared as follows:
table 1 shows the average square error of the simulation results. From table 1, it can be seen that the second order least squares method (2 LS) improves the recognition accuracy of the linear parameters by 16.85% and 21.69% and the recognition accuracy of the nonlinear parameters by 0.38% and 15.78% relative to the least squares method (LS). Compared with a third-order least square method (3 LS) and a fourth-order least square method (4 LS), the identification precision of the least square method (LS) to linear parameters is improved by 19.92% -24.61% and 24.61% -30.31%, and the identification precision to nonlinear parameters is improved by 17.76% -48.01% and 28.51% -56.19%. Therefore, the less errors are introduced, and the identification accuracy of the parameters can be improved more obviously.
Simulation 2:
aiming at a stronger nonlinear model, the simulation test is formed by conforming a plurality of exponential functions:
wherein ,the model accords with normal distribution, and is an under-measurement equation. />Is white noise with a mean value of zero and a variance of 0.2; />The linear parameter and the nonlinear parameter, respectively. Wherein the method comprises the steps of,/>. Initial data of the experiment are->,. The curves in FIGS. 10-23 all pass through 20 Monte CarloThe experimental results of the simulation are shown in Table 2, which is a comparison of the performance of different algorithms.
Table 2 the performance of the different algorithms is compared as follows:
table 2 shows the average square error of the simulation results. As can be seen from the data in table 2, the algorithm mentioned in this document has significantly improved accuracy in identifying parameters compared to the original method. Table 2 can be derived: compared with the least square method (LS), the second-order least square method (2 LS) improves the identification accuracy of the linear parameters by 55.87% on average, and improves the identification accuracy of the nonlinear parameters by 63.98% on average. Compared with the least square method (LS), the identification accuracy of the three-order least square method (3 LS) and the four-order least square method (4 LS) is averagely improved by 63.41% -69.70%, and the identification accuracy of the nonlinear parameters is averagely improved by 77.44% -89.91%.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention and not for limiting it, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that: the technical scheme of the invention can be modified or replaced by the same, and the modified technical scheme cannot deviate from the spirit and scope of the technical scheme of the invention.
Claims (6)
1. A nonlinear input/output system parameter identification method based on a high-order least square method is characterized by comprising the following specific steps:
step S1: determining parameters of a nonlinear input/output system and a system to be identified;
step S2: converting the system parameters to be identified into minimum values of residual functions to be identified;
step S3: performing high-order Taylor expansion on the residual function to be identified;
step S4: according to the type of the nonlinear input and output system, adopting a least square method to solve the update value of the parameter to be identified of the system;
step S5: and solving the parameters to be identified of the system by using an iteration formula.
2. The nonlinear input/output system parameter identification method based on the higher-order least square method as claimed in claim 1, wherein the method is characterized by comprising the following steps: in step S1, the nonlinear input-output system is a strong nonlinear system with multidimensional input and multidimensional output, and the expression is as follows:
wherein ,is a known input and output value, < >>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Dimension of input value and output value, respectively, +.>Is a time series; />Is about->Is a continuously differentiable multidimensional nonlinear function, < >> and />Respectively a linear parameter and a non-linear parameter, +.> and />For the system parameters to be identified, < > for>Represents->Is->Set of dimension real numbers->Represents->Is->Set of dimension real numbers-> and />Parameters-> and />Is a dimension of (2); />Is Gaussian white noise, obeys the expectation that 0 variance is +.>Has statistical properties->;
And (3) making:
wherein ,is about->Is a continuously differentiable multidimensional nonlinear function; />Representing a transpose of the matrix;
combining equations (1) and (2) yields:
3. the nonlinear input/output system parameter identification method based on the higher-order least square method as claimed in claim 2, wherein the method is characterized by comprising the following steps: in the step S2 of the process of the present invention,
the residual function to be identified is:
4. the nonlinear input/output system parameter identification method based on the higher-order least square method as claimed in claim 3, wherein the method is characterized by comprising the following steps: in step S3, the residual function to be identified is subjected to a higher-order taylor expansion,
assuming that the first is obtainedSolving value +.>Then is provided with->Second to->The next iteration formula is as follows:
will beAt->The high-order Taylor expansion is carried out at the position, and the expansion process is as follows:
5. The nonlinear input/output system parameter identification method based on the higher-order least square method as claimed in claim 4, wherein the method is characterized by comprising the following steps: in the step S4 of the process of the present invention,
the types of the nonlinear input-output system comprise an over-measurement system and an under-measurement system, and the over-measurement system and the under-measurement system respectively correspond to an over-measurement equation solution and an under-measurement equation solution.
6. The nonlinear input/output system parameter identification method based on the higher-order least square method as claimed in claim 5, wherein the method is characterized by comprising the following steps: in step S4, the residual function to be identified is solved and converted into the identification parameter to be updatedI.e.:
the equivalent of equation (8) is rewritten as:
and (3) recording:
when (when)When the equation (9) is an overdetering equation, the overdetering equation solution can be obtained by adopting a least square method to solve the overdetering equation:
when (when)When the equation (9) is an under-measurement equation, the solution of the under-measurement equation is obtained by adopting a least square method to solve the solution of the under-measurement equation:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310464689.2A CN116186464B (en) | 2023-04-27 | 2023-04-27 | Nonlinear input/output system parameter identification method based on high-order least square method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310464689.2A CN116186464B (en) | 2023-04-27 | 2023-04-27 | Nonlinear input/output system parameter identification method based on high-order least square method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116186464A true CN116186464A (en) | 2023-05-30 |
CN116186464B CN116186464B (en) | 2023-07-07 |
Family
ID=86449349
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310464689.2A Active CN116186464B (en) | 2023-04-27 | 2023-04-27 | Nonlinear input/output system parameter identification method based on high-order least square method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116186464B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020111758A1 (en) * | 2000-10-18 | 2002-08-15 | Qing-Guo Wang | Robust process identification and auto-tuning control |
CN105549079A (en) * | 2016-01-12 | 2016-05-04 | 中国矿业大学(北京) | Method and device for establishing full-waveform inversion model for geophysics parameters |
WO2022033183A1 (en) * | 2020-08-13 | 2022-02-17 | 重庆邮电大学 | Dynamic-static data hybrid-driven reduced-form grey box space identification method for hammerstein nonlinear industrial system |
WO2022105104A1 (en) * | 2020-11-18 | 2022-05-27 | 南通大学 | Multi-innovation recursive bayesian algorithm-based battery model parameter identification method |
CN114740496A (en) * | 2022-03-18 | 2022-07-12 | 中国人民解放军国防科技大学 | Three-dimensional wind field inversion method based on high-order Taylor expansion |
CN115453871A (en) * | 2022-09-06 | 2022-12-09 | 佛山科学技术学院 | Non-linear system modeling method based on IDE extended multidimensional Taylor network |
-
2023
- 2023-04-27 CN CN202310464689.2A patent/CN116186464B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020111758A1 (en) * | 2000-10-18 | 2002-08-15 | Qing-Guo Wang | Robust process identification and auto-tuning control |
CN105549079A (en) * | 2016-01-12 | 2016-05-04 | 中国矿业大学(北京) | Method and device for establishing full-waveform inversion model for geophysics parameters |
WO2022033183A1 (en) * | 2020-08-13 | 2022-02-17 | 重庆邮电大学 | Dynamic-static data hybrid-driven reduced-form grey box space identification method for hammerstein nonlinear industrial system |
WO2022105104A1 (en) * | 2020-11-18 | 2022-05-27 | 南通大学 | Multi-innovation recursive bayesian algorithm-based battery model parameter identification method |
CN114740496A (en) * | 2022-03-18 | 2022-07-12 | 中国人民解放军国防科技大学 | Three-dimensional wind field inversion method based on high-order Taylor expansion |
CN115453871A (en) * | 2022-09-06 | 2022-12-09 | 佛山科学技术学院 | Non-linear system modeling method based on IDE extended multidimensional Taylor network |
Non-Patent Citations (1)
Title |
---|
刘昊东 等: "基于递推最小二乘法的变体飞行器模型参数在线辨识", 空天防御, vol. 3, no. 3, pages 103 - 110 * |
Also Published As
Publication number | Publication date |
---|---|
CN116186464B (en) | 2023-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Peng et al. | A new unbiased stochastic derivative estimator for discontinuous sample performances with structural parameters | |
Dehestani et al. | Combination of Lucas wavelets with Legendre–Gauss quadrature for fractional Fredholm–Volterra integro-differential equations | |
CN110442911B (en) | High-dimensional complex system uncertainty analysis method based on statistical machine learning | |
Alagoz et al. | Time-domain identification of one noninteger order plus time delay models from step response measurements | |
CN116504341B (en) | Sequential singular value filtering method for data-driven identification partial differential equation | |
CN112764345B (en) | Strong nonlinear system Kalman filter design method based on target state tracking | |
CN113032988B (en) | Design method of high-order extended Kalman filter based on maximum correlation entropy | |
CN116186464B (en) | Nonlinear input/output system parameter identification method based on high-order least square method | |
Figueroa et al. | An approach for identification of uncertain Wiener systems | |
CN115079573A (en) | High-order expansion strong tracking filter of nonlinear system | |
CN111931301A (en) | Time-varying reliability method applied to mechanical structure | |
JP2003216658A (en) | Design support method and program | |
Sezer et al. | Solving high‐order linear differential equations by a Legendre matrix method based on hybrid Legendre and Taylor polynomials | |
CN111737883B (en) | Nonlinear double-rate circuit system robust identification method with output time lag | |
CN107992453A (en) | A kind of bayes method for the optimization of non-negative L1 norm constraints | |
Conte et al. | Robust planar tracking via a virtual measurement approach | |
Alchikh et al. | On the solutions of the fractional Bratu’s problem | |
Poskitt et al. | Bias correction of semiparametric long memory parameter estimators via the prefiltered sieve bootstrap | |
CN105608237B (en) | Rapid waveform prediction method for post-simulation stage of circuit layout | |
CN111274752B (en) | Power amplifier behavior modeling method based on two-stage open loop and binary function | |
Tronarp et al. | Updates in Bayesian filtering by continuous projections on a manifold of densities | |
CN114124026A (en) | Nonlinear system for filter based on maximum entropy high-order expansion | |
CN117236060A (en) | Novel recursion estimation method of continuous stirring reaction kettle based on auxiliary model | |
Mandal | Convergence analysis and numerical implementation of projection methods for solving classical and fractional Volterra integro-differential equations | |
Biagiola et al. | Robust control of wiener systems: a case study |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |