Parameter identification method of output error system with random time delay
Through the expected maximization algorithm combined with the linear regression equation of the output error system, the problem of random delay and noise identification in the output error model is solved, efficient parameter identification is achieved, and the accuracy and robustness of model parameters are improved.
Patent Information
- Application Number
- CN202410972851.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-11-08
AI Technical Summary
In the prior art, the parameter identification method of the output error model is difficult to effectively deal with random delays and noise, especially the least squares recursive algorithm is susceptible to colored noise and can only be recognized online. The gradient iteration algorithm cannot solve the estimation of random delays. question.
The expected maximization algorithm is used to combine the linear regression equation of the output error system, and the model parameters are iteratively calculated until converge, and random delays and noise are identified by defining noise-free output parameters and model parameters.
It realizes effective identification of random delays and noise, improves the accuracy and robustness of parameter identification, can identify and process offline data online, and improves the accuracy of model parameter estimation.
Smart Images

Figure CN118916602A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of system identification, and in particular to a parameter identification method for a random time-delay output error system. Background Art
[0002] System identification plays an important role in process industry. Conventional model structures of system identification include auto-regressive with extra inputs (ARX), finite impulse response (FIR), output error (OE) and other models. Among them, the OE model ignoring noise terms is widely used in the field of automatic control because its structure is similar to the transfer function.
[0003] However, the numerator and denominator of the OE model structure contain unknown parameters, and the colored noise problem caused by polynomial transformation needs to be solved. Therefore, a parameter identification method is needed. The methods for unknown parameter identification of the OE model in the prior art include the least squares recursive algorithm, the gradient iterative identification algorithm, etc. However, although the least squares recursive algorithm can realize online identification, its identification performance is easily affected by colored noise and has large errors; the gradient iterative algorithm mines system information by processing batch data, but can only process offline data and cannot solve the estimation problem of random delays. Summary of the invention
[0004] In order to solve the deficiencies in the prior art, the purpose of the present application is to provide a method for identifying parameters of a random time delay output error system, which can identify random time delay and noise and has a good identification effect.
[0005] To achieve the above object, the present application provides a method for identifying parameters of a random time-delay output error system, comprising:
[0006] Based on the OE model expression of the output error system, the noise-free output parameters and model parameters are defined to generate the linear regression equation of the output error system;
[0007] Based on the observed data set of the output error system and the linear regression equation, an expectation maximization algorithm is used to determine the conditional expectation of the output error system with respect to the model parameters;
[0008] Based on the conditional expectation, estimated values of the model parameters are iteratively calculated until convergence.
[0009] Furthermore, the specific steps of defining the noise-free output parameters and model parameters based on the OE model expression of the output error system and generating the linear regression equation of the output error system include:
[0010] Based on the unit shift operator in the OE model expression, the noise-free output parameter is defined using the following formula:
[0011]
[0012] Among them, k is the time, u k is the input sequence of the OE model, y k is the output sequence of the OE model, e k is zero-mean random white noise; A(z) and B(z) are polynomials about the unit shift operator, z is the unit backshift operator, is the term coefficient of A(z), n a is the order of A(z), is the term coefficient of B(z), n b is the order of B(z); τ k is the random delay at time k, which obeys uniform distribution;
[0013] Based on the noise-free output parameter, a model parameter and an information vector are defined so that the noise-free output parameter is the product of the model parameter and the information vector, and the formula is as follows:
[0014]
[0015] in, is the information vector, θ is the model parameter;
[0016] in,
[0017]
[0018] Based on the noise-free output parameters, a linear regression equation of the output error system is generated, and the formula is as follows:
[0019]
[0020]
[0021] Furthermore, the specific steps of using the expectation maximization algorithm to determine the conditional expectation of the output error system with respect to the model parameters based on the observed data set of the output error system and the linear regression equation include:
[0022] Collect the observed data set of the output error system and construct a random delay hidden data set of corresponding length:
[0023] The expectation maximization algorithm is used to construct the Q function for calculating the conditional expectation of the current model parameters;
[0024] Based on the linear regression equation, the Q function is decomposed and simplified.
[0025] Furthermore, the specific step of using the expectation maximization algorithm to construct a Q function for calculating the conditional expectation of the current model parameters adopts the following formula:
[0026]
[0027] C obs ={U,Y};
[0028] U={u1,u2,…,u L};
[0029] Y={y1,y2,…,y L};
[0030] C hid ={Γ};
[0031] Γ={τ1,τ2,…,τ L};
[0032] Among them, C obs is the observed data set, U is the input data set, u L is the Lth input data, Y is the output data set, y L is the Lth output data, L is the data length; C hid is a random time delay hidden data set, Γ is the delay at each moment, s is the number of iterations, E is the expectation, θ s is the model parameter at the iteration number s.
[0033] Furthermore, the specific step of decomposing and simplifying the Q function based on the linear regression equation adopts the following formula:
[0034]
[0035] in,
[0036]
[0037]
[0038] Where d is the value of random delay, The random delay τ at the iteration number s k = probability of d, σ 2 is the noise variance.
[0039] Furthermore, the specific steps of iteratively calculating the estimated values of the model parameters based on the conditional expectation until convergence include:
[0040] Based on the conditional expectation of the model parameters, an iteration formula is determined, which is as follows:
[0041]
[0042] in,
[0043] Initialize the model parameters and noise-free output parameters, and use the iterative formula to solve the model parameters until the set convergence conditions are met.
[0044] To achieve the above object, the present application also provides an electronic device, including:
[0045] processor;
[0046] a memory having stored thereon one or more computer program instructions executed on the processor;
[0047] When the processor runs the computer instructions, the parameter identification method of the random time-delay output error system as described above is executed.
[0048] To achieve the above objectives, the present application provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, the steps of the parameter identification method of the random time-delay output error system described above are executed.
[0049] The present application discloses a method for identifying parameters of a random time-delay output error system. By configuring noise-free output parameters, the OE model is constructed as a linear regression equation. The time delay at any time is used as a hidden variable. The expectation maximization algorithm is used to simultaneously identify the model parameters and the random time delay during the parameter identification process, and the identification effect is good.
[0050] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent from the description, or may be understood by practicing the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings:
[0052] Figure 1 A flow chart of a method for identifying parameters of a random time-delay output error system of the present application;
[0053] Figure 2 It is a schematic diagram of simulation comparison of input and output data;
[0054] Figure 3 A schematic diagram showing the comparison of the model parameters and the time delay estimation;
[0055] Figure 4 is a graph of the estimated values of model parameters;
[0056] Figure 5 Comparison graph of output data for self-verification;
[0057] Figure 6 A comparison graph of the self-verification delay estimation;
[0058] Figure 7 It is a comparison curve chart of RMSE indicator;
[0059] Figure 8 Schematic diagram for comparison of robustness;
[0060] Fig. 9 This is a comparison curve diagram of cross-validation delay estimation;
[0061] Fig.10 This is a comparison curve chart of the cross-validation output data. DETAILED DESCRIPTION
[0062] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.
[0063] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.
[0064] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.
[0065] It should be noted that the modifications of "one" and "plurality" mentioned in this application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more". "Plurality" should be understood as two or more.
[0066] Hereinafter, embodiments of the present application will be described in detail with reference to the accompanying drawings.
[0067] Example 1
[0068] An embodiment of the present application provides a method for identifying parameters of a random time-delay output error system. Figure 1 The method for identifying parameters of the output error system of the random time delay of the present application is described in detail.
[0069] Step S101: Based on the OE model expression of the output error system, define the noise-free output parameters and model parameters, and generate the linear regression equation of the output error system;
[0070] For an output error system, its OE model expression is generally:
[0071]
[0072] Among them, k is the time, u k is the input sequence of the OE model, y k is the output sequence of the OE model, e k is zero-mean random white noise with a noise variance of σ 2 ; A(z) and B(z) are polynomials about the unit shift operator, z is the unit backshift operator, is the term coefficient of A(z), n a is the order of A(z), is the term coefficient of B(z), n b is the order of B(z).
[0073] According to the OE model expression of the output error system, the specific steps of generating the linear regression equation include:
[0074] Based on the unit shift operator in the OE model expression, the noise-free output parameter is defined as follows:
[0075]
[0076] Among them, τ k is the random time delay at time k, which is uniformly distributed in the interval [α, β], and its value probability expression is:
[0077]
[0078] Among them, d is the value of random delay, and α and β are interval values.
[0079] It should be noted that due to problems such as sensor signal acquisition and network transmission, there is a time delay phenomenon in the industrial process. Therefore, in this embodiment, the time delay is taken into account.
[0080] Due to x k is the noise-free output parameter, then the model expression under white noise can be directly obtained:
[0081] y k =x k +e k ;
[0082] In order to make the expression a linear regression equation, model parameters and information vectors are defined based on the noise-free output parameters, so that the noise-free output parameters are the product of the model parameters and the information vector.
[0083] The applicant found that by multiplying both sides of the expression of the noise-free output parameter by A(z) and expanding the polynomials A(z) and B(z), we can obtain:
[0084]
[0085] Therefore, define:
[0086]
[0087] in, is the information vector and θ is the model parameter.
[0088] Furthermore, the noise-free output parameters can be rewritten as:
[0089]
[0090] Based on the rewritten noise-free output parameters, the model expression under white noise is rewritten to generate the linear regression equation of the output error system, which is as follows:
[0091]
[0092] Step S102: Based on the observed data set of the output error system and the linear regression equation, an expectation maximization algorithm is used to determine the conditional expectation of the output error system with respect to the model parameters;
[0093] First, collect the observation data set C of the output error system obs = {U, Y}, U = {u1,u2,…,u L}, Y={y1,y2,…,y L}, U is the input data set, u L is the Lth input data, Y is the output data set, y L is the Lth output data, L is the data length, and a random delay hidden data set C corresponding to the length L is constructed hid =C hid ={Γ},Γ={τ1,τ2,…,τ L}, Γ is the delay at each moment;
[0094] The expectation maximization algorithm is used to construct the Q function for calculating the conditional expectation of the current model parameters;
[0095] The Expectation-Maximization Algorithm (EM algorithm) is an iterative optimization algorithm that is mainly used to estimate the parameters of probability models containing latent variables. It has a wide range of applications in machine learning and statistics, including but not limited to Gaussian Mixture Model (GMM), Hidden Markov Model (HMM), and various clustering and classification problems. Its probability model is an iterative algorithm used to describe the relationship between observable data and latent structure to solve the maximum likelihood estimation (MLE) problem (incomplete data, containing latent variables).
[0096] In this embodiment, the random delay hidden data set is used as the hidden variable, the collected observation data set is used as the observation data, the model parameter θ is used as the estimated parameter, and the expectation maximization algorithm is used to construct the Q function. The formula is as follows:
[0097]
[0098] Where s is the number of iterations, E is the expectation, θ s is the model parameter at the iteration number s.
[0099] Based on the linear regression equation, decomposing and simplifying the Q function;
[0100] Specifically, for logp(C obs ,C hid |θ) is decomposed according to the probability chain rule to obtain:
[0101]
[0102] Among them, since the term logp(U|θ} is a constant, denoted as C, y k Depends only on τ k ,u k-1:1 ,y k-1:1 ,θ, so logp(Y|Γ,U,θ) can be expressed as:
[0103] logp(Y|Γ,U,θ)=
[0104] logp(y L |y L-1:1 ,Γ,u L:1 ,θ)+logp(y L |y L-1:1 ,Γ,uL:1 ,θ)+…+logp(y1|Γ,u L:1 ,θ);
[0105] According to the linear regression equation, y k Depends on the random time delay τ k , information vector and model parameters θ, therefore, can be further simplified to;
[0106]
[0107] Similarly, the term logp(Γ|U,θ) is independent of model parameters, inputs, and outputs. So it can be further decomposed into:
[0108]
[0109] Furthermore, it can be simplified to:
[0110]
[0111] in,
[0112]
[0113] in, The random delay τ at the iteration number s k = the probability of d.
[0114] It can be understood that step S102 corresponds to step E of the expectation-maximization algorithm.
[0115] Step s103: Based on the conditional expectation, iteratively calculate the estimated values of the model parameters until convergence.
[0116] Specifically, according to the conditional expectation of the model parameters, an iterative formula is determined to calculate the Q function for the model parameters θ and σ respectively. 2 The first-order partial derivative of and set it to 0, respectively, to obtain the model parameters θ and σ 2 The iteration formula is:
[0117]
[0118] in,
[0119]
[0120] Then, the model parameters and noise-free output parameters are initialized, and the model parameters are solved using an iterative formula until the set convergence conditions are met;
[0121] In this embodiment, due to There are unknown quantities in the noiseless output parameter, so we construct an intermediate quantity with the same structure and order Instead of the noise-free output parameters, the auxiliary model output is updated using the parameter estimates calculated in the most recent iteration, as follows:
[0122]
[0123] In this embodiment, the model parameters are initialized θ0=0.01I, I represents a matrix of all 1s;
[0124] initialization Set the iteration condition ∈, set the number of iterations S = 1, start the iteration, and the iteration process is as follows: Construct the current iteration Then calculate the Q function Then get Further update and delay estimates Compare and like Let s = s + 1 and continue iterating; otherwise, output Get the estimated values of model parameters and end the iteration.
[0125] It can be understood that step S103 corresponds to the M step of the expectation-maximization algorithm.
[0126] For example, the OE model expression of an output error system is Among them, the input data u k is a set of multi-level random hopping sequences, the noise e k is zero and has variance σ 2 =0.1 2 Gaussian sequence with time delay τ k The value range of is set to [0,3], and a batch of input and output data sets with a data length of L = 800 are collected. The input and output data of the simulation are as follows Figure 2 shown.
[0127] Figure 3 It is a schematic diagram for comparing the estimation of model parameters and time delay, which is used to compare the method (EM) of the present application with the matrix transformation expectation maximization (MTEM) algorithm in terms of parameter estimation, time delay accuracy and estimation error.
[0128] Among them, δ represents the relative error between the estimated parameters and the true parameters
[0129]
[0130] like Figure 3As shown in the figure, the parameter identification method proposed in this application has a smaller estimation error and a higher delay accuracy, where T represents the time delay accuracy. When the relative error δ of the parameter is less than 1×10 -3 $ stops the batch iteration, and the final estimate of the parameter is The length of the displayed data interval is [200, 500].
[0131] See also Figure 4 ,As the number of iterations increases, the estimated values of the model parameters gradually approach the true values.
[0132] In this embodiment, the root mean square error (RMSE) is the difference between the observed value and the estimated value.
[0133] RMSE is calculated as follows
[0134]
[0135] Comparison of the self-verification output data of the two methods can be found in Figure 5 , for a comparison of self-verified delay estimates, see Figure 6 , the iterative prediction results refer to Figure 7 The RMSE of the data predicted by the MTEM algorithm is 0.25041, and that of the method of the present application is 0.01546. The algorithm of the present application combined with the auxiliary model method has a good effect in identifying the OE model with random time lag, and the accuracy of time lag estimation is 91.43%.
[0136] In this embodiment, cross validation is also used to further test the predictive ability of the method of the present application, and the cross validation data set changes the random time delay set to [1,4]. The cross validation results of the time delay are shown in Fig. 9 As shown, the accuracy of the method of the present application is relatively high, reaching 91.43%. The cross-validation results of the output data are shown in Fig.10 The RMSE of the data predicted by the MTEM algorithm is 0.27061, and that of the method of this application is 0.03523.
[0137] See also Figure 8 , Figure 8 The robustness comparison results of the two algorithms under different noises are shown in Figure 2, among which the method of the present application has stronger robustness.
[0138] In summary, the method of the present application has a high accuracy rate in random time delay estimation, reaching 91.43%, while the accuracy rate of random time delay estimation of the MTEM algorithm is only 64.51%. The method of the present application demonstrates the ability to accurately estimate random time delays, and the method of the present application has stronger robustness.
[0139] Example 2
[0140] In this embodiment, an electronic device is also provided, including a processor and a memory. The memory is used to store non-temporary computer-readable instructions. The processor is used to run the non-temporary computer-readable instructions, and when the non-temporary computer-readable instructions are run by the processor, one or more steps of the parameter identification method of the random time-delay output error system described above can be executed. The memory and the processor can be interconnected through a bus system and / or other forms of connection mechanisms.
[0141] For example, the processor can be a central processing unit (CPU), a digital signal processor (DSP), or other forms of processing units with data processing capabilities and / or program execution capabilities, such as a field programmable gate array (FPGA); for example, the central processing unit (CPU) can be an X86 or ARM architecture, etc.
[0142] For example, the memory may include any combination of one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), portable compact disk read-only memory (CD-ROM), USB memory, flash memory, etc. One or more computer program modules may be stored on the computer-readable storage medium, and the processor may run one or more computer program modules to implement various functions of the electronic device. Various applications and various data, as well as various data used and / or generated by the application, etc. may also be stored in the computer-readable storage medium.
[0143] It should be noted that, in the embodiments of the present application, the specific functions and technical effects of the electronic device can refer to the description of the above method for identifying system parameters of the output error of the random delay, which will not be repeated here.
[0144] Example 3
[0145] In this embodiment, a computer-readable storage medium is also provided, and the storage medium is used to store non-transitory computer-readable instructions. For example, when the non-transitory computer-readable instructions are executed by a computer, one or more steps in the parameter identification method of the random time-delay output error system described above can be executed.
[0146] For example, the storage medium can be applied to the above-mentioned electronic device. For example, the storage medium can be the memory in the electronic device of Example 2. For example, the relevant description of the storage medium can refer to the corresponding description of the memory in the electronic device of Example 2, which will not be repeated here.
[0147] The above description is only a partial embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.
[0148] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.
[0149] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.
Claims
1. A parameter identification method for a random time-delay output error system, characterized in that: include: Based on the OE model expression of the output error system, the noise-free output parameters and model parameters are defined to generate the linear regression equation of the output error system; Based on the observed data set of the output error system and the linear regression equation, an expectation maximization algorithm is used to determine the conditional expectation of the output error system with respect to the model parameters; Based on the conditional expectation, estimated values of the model parameters are iteratively calculated until convergence.
2. The parameter identification method of the random time-delay output error system according to claim 1, characterized in that: The specific steps of defining the noise-free output parameters and model parameters based on the OE model expression of the output error system and generating the linear regression equation of the output error system include: Based on the unit shift operator in the OE model expression, the noise-free output parameter is defined using the following formula: Among them, k is the time, u k is the input sequence of the OE model, y k is the output sequence of the OE model, e k is zero-mean random white noise; A(z) and B(z) are polynomials about the unit shift operator, z is the unit backshift operator, a1…a na is the term coefficient of A(z), n a is the order of A(z), is the term coefficient of B(z), n b is the order of B(z); τ k is the random delay at time k, which obeys uniform distribution; Based on the noise-free output parameter, a model parameter and an information vector are defined so that the noise-free output parameter is the product of the model parameter and the information vector, and the formula is as follows: in, is the information vector, θ is the model parameter; in, Based on the noise-free output parameters, a linear regression equation of the output error system is generated, and the formula is as follows:
3. The parameter identification method of the random time-delay output error system according to claim 2, characterized in that: The specific steps of using the expectation maximization algorithm to determine the conditional expectation of the output error system with respect to the model parameters based on the observed data set of the output error system and the linear regression equation include: Collect the observed data set of the output error system and construct a random delay hidden data set of corresponding length: The expectation maximization algorithm is used to construct the Q function for calculating the conditional expectation of the current model parameters; Based on the linear regression equation, the Q function is decomposed and simplified.
4. The parameter identification method of the random time-delay output error system according to claim 3 is characterized in that: The specific steps of using the expectation maximization algorithm to construct a Q function for calculating the conditional expectation of the current model parameters are as follows: C obs ={U,Y}; U={u1,u2,…,u L }; Y={y1,y2,…,y L }; C hid ={C}; C={τ1,τ2,…,τ L }; Among them, C obs is the observed data set, U is the input data set, u L is the Lth input data, Y is the output data set, y L is the Lth output data, L is the data length; C hid is a random time delay hidden data set, Γ is the delay at each moment, s is the number of iterations, E is the expectation, θ s is the model parameter at the iteration number s.
5. The parameter identification method of the random time-delay output error system according to claim 4, characterized in that: The specific steps of decomposing and simplifying the Q function based on the linear regression equation are as follows: in, Where d is the value of random delay, The random delay τ at the iteration number s k = probability of d, σ 2 is the noise variance.
6. The parameter identification method of the random time-delay output error system according to claim 5, characterized in that: The specific steps of iteratively calculating the estimated values of the model parameters based on the conditional expectation until convergence include: Based on the conditional expectation of the model parameters, an iteration formula is determined, which is as follows: in, Initialize the model parameters and noise-free output parameters, and use the iterative formula to solve the model parameters until the set convergence conditions are met.
7. An electronic device, characterized in that: include: processor; a memory having stored thereon one or more computer program instructions executed on the processor; Wherein, when the processor runs the computer instructions, it executes the parameter identification method of the random delay output error system described in any one of claims 1-6.
8. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are executed by the processor, the steps of the parameter identification method of the random time-delay output error system described in any one of claims 1-6 are executed.
