A data-driven modeling method for secondary side voltage control of BDWPT system

Abstract

This invention discloses a data-driven modeling method for secondary-side voltage control in a BDWPT system. The method includes: transforming the equivalent circuit of a bilaterally SS-compensated BDWPT system into a circuit structure in which both the primary and secondary sides utilize series resonance; using a phase shift angle as a control variable to regulate the load voltage; and employing the SRIVC algorithm to estimate the parameter θ of the data-driven model from sampled data, where N represents the number of sampled data. Compared to traditional methods, this data-driven modeling method no longer focuses on system operation, relying solely on sampled input and output data. This modeling method has considerable versatility.

Description

Technical Field

[0001] The present invention belongs to the field of power electronics and relates to magnetic-coupled wireless power transfer (MC-WPT) technology, and specifically to a data-driven WPT system secondary side voltage control modeling method. Background Art

[0002] Wireless Power Transfer (WPT) systems primarily use frequency control and phase-shift control for voltage control. Frequency control often complicates the design of electromagnetic interference (EMI) filters, while phase-shift control, operating at a constant switching frequency, eliminates EMI issues.

[0003] Unidirectional WPT systems typically use a full-bridge inverter as the input stage and an uncontrolled rectifier as the output stage. Voltage control relies primarily on the primary-side full-bridge inverter. Unlike unidirectional WPT systems, bidirectional wireless power transfer (BDWPT) systems use an active rectifier as the output stage, allowing for additional control of the active rectifier to achieve voltage control. Compared to primary-side phase-shift control, the output voltage of secondary-side phase-shift control is directly the controlled voltage, simplifying the control process.

[0004] The WPT system has strict requirements on the output voltage and cannot do without closed-loop control with a controller. The design of the controller requires reference to a specific model. Based on the traditional circuit principle, the model is obtained. Even for the simplest SS topology, the model order reaches 7th order. At the same time, the nonlinearity of power electronic components is not considered, and the model is not suitable for dynamic analysis. The state space averaging (SSA) model takes into account the modeling of power electronic components, but it is only suitable for WPT systems with small ripple. The extended describing function (EDF) method uses a describing function to approximate nonlinear terms and obtains a dynamic model through harmonic balance. Although the feasibility of the EDF method has been proven, the model established is of high order and the derivation process is cumbersome.

[0005] In summary, both traditional circuit modeling and models derived from SSA or EDF suffer from high-order and complex derivations, and their applicability is limited. Traditional modeling methods are helpful for understanding system operating mechanisms, but are prohibitively expensive for control purposes.

[0006] Therefore, it is necessary to establish a model with lower order, simpler derivation process and lower control cost. Summary of the Invention

[0007] In view of this, the purpose of the present invention is to provide a data-driven BDWPT system secondary voltage control modeling method. This method uses the phase shift angle as input and the load voltage as output, samples the system input and output data, and uses a system identification algorithm to identify the unknown model parameters to obtain an equivalent transfer function model. This objective of the present invention is achieved through the following technical solutions:

[0008] A data-driven BDWPT system secondary side voltage control modeling method includes:

[0009] The equivalent circuit of the bilateral SS compensated BDWPT system is equivalent to a circuit structure in which both the primary and secondary sides adopt series resonance;

[0010] The phase shift angle is used as a control variable to adjust the load voltage. The phase shift angle α represents the offset angle of the secondary switch tube relative to the primary switch drive signal. The relationship between the phase shift angle and the load voltage is expressed as follows:

[0011]

[0012] in,

[0013] B(p,θ)=b0p m +b1p m-1 +...+b m ,

[0014] A(p,θ)=a0p n +a1p n-1 +...+1,(n≥m),

[0015] t k represents the sampling time, k∈N + ,u(t k )=α(t k ), indicating t k The phase shift angle at the moment, G(p,θ) represents the data-driven model, x(t k ) represents the output of the data-driven model, e(t k ) are two zero-mean, uncorrelated measurement noise sequences, y(t k ) represents t k The load voltage output by the system at the moment, θ=[a0…a n-1 b0…b m ] T ∈R n+m+2 The parameters of the data-driven model, n and m are the polynomial orders;

[0016] Using the SRIVC algorithm, using sampled data Estimate the parameters θ of the data-driven model, where N represents the number of sampled data.

[0017] Furthermore, estimating the data-driven model parameters θ includes:

[0018] S1: Initialization, that is, setting the initial filter F(p) and using the least squares method to calculate the initial parameters in, ω f Indicates the system cutoff frequency;

[0019] S2: Iterative estimation, including:

[0020] For the jth iteration,

[0021] Using the j-1th iteration based on the relationship The obtained polynomial and The established auxiliary model Generate auxiliary variable sequence

[0022] use For input u(t k ), output y(t k ) and auxiliary variables Perform filtering;

[0023] Sampling the filtered signal yields the vector φ f (t k ), IV vector and the nth-order output filter derivative

[0024] Based on the above data, generate the latest estimate of the data-driven model parameter θ Continue iterating until convergence.

[0025] in,

[0026]

[0027]

[0028] (·) f represents the result after filtering using the filter F(p,θ). Indicates t k For the output y(t k )’s i-th order filter derivative; Indicates t kThe output x(t k ) is the i-th order filtered derivative of .

[0029] Furthermore, the error minimization is used to estimate the data-driven model parameters θ. Specifically, the latest estimate of the data-driven model parameters θ is generated. include:

[0030] It is estimated according to the following formula:

[0031]

[0032] Where: ||x|| 2 =x T Qx,Q=I,

[0033] Obtain:

[0034]

[0035] The beneficial effects of the present invention are:

[0036] The method provided by the present invention uses the phase shift angle as input and the load voltage as output, samples the system input and output data, and uses the system identification algorithm to identify the unknown parameters of the model to obtain an equivalent transfer function model. This method is a dynamic modeling method that can predict the system output and be used for dynamic analysis of system operation. Moreover, compared with traditional methods, data-driven modeling no longer focuses on the system operation status and only relies on the system sampled input and output data. Therefore, the modeling method has a certain degree of versatility.

[0037] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0039] Figure 1a ) is the circuit diagram of the BDWPT system with bilateral SS compensation;

[0040] Figure 1b ) is the equivalent circuit diagram of the BDWPT system with bilateral SS compensation;

[0041] Figure 2 is a schematic diagram of the phase shift angle α;

[0042] Figure 3is the phase shift angle α to the load voltage U dc2 Relationship diagram;

[0043] Figure 4 It is a schematic block diagram of voltage control;

[0044] Figure 5 It is a data-driven model framework diagram;

[0045] Figure 6 It is a Simulink simulation diagram;

[0046] Figure 7 It is the input and output data diagram of Simulink simulation;

[0047] Figure 8 This is a comparison chart of the Simulink model and the data-driven model results. DETAILED DESCRIPTION

[0048] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the preferred embodiments are only for illustrating the present invention, and are not intended to limit the scope of protection of the present invention.

[0049] The resonant topology of the WPT system with bilateral SS compensation uses series resonance on both the primary and secondary sides. Its structure is shown in Figure 1, where Figure 1a) shows the circuit topology of the BDWPT system, and Figure 1b) shows the equivalent diagram of the basic structure of the BDWPT system. This application uses a new method to establish a dynamic model of the voltage control of the active rectifier on the secondary side of the BDWPT, using a phase shift angle α (such as Figure 2 As shown) as the control variable to adjust the load voltage U dc2 The phase shift angle α represents the offset angle of the secondary switch tube relative to the primary switch drive signal.

[0050] Phase shift angle α and load voltage U dc2 The relationship between Figure 3 As shown. Figure 3 It can be seen that the phase shift angle α is related to the load voltage U dc2 There is a certain relationship between them, which can be represented by a control model. Figure 4 It is the voltage control flow chart.

[0051] The present invention determines the phase shift angle α and the load voltage U by a data-driven method. dc2 The most relevant model G(s). This method relies only on the input and output data of the system and does not require accurate circuit parameters. The controlled input variables are generated by the controller, so it can be assumed that the measured input is noise-free, that is,

[0052] u(t k )=α(t k ) (1)

[0053] Among them, u(t k ) represents t k The input at time α(t k ) represents t k The phase shift angle at time.

[0054] In the absence of noise, the output voltage V0 can be expressed as:

[0055] x(t k )=V0(t k ) (2)

[0056] Among them, x(t k ) represents t k The output of the model is driven by the data at each moment.

[0057] However, disturbances in the system output are unavoidable, so the measured output is expressed as follows:

[0058] y(t k )=x(t k )+e(t k ) (3)

[0059] where t k =kT is the sampling time, T is the sampling period, k∈N + ; and e(t k ) are two zero-mean, uncorrelated measurement noise sequences.

[0060] The model can be re-described based on the input and output, such as Figure 5 shown.

[0061] The model can be expressed as follows:

[0062]

[0063] Where B(p,θ) and A(p,θ) are the following polynomials:

[0064] B(p,θ)=b0p m +b1p m-1 +…+b m (5)

[0065] A(p,θ)=a0p n +a1p n-1 +…+1,(n≥m) (6)

[0066] in θ=[a0…a n-1 b0…b m ] T ∈R n+m+2is the unknown parameter vector, and n and m are the polynomial orders.

[0067] Using the SRIVC algorithm, using sampled data Estimate the parameters θ of the data-driven model, where N represents the number of sampled data.

[0068] In some embodiments, estimating the data-driven model parameters θ may include:

[0069] S1: Initialization, that is, setting the initial filter F(p) and using the least squares method to calculate the initial parameters in, ω f Indicates the system cutoff frequency;

[0070] S2: Iterative estimation, including:

[0071] For the jth iteration,

[0072] Using the j-1th iteration based on the relationship The obtained polynomial and The established auxiliary model Generate auxiliary variable sequence

[0073] use For input u(t k ), output y(t k ) and auxiliary variables Perform filtering;

[0074] Sampling the filtered signal yields the vector φ f (t k ), IV vector and the nth-order output filter derivative

[0075] Based on the above data, generate the latest estimate of the data-driven model parameter θ Continue iterating until convergence.

[0076] For the SRIVC algorithm, consider the following output error minimization problem:

[0077]

[0078] ε(t k )=y(t k )-x(t k ) (8)

[0079] Among them, ε(t k ) is the output error.

[0080] It can be written in filtered linear regression form:

[0081]

[0082] in:

[0083]

[0084] in,(·) f represents the result obtained by using the filtered form of the following filter, which can be expressed as:

[0085]

[0086] The SRIVC method uses an iterative procedure for estimation. In each iteration, the auxiliary model is used to generate instrumental variables and the prefilter is updated based on the estimated parameters of the previous iteration. The auxiliary model used in the jth iteration is based on The output of the auxiliary model is:

[0087]

[0088] is φ T (t) Noise-free form.

[0089]

[0090] in, Indicates t k For the output y(t k )’s i-th order filter derivative; Indicates t k The output x(t k ) is the i-th order filtered derivative of .

[0091] According to formula (7), the IV optimization problem is expressed in the following form:

[0092]

[0093] Where: ||x|| 2 =x T Qx,Q=I.

[0094] The solution of formula (13) is:

[0095]

[0096] The accuracy of the model of the present invention is verified by using specific examples below.

[0097] Figure 6It is the Simulink simulation diagram. Table 1 shows the main parameters in the simulation diagram.

[0098] Table 1 Main parameters in the simulation diagram

[0099]

[0100] The input of the system is selected by the user, but in order to fully stimulate the system and meet the input requirements of the actual system, white noise is selected as the input of the system:

[0101]

[0102] rand(t k ) represents time t k A random value between [0,1].

[0103] Add DT white noise with a variance of 0.001 to the output measurement, and the input and output data of the Simulink simulation used for estimation are as follows Figure 7 shown.

[0104] In the presence of output measurement noise, the algorithm still achieves unbiased estimation of model parameters because the auxiliary variables can eliminate the influence of noise. Finally, the system is determined to be a second-order model, that is, n = 2 and m = 1, and

[0105]

[0106] The simulation time of the Simulink simulation system is 0.2 seconds, and a total of N = 2001 data are obtained. The algorithm converges after 17 iterations, and the parameter vector obtained is

[0107] θ=[1 1018 6.567×10 4 1.269×10 6 9.189×10 6 ] T (18)

[0108] Figure 8 The output of the estimated model and the real system are compared. The accuracy of the model is evaluated by performance indicators:

[0109]

[0110] Among them, y s (t k ) is the real system output, y d (t k ) is based on the output of the data-driven model.

[0111] according to Figure 8As can be seen, the comparison between the output of the estimated model and the real system shows that the model achieves very high accuracy (i.e., ), thus confirming the effectiveness of the data-driven model of the present invention.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A data-driven BDWPT system secondary side voltage control modeling method, characterized in that: include: The equivalent circuit of the bilateral SS compensated BDWPT system is equivalent to a circuit structure in which both the primary and secondary sides adopt series resonance; The phase shift angle is used as a control variable to adjust the load voltage. The phase shift angle represents the offset angle of the secondary switch tube relative to the primary switch drive signal. The relationship between the phase shift angle and the load voltage is expressed as follows: in, B(p,θ)=b0p m +b1p m-1 +...+b m , A(p,θ)=a0p n +a1p n-1 +…+1,(n≥m), t k represents the sampling time, k∈N + ,u(t k )=α(t k ), indicating t k The phase shift angle at the moment, G(p,θ) represents the data-driven model, x(t k ) represents the output of the data-driven model, e(t k ) are two zero-mean, uncorrelated measurement noise sequences, y(t k ) represents t k The load voltage output by the system at the moment, θ=[a0…a n-1 b0…b m ] T ∈R n+m+2 The parameters of the data-driven model, n and m are the polynomial orders; Using the SRIVC algorithm, using sampled data Estimate the parameters θ of the data-driven model, where N represents the number of sampled data; Estimating the data-driven model parameters θ involves: S1: Initialization, that is, setting the initial filter F(p) and using the least squares method to calculate the initial parameters in, ω f Indicates the system cutoff frequency; S2: Iterative estimation, including: For the jth iteration, Using the j-1th iteration based on the relationship The obtained polynomial and The established auxiliary model Generate auxiliary variable sequence use For input u(t k ), output y(t k ) and auxiliary variables Perform filtering; Sampling the filtered signal yields the vector φ f (t k ), IV vector and the nth-order output filter derivative Based on the above data, generate the latest estimate of the data-driven model parameter θ Continue iterating until convergence. in, (·) f represents the result after filtering using the filter F(p,θ). Indicates t k For the output y(t k )’s i-th order filter derivative; Indicates t k The output x(t k ) is the i-th order filtered derivative of .

2. The data-driven BDWPT system secondary voltage control modeling method according to claim 1 is characterized in that: The method includes estimating data-driven model parameters θ using error minimization.

3. The data-driven BDWPT system secondary voltage control modeling method according to claim 2 is characterized in that: Generate the latest estimate of the parameters θ of the data-driven model include: It is estimated according to the following formula: Where: ‖x‖ 2 =x T Qx,Q=I, Obtain:

Images