A design method and system of LMS-SOGI three-phase phase-locked loop suitable for non-ideal power grid

Abstract

The application provides a LMS-SOGI three-phase phase-locked loop design method and system suitable for a non-ideal power grid, and the method comprises the following steps: sampling three-phase grid-connected voltages of a grid-connected inverter to obtain voltage components through Clark transformation; the voltage components are respectively sent into two SOGI structures to extract positive sequence components of the grid-connected voltages; a least mean square LMS filter link mathematical model is constructed to design a least mean square LMS filter link step parameter; then, the positive sequence components of the grid-connected voltages are subjected to LMS filtering to obtain d-axis and q-axis positive sequence components of a fundamental frequency grid-connected voltage, the q-axis component of the fundamental frequency grid-connected voltage is subtracted by 0 to obtain a q-axis voltage phase-locked error signal, an output signal angular frequency adjustment amount ω is obtained through a PI controller, and a phase θ is obtained through integration. Finally, a three-phase grid voltage LMS-SOGI phase-locked model is built in MATLAB / Simulink simulation software, and different grid voltages are selected for comparative analysis of the filtering effect. The application solves the technical problem that the phase-locked technology of the grid-connected inverter is difficult to accurately lock the voltage frequency and phase in a specific scene.

Description

Technical Field

[0001] This invention relates to the field of phase-locked loop control for grid-connected inverters, and specifically to an LMS-SOGI three-phase phase-locked loop design method and system suitable for non-ideal power grids. Background Technology

[0002] Currently, the demand for energy resources is constantly increasing due to socio-economic development. To reduce dependence on traditional fossil fuels, my country is continuously expanding the scale of new energy power generation connected to the grid. With the large-scale integration of new energy inverters, the equivalent impedance of the grid from the inverter's perspective increases, leading to increased voltage harmonic content and frequency changes in the grid. This affects the stability of grid operation and degrades power quality.

[0003] Most power electronic devices are connected to a three-phase power grid, and three-phase phase-locked loops (PLLs) are required for synchronization with the grid. Traditional PLLs use hardware-based PLLs to track the voltage phase by detecting zero-crossing points. This method is unsuitable for applications with severe harmonic pollution and can only achieve single-phase PLLs, failing to track the positive-sequence components of the system. Currently, the main three-phase software phase-locked loop methods are: (1) single / dual synchronous coordinate system software phase-locked loops, such as the existing invention patent application document "Second-order decoupled dual synchronous coordinate system phase-locked loop control method for distribution network loop-closing equipment" published by CN115000945A, which includes: representing the three-phase grid voltage based on the fact that the three-phase grid voltage contains only positive and negative sequence fundamental components; transforming the grid voltage from the three-phase natural coordinate system to the two-phase αβ coordinate system; the positive and negative sequence Park converter is a mathematical transformation from the two-phase stationary coordinate system to the two-phase rotating coordinate system; obtaining the transfer function of the second-order low-pass filter SOLPF; combining the second-order low-pass filter SOLPF in step 3, deriving the transfer function expression corresponding to the decoupling network DN; the positive and negative sequence d and q axis voltages output by the decoupling network DN; using the loop filter LF as a PI regulator, and obtaining the transfer function. The aforementioned prior art can only accurately detect the phase, frequency, and amplitude of the grid voltage when the grid voltage is balanced; (2) A single synchronous coordinate system software phase-locked loop system based on the symmetrical component method, such as the prior invention patent application document "A DSTATCOM Current Detection Method Based on Instantaneous Symmetrical Component Method" published by CN102081114A, includes: 1. First, determining the symmetrical component method and the representation method of instantaneous values ​​in the phasor time domain; 2. Determining the instantaneous values; 3. Improving the implementation of the instantaneous symmetrical component method; 4. Using MATLAB simulation tools to establish a model to simulate and analyze the above conclusions; 5. Filtering, conversion, and processing of data; 6. Subtracting the obtained three-phase fundamental positive sequence active currents iafp+, ibfp+, and icfp+ from the load current yields the required comprehensive compensation command currents iac, ibc, and icc, which include harmonics, negative sequence, and reactive power. The aforementioned existing technologies can only suppress the second harmonic effect caused by the negative sequence component in the grid voltage; (3) The decoupled software phase-locked loop based on the dual synchronous coordinate system can effectively overcome the influence of frequency change on the phase-locked loop, but the harmonic suppression mainly relies on the first-order low-pass filter composed of PI control, and the filtering effect is generally poor; (4) The software phase-locked loop (SOGI) based on the dual second-order generalized integral uses the characteristics of the trigonometric function itself to filter out the harmonic influence through the orthogonal signal generator, and at the same time generates orthogonal signals with a phase difference of 90° to separate the positive and negative sequence components. Its phase-locked loop can meet the phase-locking requirements under normal conditions, but when the grid quality deteriorates, the grid is unbalanced, or the amplitude of a large number of harmonics is not ideal, the above-mentioned traditional phase-locked loops are difficult to meet the requirements, and even SOGI phase-locking is difficult to meet its filtering effect.Currently, several academic papers have analyzed and reported on improved software phase-locked loop (PLL) control schemes based on dual second-order generalized integrators. Examples include: 1. Adding an integrator to the original architecture unit to form a third-order generalized integrator, used to suppress harmonics and high-frequency signals; 2. Adding a low-pass filter before the SOGI architecture unit to enhance its filtering function. For instance, the existing invention patent document CN107623522A, "A Dual Second-Order Generalized Integrator PLL Control Method Based on dq Transform," describes steps A: obtaining voltages Uα and Uβ from the sampled three-phase voltages Ua, Ub, and Uc through Clark transformation; Step B: feeding uα and uβ into two SOGI structures to extract the positive-sequence component of the grid voltage; Step C: filtering the positive-sequence component of the grid voltage; and Step D: finally feeding the filtered positive-sequence component of the grid voltage into a PLL based on the dq coordinate system as a reference for the PLL. As can be seen from the specific implementation of the aforementioned prior art, in order to reduce the influence of negative sequence components and obtain better steady-state accuracy, the cutoff frequency of the loop filter must be very low, which greatly affects the speed of dynamic response.

[0004] In summary, existing technologies have the technical problem that the phase-locked loop technology of grid-connected inverters is difficult to accurately lock the voltage frequency and phase in specific scenarios. Summary of the Invention

[0005] The technical problem to be solved by this invention is how to solve the problem that the phase-locked loop technology of grid-connected inverters in the prior art is difficult to accurately lock the voltage frequency and phase in specific scenarios.

[0006] This invention solves the above-mentioned technical problems by employing the following technical solution: A design method for an LMS-SOGI three-phase phase-locked loop suitable for non-ideal power grids, comprising:

[0007] S1. Sample and obtain the three-phase grid-connected voltage U of the grid-connected inverter. a (t), U b (t), U c (t), based on which Clark transformation operation is performed to obtain the grid-connected voltage α and β axis components U. α (t), U β (t);

[0008] S2, convert the grid-connected voltage α and β axis components U α (t), U β (t) The components are fed into two SOGI structures respectively, and the positive sequence components of the grid voltage in the two-phase stationary coordinate system αβ coordinate system are extracted through the SOGI dual second-order generalized integrator transformation operation.

[0009] S3. Construct a mathematical model for the Least Mean Square (LMS) filter stage and design the step size parameters for the LMS filter stage. Step S3 further includes:

[0010] S31. Using the preset LMS adaptive filtering module, the mathematical model relationship of the LMS filtering link is obtained.

[0011] S32. Using the Least Mean Square (LMS) adaptive filtering module, a continuous mathematical model in the time domain is obtained. The preset LMS adaptive filtering module includes: input matrix X(n), expected response input matrix d(n), error vector matrix ε(n), weight vector matrix W(n), step size μ(n), and output matrix y(n).

[0012] S4. Using the mathematical model of the Least Mean Square (LMS) filter stage and based on the step size parameter of the LMS filter stage, the LMS filter processes the positive sequence component of the grid voltage. To obtain the d-axis positive sequence component of the base frequency grid-connected voltage and the positive sequence q-axis component of the base frequency grid-connected voltage

[0013] S5. Convert the q-axis component of the base frequency grid-connected voltage. Subtracting from 0 yields the q-axis voltage phase-locked error signal e. q (t), based on which the angular frequency adjustment ω of the output signal is obtained by processing with a PI controller, and the phase θ is obtained by integration;

[0014] S6. Build a three-phase grid voltage LMS-SOGI phase-locked model in the pre-set simulation software. Based on the output signal angular frequency adjustment ω and phase θ, take different grid voltage scenarios and compare and analyze the filtering effect.

[0015] This invention adds an adaptive filtering logic within the SOGI system to improve the voltage filtering effect of the three-phase power grid and the steady-state accuracy of the phase-locked loop. The SOGI phase-locked loop based on least mean square (LMS) adaptive filtering of this invention effectively avoids the influence of poor loop filter performance on the output frequency and phase of the phase-locked loop under non-ideal power grid conditions, thus improving the stability of the inverter. It can be widely used in various harsh operating conditions and applications requiring high detection speeds, and improves the locking accuracy for voltage frequency and phase.

[0016] In a more specific technical solution, in step S1, based on the real-time three-phase grid-connected voltage U... a (t), U b (t), U c (t), the fundamental voltage in the three-phase stationary coordinate system abc is expressed using the following logic:

[0017]

[0018] In the formula, U a0 (t), U b0 (t), U c0 (t) represents the three-phase grid-connected fundamental voltage in the three-phase stationary coordinate system abc, U N k represents the amplitude of the three-phase balanced voltage of the power grid. b k c ω0 represents the unbalance coefficients of phases B and C in the three-phase unbalanced voltage, and ω0 represents the fundamental angular frequency.

[0019] In a more specific technical solution, step S1 utilizes the following logical Clark transformation operation:

[0020]

[0021] In the formula, U α (t), U β (t) represents the α and β components of the grid voltage in the two-phase stationary coordinate system αβ, T αβ This is the Clark transformation coefficient matrix.

[0022] In a more specific technical solution, in step S2, the SOGI dual second-order generalized integrator transformation operation is performed using the following logic:

[0023]

[0024] In the formula, The positive sequence components of the grid voltage are defined in the two-phase stationary coordinate system αβ. T represents the positive sequence component of the three-phase grid-connected fundamental voltage in the three-phase stationary coordinate system abc coordinate system. + For the symmetric component method coefficient matrix, q refers to the phase shift of the original signal by 90° in the time domain.

[0025] In a more specific technical solution, step S2 is represented by the following logic for the transfer function of the SOGI structure:

[0026]

[0027]

[0028] In the formula, v' and v are the two outputs of the SOGI dual second-order generalized integrator module, ω' is the resonant frequency of the SOGI dual second-order generalized integrator module, and k is the reciprocal of the quality factor.

[0029] This invention includes a three-phase grid voltage adaptive filtering and phase-locked loop (PLL) technology for non-ideal grids. It improves the internal structure of the double second-order generalized integrator SOGI, which is different from the improvement of adding a pre-filter and increasing the order. The LMS least mean square root adaptive filtering module replaces the Park transform module. It is not only equivalent to the function of the Park transform, but also achieves better filtering effect, enabling the subsequent PLL module to track the grid voltage more accurately.

[0030] This module not only constructs two-phase quadrature voltage signals by offsetting the input voltage signal by 90° to obtain the positive sequence component of the grid voltage, but also filters out high-frequency interference signals.

[0031] In a more specific technical solution, in step S31, based on the input matrix X(n), the desired response input matrix d(n), the error vector matrix ε(n), the weight vector matrix W(n), the step size μ(n), and the output matrix y(n), the following logic is used to obtain the mathematical model relationship of the least mean square (LMS) filtering stage:

[0032] y(n)=X T (n)*W(n)

[0033] ε(n)=d(n)-y(n)=d(n)-X T (n)*W(n).

[0034] In order to reduce the influence of negative sequence components and provide steady-state accuracy, the loop filter cutoff frequency must be set to a low value in the traditional technology, which leads to a slow dynamic response. In this invention, μ(n) adopts a variable step size method to overcome the contradiction between dynamic response speed and steady-state accuracy.

[0035] In a more specific technical solution, step S32 uses the following variable step size method to determine the step size:

[0036] p(n)=βp(n-1)+(1-β)ε(n)ε(n-1)

[0037] μ(n+1)=αμ(n)+γp 2 (n)

[0038] In the formula, p(n) is the step size iteration compensation matrix, and α, β, and γ are the step size iteration coefficients.

[0039] The LMS least mean square adaptive filtering module in this invention does not adopt a fixed step size or single output model, but extends to a variable step size and multiple output form. Its high-speed convergence and low steady-state offset characteristics enable high-speed and high steady-state accuracy detection of the amplitude and phase angle of the positive sequence voltage component of the three-phase power grid. Therefore, this phase-locked system can maintain good performance in situations with severe harmonic distortion and three-phase imbalance, and can quickly detect abrupt changes in the amplitude and phase of the positive sequence voltage component of the three-phase voltage, as well as the amplitude and phase angle of the positive sequence voltage component of the three-phase power grid.

[0040] In a more specific technical solution, based on the weight vector matrix W(n), the following logic is used to perform iterative weight transformation to obtain the iterative weights:

[0041] W(n+1)=W(n)+μ(n)ε(n)X(n).

[0042] In a more specific technical solution, step S32 utilizes the following logic: Obtain the continuous time-domain mathematical model of the corresponding phase-locked loop in the current LMS minimum mean square root adaptive filtering module, wherein the continuous time-domain mathematical model includes:

[0043] Input matrix:

[0044]

[0045] Weight vector matrix:

[0046]

[0047] Expected response input matrix:

[0048]

[0049] In the formula, cosωt and sinωt are the phase angle trigonometric functions of the phase-locked loop output. The positive and negative sequence components of the grid voltage in the two-phase rotating coordinate system dq are given.

[0050] In a more specific technical solution, an LMS-SOGI three-phase phase-locked loop design system suitable for non-ideal power grids includes:

[0051] The Clark converter module is used to sample and obtain the three-phase grid-connected voltage U of the grid-connected inverter. a (t), U b (t), U c (t), based on which Clark transformation operation is performed to obtain the grid-connected voltage α and β axis components U. α (t), U β (t);

[0052] SOGI dual second-order generalized integrator converter module is used to convert the grid-connected voltage α and β axis components Uα (t), U β (t) The components are fed into two SOGI structures respectively, and the positive sequence components of the grid voltage in the two-phase stationary coordinate system αβ coordinate system are extracted through the SOGI dual second-order generalized integrator transformation operation. The SOGI dual second-order generalized integrator transform module is connected to the Clark transform module;

[0053] The Least Mean Square (LMS) filter model construction module is used to construct the mathematical model of the LMS filter stage and design the step size parameters of the LMS filter stage. The LMS filter module also includes:

[0054] The LMS filter model relationship acquisition module is used to process and obtain the mathematical model relationship of the minimum mean square LMS filter stage using the preset minimum mean square LMS adaptive filter module.

[0055] The time-domain continuous mathematical model acquisition module is used to acquire the time-domain continuous mathematical model using the least mean square (LMS) adaptive filtering module. The preset LMS adaptive filtering module includes: input matrix X(n), expected response input matrix d(n), error vector matrix ε(n), weight vector matrix W(n), step size μ(n), and output matrix y(n).

[0056] The filtering module utilizes the mathematical model of the Least Mean Square (LMS) filter stage and, based on the step size parameters of the LMS filter stage, performs LMS filtering to process the positive-sequence component of the grid voltage. To obtain the d-axis positive sequence component of the base frequency grid-connected voltage and the positive sequence q-axis component of the base frequency grid-connected voltage The filtering operation module is connected to the Least Mean Square (LMS) filtering model construction module;

[0057] The phase processing module is used to process the q-axis component of the base frequency grid-connected voltage. Subtracting from 0 yields the q-axis voltage phase-locked error signal e. q (t), based on which the angular frequency adjustment ω of the output signal is obtained by processing with a PI controller, and the phase θ is obtained by integration. The phase processing module is connected to the filtering operation module.

[0058] The filtering effect comparison and analysis module is used to build a three-phase power grid voltage LMS-SOGI phase-locked model in the preset simulation software. Based on the output signal angular frequency adjustment ω and phase θ, different power grid voltage scenarios are selected to compare and analyze the filtering effect. The filtering effect comparison and analysis module is connected to the phase processing module.

[0059] Compared with existing technologies, this invention has the following advantages: It adds an adaptive filtering logic within the SOGI system to improve the voltage filtering effect of the three-phase power grid and the steady-state accuracy of the phase-locked loop. The SOGI phase-locked loop based on least mean square (LMS) adaptive filtering effectively avoids the influence of poor loop filter performance on the output frequency and phase of the phase-locked loop under non-ideal power grid conditions, improving the stability of the inverter. It can be widely used in various harsh operating conditions and applications requiring high detection speeds, improving the locking accuracy for voltage frequency and phase.

[0060] This invention includes a three-phase grid voltage adaptive filtering and phase-locked loop (PLL) technology for non-ideal grids. It improves the internal structure of the double second-order generalized integrator SOGI, which is different from the improvement of adding a pre-filter and increasing the order. The LMS least mean square root adaptive filtering module replaces the Park transform module. It is not only equivalent to the function of the Park transform, but also achieves better filtering effect, enabling the subsequent PLL module to track the grid voltage more accurately.

[0061] In order to reduce the influence of negative sequence components and provide steady-state accuracy, the loop filter cutoff frequency must be set to a low value in the traditional technology, which leads to a slow dynamic response. In this invention, μ(n) adopts a variable step size method to overcome the contradiction between dynamic response speed and steady-state accuracy.

[0062] The LMS (Least Mean Square) adaptive filtering module in this invention does not employ a fixed-step-size or single-output model, but rather extends to a variable-step-size and multi-output form. Its high-speed convergence and low steady-state offset characteristics enable high-speed and high-steady-state accuracy detection of the amplitude and phase angle of the positive-sequence voltage components of the three-phase power grid. Therefore, this phase-locked system maintains good performance even in situations with severe harmonic distortion and three-phase imbalance, and can quickly detect abrupt changes in the amplitude and phase of the positive-sequence voltage components of the three-phase power grid, as well as the amplitude and phase angle of the positive-sequence voltage components of the three-phase power grid. This invention solves the technical problem in existing grid-connected inverter phase-locked technology that makes it difficult to accurately lock the voltage frequency and phase in specific scenarios. Attached Figure Description

[0063] Figure 1 This is a flowchart of a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 1 of the present invention;

[0064] Figure 2 This is a schematic diagram of the phase-locked loop model in a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 1 of the present invention;

[0065] Figure 3 This is a simulated non-ideal three-phase voltage waveform diagram of a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 1 of the present invention.

[0066] Figure 4 This is a schematic diagram of the SOGI dual second-order generalized integrator module in a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 1 of the present invention.

[0067] Figure 5 is a Bode plot of the SOGI dual second-order generalized integrator module in a three-phase power grid voltage adaptive filtering and phase-locking technology under a non-ideal power grid according to Embodiment 1 of the present invention.

[0068] Figure 6 This is a schematic diagram of the LMS least mean square root adaptive filtering module in a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 1 of the present invention.

[0069] Figure 7 This is a comparison diagram of the filtering effects of each module in a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 2 of the present invention;

[0070] Figure 8 This is a steady-state effect diagram of the phase-locked loop output in a phase-locked loop output of a phase-locked loop in a three-phase power grid voltage adaptive filtering and phase-locked loop technology under a non-ideal power grid according to Embodiment 2 of the present invention;

[0071] Figure 9 This is a diagram showing the transient effect of the phase-locked loop output in a non-ideal three-phase power grid voltage adaptive filtering and phase-locked loop technology according to Embodiment 2 of the present invention. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0073] Example 1

[0074] like Figure 1 As shown, this invention provides a design method for an LMS-SOGI three-phase phase-locked loop suitable for non-ideal power grids, comprising the following basic steps:

[0075] S1. Obtain the three-phase grid-connected voltage U of the grid-connected inverter through sampling. a (t), U b (t), U c (t), the grid-connected voltage α and β axis components U are obtained after Clark transformation. α (t), U β (t);

[0076] S2, then U α (t), U β (t) The data are fed into two SOGI structures respectively, and the positive sequence components of the grid voltage in the two-phase stationary coordinate system αβ coordinate system are extracted.

[0077] S3. Construct a mathematical model for the least mean square LMS filter stage and design the step size parameters for the least mean square LMS filter stage.

[0078] S4. Then analyze the positive sequence component of the grid voltage. LMS filtering is performed, and the Least Mean Square LMS filtering stage is equivalent to Park transform, yielding the positive sequence components of the fundamental frequency grid-connected voltage along the d and q axes.

[0079] S5. Obtain the q-axis component of the fundamental frequency grid-connected voltage. Subtracting from 0 yields the q-axis voltage phase-locked error signal e. q (t) is obtained by the angular frequency adjustment of the output signal ω through the PI controller, and then the phase θ is obtained by integration.

[0080] S6. Build a three-phase grid voltage LMS-SOGI phase-locked model in MATLAB / Simulink simulation software, and compare and analyze its filtering effect under different grid voltage conditions.

[0081] like Figure 2 As shown, in this embodiment, a phase-locked loop for adaptive filtering of three-phase grid voltage under non-ideal grid conditions is constructed, including: module ① grid-connected three-phase voltage module, module ② Clark transform module, module ③ SOGI second-order generalized integrator generator module, module ④ LMS least mean square root adaptive filtering module, module ⑤ PI control module, and module ⑥ simulation result observation module.

[0082] like Figure 3 As shown, in this embodiment, the grid-connected three-phase voltage simulated by module ① is divided into three stages: 0-2s and 4-6s are normal grid voltages: the amplitude of the three-phase voltages A, B, and C is 380V, and the phase difference is 120°; 2-4s is a three-phase unbalanced voltage with added harmonics, the unbalanced voltages are 380V for phase A, 450V for phase B, and 300V for phase B, and the added harmonics are the 2nd, 3rd, 5th, 7th, and 10th harmonics with an amplitude of 38V.

[0083] In this embodiment, the three-phase voltage sampling module specifically performs the following steps: in step S1, it samples and obtains the real-time three-phase grid-connected voltage U of the grid-connected inverter. a (t), U b (t), U c(t), whose fundamental voltage can be expressed in the three-phase stationary coordinate system abc as:

[0084]

[0085] Among them U a0 (t), U b0 (t), U c0 (t) represents the three-phase grid-connected fundamental voltage in the three-phase stationary coordinate system abc, and the three-phase balanced voltage amplitude U of the grid voltage. N =380V, unbalance coefficient k of phases B and C of the three-phase unbalanced voltage b =1.184, k c =0.789, fundamental angular frequency ω0 = 100π, assume the initial phase angle of phase a voltage is 0.

[0086] The Clark transformation in step S1, i.e., module ② Clark transformation module, adopts the following formula:

[0087]

[0088] U α (t), U β (t) represents the α and β components of the grid voltage in the two-phase stationary coordinate system αβ, T αβ The Clark transformation coefficient matrix is, i.e.

[0089] In this embodiment, the SOGI dual second-order generalized integrator transform in step S2 adopts the following formula:

[0090]

[0091] in The positive sequence components of the grid voltage are defined in the two-phase stationary coordinate system αβ. T represents the positive sequence component of the three-phase grid-connected fundamental voltage in the three-phase stationary coordinate system abc coordinate system. + The coefficient matrix of the symmetric component method is, i.e. in q refers to a 90° phase shift of the original signal in the time domain, i.e.

[0092] like Figure 4 As shown, in this embodiment, module ③, the SOGI second-order generalized integrator generator module, not only constructs two-phase quadrature voltage signals by shifting the input voltage signal by 90° to obtain the positive-sequence component of the grid voltage, but also filters out high-frequency interference signals. The transfer function of this SOGI system is:

[0093]

[0094]

[0095] As shown in Figure 5, in this embodiment, v' and v are the two outputs of the SOGI dual second-order generalized integrator module. The resonant frequency of the SOGI dual second-order generalized integrator module is ω' = 50Hz, and the reciprocal of the quality factor is k = 1..414.

[0096] like Figure 6 As shown, in this embodiment, module ④, the LMS (Least Mean Square) adaptive filtering module, includes: an input matrix X(n), a desired response input matrix d(n), an error vector matrix ε(n), a weight vector matrix W(n), a step size μ(n), and an output matrix y(n). Its basic mathematical model relationship is as follows:

[0097] y(n)=X T (n)*W(n)

[0098] ε(n)=d(n)-y(n)=d(n)-X T (n)*W(n)

[0099] In this embodiment, the weight iteration formula becomes:

[0100] W(n+1)=W(n)+μ(n)ε(n)X(n)

[0101] In this embodiment, to overcome the contradiction between dynamic response speed and steady-state accuracy, μ(n) adopts a variable step size approach, the formula of which is:

[0102] p(n)=βp(n-1)+(1-β)ε(n)ε(n-1)

[0103] μ(n+1)=αμ(n)+γp 2 (n)

[0104] In this embodiment, p(n) is the step-size iteration compensation matrix, and the step-size iteration coefficients are α = 0.9, β = 0.988, and γ = 0.03.

[0105] In this embodiment, the time-domain continuous mathematical model of phase-locked loop in this LMS least mean square root adaptive filtering module is as follows:

[0106] Input matrix

[0107] Weight vector matrix

[0108] Expected response input matrix

[0109] Where cosωt and sinωt are the phase angle trigonometric functions of the phase-locked loop output. The positive and negative sequence components of the grid voltage in the two-phase rotating coordinate system dq are given.

[0110] In this embodiment, the above mathematical model assigns the module output to the module's desired input, enabling the LMS least mean square root adaptive filtering module to adaptively train weights during self-tracking. Through error feedback, W(t) is adjusted to approximate the time-varying optimal weight matrix, thus obtaining the optimized parameter values ​​required in step S4.

[0111] In this embodiment, module ⑤, the PI control module, converts the positive-sequence q-axis component of the base frequency grid-connected voltage obtained in step S4. Subtracting U from 0 yields the q-axis voltage phase-locked error signal e. q , The q-axis voltage phase-locked error signal e q As the input signal to the PI controller, the output signal of the PI controller is the angular frequency adjustment ω, which is then integrated to obtain the phase θ.

[0112] In this embodiment, step S5 involves building a three-phase grid voltage LMS-SOGI phase-locked model in MATLAB / Simulink simulation software. Module ⑥, the simulation result observation module, uses Clark and Park transformations to analyze the positive sequence components of the grid voltage in the two-phase stationary coordinate system αβ coordinate system after SOGI filtering. With LMS adaptive filtering, the positive sequence components of the grid voltage in the two-phase rotating dq coordinate system The filtering effect was compared and analyzed by transforming the three-phase stationary coordinate system abc to the THD (Total Harmonic Distortion) comparative system.

[0113] like Figure 7 As shown, in this embodiment, sparse dashed lines represent the THD of the grid-connected voltage waveform, and dense dashed lines represent the positive sequence component of the grid voltage after SOGI filtering. The THD waveform after transformation to the three-phase stationary coordinate system abc is represented by the solid line, which indicates the positive sequence component of the grid voltage after LMS adaptive filtering. The THD waveform after transformation to the three-phase stationary coordinate system abc.

[0114] During the 0–2s and 4–6s periods under ideal grid voltage conditions, the steady-state values ​​of all three parameters are close to 0. Under the non-ideal grid conditions (2–4s), the steady-state values ​​are: THD of the grid-connected voltage waveform = 14.14%, THD of the voltage waveform after SOGI filtering = 4.97%, and THD of the voltage waveform after LMS filtering = 1.79%. This demonstrates that the LMS-SOGI phase-locked loop significantly improves the filtering effect compared to SOGI, achieving the expected goal.

[0115] like Figure 8 , Figure 9 As shown in this embodiment, the steady-state and transient effect diagrams of the phase-locked loop (PLL) output of the three-phase grid voltage adaptive filter under non-ideal grid conditions are presented. The solid line in the diagram represents the grid fundamental voltage, and the dashed line represents the PLL output angle. The steady-state effect diagram is taken from 2.6997s to 2.7002s during the 2s-4s non-ideal grid stage. The steady-state error is 0.10ms, approximately 0.5%. The transient effect diagram is taken from 1.95s to 2.35s, representing the transition from an ideal to a non-ideal grid voltage state. The transition point is 2s. After adopting a variable step size design, the dynamic response time of the PLL is approximately 0.2s.

[0116] In summary, this invention adds an adaptive filtering logic within the SOGI system to improve the voltage filtering effect of the three-phase power grid and the steady-state accuracy of the phase-locked loop. The SOGI phase-locked loop based on least mean square (LMS) adaptive filtering of this invention effectively avoids the influence of poor loop filter performance on the output frequency and phase of the phase-locked loop under non-ideal power grid conditions, thus improving the stability of the inverter. It can be widely used in various harsh operating conditions and applications requiring high detection speeds, and improves the locking accuracy for voltage frequency and phase.

[0117] This invention includes a three-phase grid voltage adaptive filtering and phase-locked loop (PLL) technology for non-ideal grids. It improves the internal structure of the double second-order generalized integrator SOGI, which is different from the improvement of adding a pre-filter and increasing the order. The LMS least mean square root adaptive filtering module replaces the Park transform module. It is not only equivalent to the function of the Park transform, but also achieves better filtering effect, enabling the subsequent PLL module to track the grid voltage more accurately.

[0118] In order to reduce the influence of negative sequence components and provide steady-state accuracy, the loop filter cutoff frequency must be set to a low value in the traditional technology, which leads to a slow dynamic response. In this invention, μ(n) adopts a variable step size method to overcome the contradiction between dynamic response speed and steady-state accuracy.

[0119] The LMS (Least Mean Square) adaptive filtering module in this invention does not employ a fixed-step-size or single-output model, but rather extends to a variable-step-size and multi-output form. Its high-speed convergence and low steady-state offset characteristics enable high-speed and high-steady-state accuracy detection of the amplitude and phase angle of the positive-sequence voltage components of the three-phase power grid. Therefore, this phase-locked system maintains good performance even in situations with severe harmonic distortion and three-phase imbalance, and can quickly detect abrupt changes in the amplitude and phase of the positive-sequence voltage components of the three-phase power grid, as well as the amplitude and phase angle of the positive-sequence voltage components of the three-phase power grid. This invention solves the technical problem in existing grid-connected inverter phase-locked technology that makes it difficult to accurately lock the voltage frequency and phase in specific scenarios.

[0120] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A design method for an LMS-SOGI three-phase phase-locked loop suitable for non-ideal power grids, characterized in that, The method includes: S1. Sample and obtain the three-phase grid-connected voltage of the grid-connected inverter. Based on this, Clark transformation operation is performed to obtain the grid-connected voltage. Axial components ; S2, the grid-connected voltage Axial components Two SOGI structures are fed into the system respectively, and two-phase stationary coordinate systems are extracted through SOGI dual second-order generalized integrator transformation operation. Positive sequence components of grid voltage in coordinate system ; S3. Construct a mathematical model for the least mean square (LMS) filter stage and design the step size parameters for the LMS filter stage. Step S3 further includes: S31. Using the preset LMS adaptive filtering module, the mathematical model relationship of the LMS filtering link is obtained. Based on the input matrix Expected response input matrix Error vector matrix Weight vector matrix The step size Output matrix Using the following logic, the mathematical model relationship of the Least Mean Square (LMS) filtering stage is obtained: ； S32. Using the Least Mean Square (LMS) adaptive filtering module, obtain a continuous mathematical model in the time domain, wherein the preset LMS adaptive filtering module includes: an input matrix. Expected response input matrix Error vector matrix Weight vector matrix Step length Output matrix ; S4. Using the mathematical model of the Least Mean Square (LMS) filter stage and according to the step size parameter of the LMS filter stage, the positive sequence component of the grid voltage is processed by LMS filtering. To obtain the d-axis positive sequence component of the base frequency grid-connected voltage. and the positive sequence q-axis component of the base frequency grid-connected voltage ; S5. Convert the q-axis component of the base frequency grid-connected voltage. Subtracting from 0 yields the q-axis voltage phase-locked error signal. The output signal angular frequency adjustment amount is obtained by using a PI controller. The phase is obtained by integration. ; S6. Build a three-phase power grid voltage LMS-SOGI phase-locked model in the pre-set simulation software, and adjust the output signal angular frequency according to the adjustment amount. and the phase We selected scenarios with different grid voltages and compared and analyzed the filtering effects.

2. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S1, based on the real-time voltage of the three-phase grid connection... The fundamental voltage in the three-phase stationary coordinate system abc is expressed using the following logic: In the formula, The three-phase grid-connected fundamental voltage is given by the three-phase stationary coordinate system abc coordinate system. The three-phase balanced voltage amplitude of the power grid. , This represents the unbalance coefficients for phases B and C of the three-phase unbalanced voltage. Fundamental angular frequency.

3. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S1, the Clark transformation operation is performed using the following logic: In the formula, Two-phase stationary coordinate system Grid voltage in coordinate system Quantity, This is the Clark transformation coefficient matrix.

4. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S2, the SOGI double second-order generalized integrator transformation operation is performed using the following logic: In the formula, Two-phase stationary coordinate system Positive sequence components of grid voltage in coordinate system The positive sequence component of the three-phase grid-connected fundamental voltage in the three-phase stationary coordinate system abc coordinate system. The coefficient matrix is ​​the symmetric component method, where q represents the phase shift of the original signal in the time domain. .

5. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S2, the transfer function of the SOGI structure is represented by the following logic: In the formula, These are the two outputs of the SOGI dual second-order generalized integrator module. The resonant frequency of the SOGI dual second-order generalized integrator module is given. It is the reciprocal of the quality factor.

6. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S32, the step size is obtained using the following variable step size method: In the formula, The step-size iterative compensation matrix is... The step size is the iteration coefficient.

7. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, According to the weight vector matrix The following logic is used to perform iterative weight transformation to obtain the iterative weights: 。 8. The LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids according to claim 1, characterized in that, In step S32, the following logic is used to obtain the time-domain continuous mathematical model of the corresponding phase-locked loop in the current LMS least mean square root adaptive filtering module, wherein the time-domain continuous mathematical model includes: Input matrix: ； Weight vector matrix: ； Expected response input matrix: ； In the formula, This is the phase angle trigonometric function output by the phase-locked loop. Two-phase rotating coordinate system Positive and negative sequence components of grid voltage in the coordinate system.

9. A three-phase phase-locked loop (LMS-SOGI) design system for non-ideal power grids, used to execute the LMS-SOGI three-phase phase-locked loop design method for non-ideal power grids as described in any one of claims 1 to 8, characterized in that, The system includes: The Clark converter module is used to sample and acquire the three-phase grid-connected voltage of the grid-connected inverter. Based on this, Clark transformation operation is performed to obtain the grid-connected voltage. Axial components ; SOGI dual second-order generalized integrator converter module, used to convert the grid-connected voltage Axial components Two SOGI structures are fed into the system respectively, and two-phase stationary coordinate systems are extracted through SOGI dual second-order generalized integrator transformation operation. Positive sequence components of grid voltage in coordinate system The SOGI dual second-order generalized integrator transform module is connected to the Clark transform module; A least mean square (LMS) filter model construction module is used to construct the mathematical model of the least mean square (LMS) filter stage and design the step size parameters of the least mean square (LMS) filter stage. The least mean square (LMS) filter model construction module also includes: The LMS filter model relationship acquisition module is used to process and obtain the mathematical model relationship of the LMS filter stage by utilizing the preset minimum mean square LMS filter model construction module. A time-domain continuous mathematical model acquisition module is used to acquire a time-domain continuous mathematical model using the Least Mean Square (LMS) filter model construction module. The preset LMS filter model construction module includes an input matrix. Expected response input matrix Error vector matrix Weight vector matrix Step length Output matrix ; The filtering module is used to perform LMS filtering on the positive-sequence component of the grid voltage using the mathematical model of the Least Mean Square (LMS) filter stage and according to the step size parameter of the LMS filter stage. To obtain the d-axis positive sequence component of the base frequency grid-connected voltage. and the positive sequence q-axis component of the base frequency grid-connected voltage The filtering operation module is connected to the Least Mean Square (LMS) filtering model construction module; The phase processing module is used to process the q-axis component of the fundamental frequency grid-connected voltage. Subtracting from 0 yields the q-axis voltage phase-locked error signal. The output signal angular frequency adjustment amount is obtained by using a PI controller. The phase is obtained by integration. The phase processing module is connected to the filtering operation module; The filtering effect comparison and analysis module is used to build a three-phase power grid voltage LMS-SOGI phase-locked model in the preset simulation software, and adjust the output signal angular frequency according to the adjustment amount. and the phase The filtering effect is compared and analyzed in scenarios with different grid voltages. The filtering effect comparison and analysis module is connected to the phase processing module.

Images