Unscented Kalman filter and grid-connected inverter model-free prediction control method and device

Abstract

The invention discloses an unscented Kalman filter and a grid-connected inverter model-free prediction control method and device, and the method comprises the steps: carrying out the state observation of a lumped disturbance quantity F alpha beta of a system through an improved unscented Kalman filter, and carrying out the parameter identification of a proportionality coefficient alpha in a super-local model through a weight-free adaptive algorithm; and the adaptive optimal value of the estimated identification value # imgabs0 # proportionality coefficient alpha is fed back to a model-free prediction control module, optimization is carried out in a weight factor-free cost function, and stable operation of the grid-connected inverter can be predicted without any model parameter when grid-connected current control is carried out to solve system parameter mismatch. According to the method, the adaptive Sigma method is introduced, a noise model and a weighted observer are enhanced, voltage fluctuation coupling, dynamic gamma optimization and a nonlinear dynamic model are introduced into a model parameter adaptive algorithm, the calculation complexity is kept relatively low, and meanwhile, the robustness of the model parameter adaptive algorithm is improved. And the estimation precision of the lumped disturbance quantity F alpha beta of the system and the identification degree of the proportionality coefficient alpha in the hyper-local model are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of grid-connected inverter control, and in particular relates to an unscented Kalman filter, a model-free predictive control method and a device for a grid-connected inverter. Background Art

[0002] With the rapid development of renewable energy, distributed generation systems such as solar and wind power are becoming increasingly popular and have become an important part of the modern power system. As the core device connecting the distributed generation system and the power grid, the control performance of the three-level grid-connected inverter directly affects the stability and efficiency of the entire system. The connection of a large number of power electronic devices will cause the increase of grid harmonics, and in more serious cases, it will lead to a decrease in power quality and the collapse of grid-connected equipment. Therefore, how to connect it to the grid efficiently and stably has become a key challenge for grid-connected inverters. Although the traditional grid-connected inverter control method can meet the performance requirements to a certain extent, it still has limitations in the dynamic response and robustness of the system.

[0003] Model-Free Predictive Control (MFPC) is an emerging control strategy that has received widespread attention in recent years because it does not rely on accurate system models and can predict the future behavior of the system in real time and perform optimal control. Currently, model-free predictive control in grid-connected inverters has problems such as delayed response time, insufficient control accuracy, sensitivity to interference, and poor adaptability under different operating conditions. These problems have led to reduced system stability and efficiency, thus affecting the reliability and economy of power generation.

[0004] The prior art proposes model-free predictive control based on the Runge-Kutta Algorithm (RKA) and the Lagrangian difference algorithm. This method has high computational complexity. At the same time, the model parameters of the fourth-order RKA are greatly affected by the time delay of the current response, and the prediction will be affected under poor operating conditions. There is also a model-free predictive control method based on a sliding mode observer, but this method is more complicated to design when the sliding surface and switching gain are reasonably selected. Rapid switching near the sliding surface may cause high-frequency oscillations in the system, and the interference sensitivity to random noise is poor. Summary of the invention

[0005] In order to solve the limitation problem of model-free predictive control in grid-connected inverters and improve the control performance of the system, the present invention proposes an unscented Kalman filter, a model-free predictive control method and a device for grid-connected inverters. The present invention improves the lumped disturbance F of the system while maintaining relatively low computational complexity. αβ Estimation accuracy and discrimination of the scaling factor α in the hyperlocal model.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] In a first aspect, the present invention provides an unscented Kalman filter, the unscented Kalman filter comprising:

[0008] Sigma point generation module is used to estimate the current state and capture the characteristics of state distribution; dynamic adjustment module is used to calculate new Sigma points based on real-time covariance and adjust the covariance Dynamic adjustment to incorporate historical errors into covariance updates;

[0009] A weighted calculation module, used to generate effective Sigma points that are conducive to covering the state space by dynamically adjusting the covariance;

[0010] The noise model module is used to obtain a noise model suitable for the actual state through the autoregressive model and the multiplicative noise theory modeling;

[0011] An enhancement module, used to adjust the noise covariance by enhancing the noise model, and update the process noise and observation noise;

[0012] Weighted observability tool, used to judge the observability of the system.

[0013] In one embodiment, the Sigma point generation formula is:

[0014]

[0015] In the formula,

[0016] The noise model is:

[0017]

[0018] In the formula, a i , b i are the model coefficients, p is the lag order; y(t) and u(t) are enhanced noise models;

[0019] The process noise update and observation noise update are:

[0020]

[0021] In the formula, φ and are the adjustment coefficients for process noise and observation noise, Q and R are the process noise covariance and measurement noise covariance, respectively; and are the state estimate and the measurement estimate respectively; X pred and Z pred are the initial values ​​of the state quantity and the measurement quantity respectively.

[0022] In one embodiment, the matrix O of the weighted observability w for:

[0023]

[0024] In the formula, w i is the weighting coefficient;

[0025] Through singular value decomposition calculation, O can be quickly determined w The rank of is:

[0026] O w =UΣV T

[0027] In the formula, O w is the weighted observable matrix, U and V are the left and right singular vectors in the singular value decomposition method respectively;

[0028] Count the number of non-zero singular values ​​of Σ to determine O w rank.

[0029] In a second aspect, the present invention provides a weight-free adaptive model-free predictive control method for a grid-connected inverter, wherein the method converts the system's lumped disturbance F αβ The state observation is performed by an unscented Kalman filter, and the parameter identification of the proportional coefficient α in the hyperlocal model is performed by an unweighted adaptive algorithm; the method comprises:

[0030] Obtain the three-phase voltage and current on the output side of the grid-connected inverter, and establish a mathematical model of the three-level grid-connected inverter based on Kirchhoff's law;

[0031] Clark transformation is performed on the three-phase voltage and current to obtain the mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system, and the three-phase grid voltage is passed through a phase-locked loop to obtain the current reference value;

[0032] Substituting the grid-side current and voltage at time k as initial values ​​into the unscented Kalman filter described in any one of claims 1 to 3 to obtain an estimated identification value;

[0033] According to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value, a weightless adaptive optimal value of the proportional coefficient α is obtained through a weightless adaptive algorithm;

[0034] The output current of the grid-connected inverter at time k+1 is obtained by model-free prediction of the estimated identification value, the unweighted adaptive optimal value of the proportional coefficient α and the grid voltage in the two stationary coordinate systems;

[0035] The output current and current reference value of the grid-connected inverter at time k+1 are substituted into the unweighted cost function to find the optimal value, and the optimal voltage vector is selected to control the switching state of the grid-connected inverter at the next moment, so as to realize model-free predictive control of the grid-connected inverter under the condition of system parameter mismatch.

[0036] In one implementation, the estimated identification value is:

[0037] In the formula, i αβ (k), i αβ (k+1) are the output currents of the grid-connected inverter at time k and k+1 respectively, are the estimated identification values ​​of the grid-connected inverter at time k and k+1 respectively; A is the state transfer matrix, T is the sampling period, ω is the grid angular frequency; Z(k) is the observation equation; B is the noise driving matrix, α is the proportionality coefficient, usually set to α = 1 / L; H is the measurement matrix, H =

[10] .

[0038] In one embodiment, the method of obtaining the unweighted adaptive optimal value of the proportional coefficient α by an unweighted adaptive algorithm based on the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value includes:

[0039] Define a control error and select the target signal to be tracked as the reference current i in the stationary coordinate system αβref (t), the error state equation is defined as:

[0040] e(t)=i αβref (t)-i αβ (t)

[0041] According to the above formula, the hyperlocal model Substituting it in, we get the error dynamic equation:

[0042]

[0043] The error dynamic equation is used as the basis of the adaptive law, and the adaptive law is established based on the gradient descent method. Its expression is:

[0044]

[0045] Where γ is the learning rate, which needs to be adjusted properly. Too large a value may lead to instability, while too small a value may result in slow convergence. k is the Δu balance gain coefficient.

[0046] After the proportional coefficient α is discretized, a nonlinear dynamic model is introduced, and the control parameters are dynamically adjusted in the αβ axis direction through a nonlinear feedback mechanism. The specific adjustment is:

[0047] α α(k+1)=α α (k)-Asgn(e β (k))γ|e β (k)| λ u α (k)T+ksgn(Δu)T

[0048] α β (k+1)=α β (k)-Bsgn(e α (k))γ|e α (k)| λ u β (k)T+ksgn(Δu)T

[0049] Where sgn(e(k)) represents the sign of the error to ensure that the update direction is consistent with the error; |e(k)| λ is the power of the absolute value of the error (usually λ>0); A and B are the error coupling coefficients respectively;

[0050] The Lyapunov function is designed as a polynomial form, combining the changes in error and control parameters, specifically:

[0051]

[0052] In the formula, the initial value α0 is set to 50; Improve the capture of transient dynamic behavior, k is a positive weight coefficient used to adjust the error state The impact of is the derivative of the voltage fluctuation, further providing information about the midpoint potential fluctuation;

[0053] According to Lyapunov's stability principle, the error state e(t), and In the adjustment process, the convergence is uniform to zero, and the control parameter α fluctuates around α0. The system remains stable, ensuring that the error state and parameter α converge to the optimal value α opt .

[0054] In one embodiment, the learning rate γ is dynamically adjusted according to the deviation of the target performance indicator e(t), and its dynamic adjustment strategy is: if the current error e k Less than the error e at the previous moment k-1 , then increase the learning rate: γ=min(1.05γ,γ max ); if the current error increases, reduce the learning rate: γ=max(0.95γ,γ min ).

[0055] In one embodiment, the identification value F αβ, unweighted adaptive optimal value α opt and u αβ After discretization by the model-free prediction module, we get i αβ , whose expression is:

[0056]

[0057] In one implementation, the unweighted cost function is:

[0058] g=(i αref -i α (k+1)) 2 +(i βref -i β (k+1)) 2

[0059] In the formula, i αβref is the reference current in the stationary coordinate system.

[0060] In a third aspect, the present invention provides a weight-free adaptive grid-connected inverter model-free predictive control device, the device comprising:

[0061] A mathematical model module is used to obtain the three-phase voltage and current at the output side of the grid-connected inverter and establish a mathematical model of the three-level grid-connected inverter according to Kirchhoff's law;

[0062] A coordinate system transformation module is used to perform Clark transformation on the three-phase voltage and current to obtain the mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system;

[0063] A current reference value module is used to obtain a current reference value by passing the three-phase grid voltage through a phase-locked loop;

[0064] The above-mentioned unscented Kalman filter is used to obtain an estimated identification value based on the grid-side current and voltage of the grid-connected inverter at time k;

[0065] A parameter adaptive calculation module is used to obtain a weightless adaptive optimal value of a proportional coefficient α through a weightless adaptive algorithm according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value;

[0066] A model-free prediction module is used to obtain the output current of the grid-connected inverter at time k+1 by model-free prediction of the estimated identification value, the weight-free adaptive optimal value of the proportional coefficient α and the grid voltage in the two stationary coordinate systems;

[0067] The optimal switching state selection module is used to substitute the output current and current reference value of the grid inverter at time k+1 into the unweighted cost function to find the optimal value, select the optimal voltage vector to control the switching state of the grid-connected inverter at the next moment, and realize model-free predictive control of the grid-connected inverter under the condition of system parameter mismatch.

[0068] The beneficial effects of the solution proposed by the present invention are as follows:

[0069] The present invention proposes a model-free predictive control strategy for grid-connected inverters based on a weight-free adaptive unscented Kalman filter. The strategy mainly converts the system's lumped disturbance F αβ The state observation is carried out by using an improved unscented Kalman filter. The adaptive Sigma method, enhanced noise model and weighted observable are introduced to keep the computational complexity relatively low while providing high-precision state estimation results in nonlinear models. The unweighted adaptive optimal value α is obtained by introducing voltage fluctuation coupling, dynamic γ optimization and nonlinear dynamic model in the model parameter adaptive algorithm. opt , the weight influence of the DC side capacitor voltage fluctuation in the cost function is optimized, so that the predicted current has good performance in a wider amplitude range; under the condition of system parameter mismatch, it not only greatly reduces the calculation cost to ensure the stable operation of the system, but also has better current tracking performance and lower harmonic distortion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings are part of the present invention and are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but do not constitute an improper limitation of the present invention. Obviously, the accompanying drawings described below are only some embodiments. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0071] Figure 1 A schematic diagram of the structure of an unscented Kalman filter provided by an embodiment of the present invention;

[0072] Figure 2 A flow chart of a non-weighted adaptive model-free predictive control method for a grid-connected inverter provided by an embodiment of the present invention;

[0073] Figure 3 It is a three-level grid-connected inverter topology;

[0074] Figure 4 is the unweighted adaptive optimal value α provided in the embodiment of the present invention opt Waveform diagram of

[0075] Figure 5: is the three-phase grid-connected current waveform of the traditional MPC in steady state under the inductance parameter mismatch condition provided in the embodiment of the present invention, wherein Figure 5 (a) is the simulated waveform of the three-phase grid-connected current in the traditional MPC steady state under the condition of inductance parameter mismatch. Figure 5 (b) is the simulated waveform of the grid-connected current error of the traditional MPC in steady state under the condition of inductance parameter mismatch. Figure 5 (c) is a schematic diagram of the three-phase grid-connected current THD in the traditional MPC steady state under the condition of inductance parameter mismatch;

[0076] Figure 6 It is the current simulation waveform of the model-free predictive control method of the sliding mode observer in the prior art, wherein: Figure 6 (a) is the simulated waveform of the three-phase grid-connected current in steady state under the condition of inductance parameter mismatch. Figure 6 (b) is the simulated waveform of grid-connected current error in steady state under the condition of inductor parameter mismatch. Figure 6 (c) is a schematic diagram of the three-phase grid-connected current THD in steady state under the condition of inductance parameter mismatch;

[0077] Figure 7 is a current simulation waveform using the method of the present invention, wherein Figure 7 (a) is the simulated waveform of the three-phase grid-connected current in steady state using the method of the present invention under the condition of inductance parameter mismatch, Figure 7 (b) is the grid-connected current error simulation waveform of the control strategy of the present invention in steady state under the condition of inductance parameter mismatch, Figure 7 (c) is a schematic diagram of the three-phase grid-connected current THD in steady state based on the control strategy of the present invention under the condition of inductance parameter mismatch;

[0078] Figure 8 This is a comparison diagram of the grid-connected current simulation of the method of the present invention and the prior art, where Figure 8 (a) is a simulation comparison diagram of the traditional MPC and the method of the present invention when the grid-connected current is 10A. Figure 8 (b) is a simulation comparison diagram of the model-free predictive control method of the sliding mode observer and the method of the present invention when the grid-connected current is 10A;

[0079] Fig. 9 is the dynamic performance simulation waveform under the condition of inductor parameter mismatch, where: Fig. 9 (a) is the dynamic performance simulation waveform of the model-free predictive control method of the sliding mode observer under the condition of inductor parameter mismatch. Fig. 9 (b) is a dynamic performance simulation waveform of the method of the present invention under the condition of inductance parameter mismatch;

[0080] Fig.10 A block diagram of a model-free predictive control system for a grid-connected inverter provided by an embodiment of the present invention;

[0081] It should be noted that these drawings and textual descriptions are not intended to limit the conceptual scope of the present invention in any way, but are intended to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0082] The model-free predictive control method for the grid-connected inverter of the present invention will be further described in detail below in conjunction with the accompanying drawings. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other, and the technical solution of the present application will be described in detail with reference to the accompanying drawings and in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0083] like Figure 1 As shown, an embodiment of the present invention shows an unscented Kalman filter, the unscented Kalman filter comprising:

[0084] Adaptive Sigma point generation module, used to estimate the current state and capture the characteristics of state distribution;

[0085] The prediction module is used to calculate the predicted values ​​of the system state quantity and the observed quantity according to the generated Sigma points;

[0086] Dynamic adjustment module, used to dynamically adjust the mean and covariance according to the predicted value, and incorporate historical errors into the covariance update;

[0087] The enhanced noise model module is used to obtain a noise model suitable for the actual state through the autoregressive model and multiplicative noise theory modeling, and then use the enhanced noise model to adjust the noise covariance and update the process noise and observation noise;

[0088] Weighted observable, used to determine the observability of the system.

[0089] In the standard unscented Kalman filter, Sigma points are generated around the current state estimate to capture the characteristics of the state distribution. The Sigma point generation formula is:

[0090]

[0091] In order to dynamically adjust the position of the Sigma point, a new Sigma point can be calculated based on the real-time covariance.

[0092] Covariance Dynamic adjustment, incorporating historical errors into covariance updates, using a weighted average approach:

[0093]

[0094] Where σ is the adjustment factor. When the system changes dynamically, the covariance can be increased to generate more dispersed Sigma points; conversely, the covariance can be reduced to focus on more accurate estimates. The adaptive Sigma point method uses dynamic adjustment of the covariance to generate more effective Sigma points, ensuring that they can better cover the state space.

[0095] Gaussian white noise is usually independent and identically distributed in time, and the process noise w(t) and observation noise follow a normal distribution:

[0096] w(t)~N(0,Q)

[0097] v(t)~N(0,R)

[0098] In practical applications, the noise in the system is often time-dependent. Through the autoregressive model and multiplicative noise theory modeling, a noise model suitable for the actual state can be obtained:

[0099]

[0100] In the formula, a i , b i are the model coefficients, p is the lag order; y(t) and u(t) are enhanced noise models.

[0101] By enhancing the noise model to adjust the noise covariance, the process noise update and observation noise update are:

[0102]

[0103] In the formula, φ and are the adjustment coefficients for process noise and observation noise, Q and R are the process noise covariance and measurement noise covariance, respectively; and are the state estimate and the measurement estimate respectively; X pred and Z pred are the initial values ​​of the state quantity and the measurement quantity respectively.

[0104] In the case of dynamic changes in the system, the conventional observability matrix of the unscented Kalman filter observability may not effectively reflect the observability of the system. To this end, the weighted observability matrix O is introduced. w for:

[0105]

[0106] In the formula, w i is the weighting coefficient.

[0107] And through singular value decomposition calculation, O can be quickly determinedw The rank of is:

[0108] O w =UΣV T

[0109] In the formula, O w is the weighted observable matrix, U and V are the left and right singular vectors in the singular value decomposition method respectively;

[0110] Count the number of non-zero singular values ​​of Σ to determine O w rank.

[0111] like Figure 2 As shown, an embodiment of the present invention shows a non-weighted adaptive model-free predictive control method for a grid-connected inverter, which converts the system's lumped disturbance F αβ The state observation is performed by using an unscented Kalman filter, and the parameter identification of the proportional coefficient α in the hyperlocal model is performed by using an unweighted adaptive algorithm; the method includes:

[0112] Step S100: acquiring the three-phase voltage and current at the output side of the grid-connected inverter, and establishing a mathematical model of the three-level grid-connected inverter according to Kirchhoff's law.

[0113] Step S200: Clark transformation is performed on the three-phase voltage and current to obtain a mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system, and the three-phase grid voltage is passed through a phase-locked loop to obtain a current reference value.

[0114] Combination Figure 3 ,In the three-level grid-connected inverter topology, according to Kirchhoff’s law, a mathematical model of the grid-connected inverter is established.

[0115] The three-phase current i on the output side of the grid-connected inverter is obtained by sampling and calculation a 、i b 、i c , and then perform Clark transformation to get i αβ (k);

[0116] The expressions are as follows:

[0117]

[0118] In the formula, i a 、i b 、i c is the grid-connected inverter output current, u a 、u b 、u c is the inverter output voltage, L is the filter inductor, R is the parasitic resistance, e a 、e b、e c is the grid voltage.

[0119] The Clark transformation formula is:

[0120]

[0121] The three-phase voltage and current are transformed by Clark transformation, and the mathematical model of the three-level grid-connected inverter is expressed in the αβ-term stationary coordinate system as follows:

[0122]

[0123] In the formula, i αβ =[i α (k),i β (k]) T ,i α (k) and i β (k) is the output current of the grid-connected inverter at time (k); u αβ =[u α (k),u β (k)] T ,u α (k) and u β (k) is the output voltage of the grid-connected inverter at time (k); e αβ =[e α (k),e β (k)] T ,e α (k) and e β (k) is the grid voltage at time (k).

[0124] The three-phase grid voltage is passed through the phase-locked loop to obtain the reference angle θ and the given current value, and the current reference value i is obtained through coordinate transformation. αβref .

[0125] Step S300: Substituting the grid-side current and voltage at time k as initial values ​​into the unscented Kalman filter to obtain an estimated identification value.

[0126] Let i at time (k) αβ (k),u αβ (k) Substitute into the unscented Kalman filter algorithm to update the estimated identification value Its expression is:

[0127]

[0128] In the formula, i αβ (k), i αβ (k+1) are the output currents of the grid-connected inverter at time k and k+1 respectively, are the estimated identification values ​​of the grid-connected inverter at time k and k+1 respectively; A is the state transfer matrix, T is the sampling period, ω is the grid angular frequency; Z(k) is the observation equation; B is the noise driving matrix, α is the proportionality coefficient, usually set to α = 1 / L; H is the measurement matrix, H =

[10] .

[0129] Step S400: according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value, a weightless adaptive optimal value of the proportional coefficient α is obtained by a weightless adaptive algorithm.

[0130] α is the proportionality factor when designing the hyperlocal model, usually set to α = 1 / L, but it will be affected by i αβ 、F αβ 、u αβ At the same time, if the parameter α is too large or too small, it will affect the performance of the model-free predictive control of the grid-connected inverter. For this reason, the present invention designs an adaptive controller based on error feedback and voltage fluctuation coupling to determine the optimal value α of the proportional coefficient α in the hyperlocal model. opt .

[0131] In the embodiment of the present application, according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value, the weightless adaptive optimal value of the proportional coefficient α is obtained by a weightless adaptive algorithm, and the specific steps include:

[0132] Define a control error and select the target signal you want to track as i αβref (t), then the error state equation can be defined as:

[0133] e(t)=i αβref (t)-i αβ (t)

[0134] According to the above formula, the hyperlocal model Substituting it in, we get the error dynamic equation:

[0135]

[0136] The error dynamic equation can be used as the basis for designing the adaptive law. In order to dynamically adjust the parameter α to minimize the error dynamic equation, it is necessary to realize the optimization of the unweighted factor in the cost function. The adaptive law can be established based on the gradient descent method, and its expression is:

[0137]

[0138] Where γ is the learning rate, which needs to be adjusted properly. Too large a value may lead to instability, while too small a value may result in slow convergence. k is the Δu balance gain coefficient.

[0139] Among them, the learning rate γ can be dynamically adjusted according to the deviation of the target performance indicator e(t). Its dynamic adjustment strategy is: if the current error e k Less than the error e at the previous moment k-1 , then increase the learning rate: γ=min(1.05γ,γ max ); if the current error increases, reduce the learning rate: γ=max(0.95γ,γ min ).

[0140] After the proportional coefficient α is discretized, a nonlinear dynamic model is introduced, and the control parameters are dynamically adjusted in the αβ axis direction through a nonlinear feedback mechanism, which can be adjusted as follows:

[0141] α α (k+1)=α α (k)-Asgn(e β (k))γ|e β (k)| λ u α (k)T+ksgn(Δu)T

[0142] α β (k+1)=α β (k)-Bsgn(e α (k))γ|e α (k)| λ u β (k)T+ksgn(Δu)T

[0143] Where sgn(e(k)) represents the sign of the error to ensure that the update direction is consistent with the error; |e(k)| λ It is the power of the absolute value of the error (usually λ>0); A and B are the error coupling coefficients respectively.

[0144] In order to ensure the stability of the system, the Lyapunov function is designed as a polynomial form, combining the changes of errors and control parameters:

[0145]

[0146] In the formula, the initial value α0 is set to 50; Improve the capture of transient dynamic behavior, k is a positive weight coefficient used to adjust The impact of is the derivative of the voltage fluctuation, providing further information about the midpoint potential fluctuation.

[0147] According to Lyapunov's stability principle, the error state e(t), Δu and In the adjustment process, the convergence is uniform to zero, and the control parameter α fluctuates around α0. The system remains stable, ensuring that the error state and parameter α converge to the optimal value α opt .

[0148] Figure 4 is the unweighted adaptive optimal value α obtained by an adaptive controller based on error feedback and voltage fluctuation coupling. opt waveform.

[0149] Step S500: The estimated identification value, the weightless adaptive optimal value of the proportionality coefficient α and the two grid voltages u in the stationary coordinate system are calculated. αβ The output current of the grid-connected inverter at time k+1 obtained by model-free prediction.

[0150] The estimated identification value, the unweighted adaptive optimal value of the proportional coefficient α and the grid voltage u αβ The output current of the grid-connected inverter at time k+1 obtained by model-free prediction is:

[0151]

[0152] Step S600: Substitute the output current and current reference value of the grid inverter at time k+1 into the unweighted cost function to find the optimal value, select the optimal voltage vector to control the switching state of the grid-connected inverter at the next moment, and realize model-free predictive control of the grid-connected inverter under the system parameter mismatch condition.

[0153] The predicted current i of the grid-connected inverter at time (k+1) αβ (k+1) and the grid voltage through the phase-locked loop to obtain the current reference value i αβref Substitute them into the cost function without weight factors to find the best solution and select the optimal voltage vector to control the switching state of the grid-connected inverter at the next moment. The cost function without weight factors is:

[0154] g=(i αref -i α (k+1)) 2 +(i βref -i β (k+1)) 2

[0155] In the formula, i αβref is the reference current in the stationary coordinate system.

[0156] In order to verify the effectiveness of the present invention, simulation was carried out in MATLAB / Simulink environment, and the system parameters are shown in the following table.

[0157] Table system parameters

[0158]

[0159] In order to verify the feasibility and parameter robustness of the proposed control strategy, the present invention studies its steady-state performance in simulation and compares and analyzes it with the traditional MPC and the control strategy proposed in Reference 2. v =0.1H, the reference value of the grid-connected inverter output current is 10A, and the grid current base frequency is 50Hz. Figure 5 , Figure 6 and Figure 7 The three-phase grid-connected current waveform, current error and THD schematic diagrams of the traditional MPC, the control strategy proposed in document 2 and the control strategy described in the present invention in steady state are respectively shown.

[0160] from Figure 5 , Figure 6 and Figure 7 It can be seen that when the reference current is 10A, the grid-connected current obtained by the traditional MPC is relatively chaotic, the THD is 7.10%, and the current error value is large, fluctuating at ±3A; the grid-connected current THD of the control strategy proposed in Document 2 is 3.72%, and the current error value fluctuates at ±2A; the control strategy described in the present invention has significantly reduced the grid-connected current THD and error value compared to the control strategy proposed in Document 2. The THD of this method is 1.15%, and the current error value fluctuates at ±1A. It can be seen that this verifies the feasibility of the control strategy described in the present invention in steady state, and avoids the influence of model parameter mismatch on the grid-connected inverter system.

[0161] Under the same simulation conditions, the steady-state control performance of the traditional MPC and the control strategy described in the present invention is compared. Figure 8 As shown in a, the steady-state control performance comparison between the control strategy proposed in document 2 and the control strategy described in the present invention is as follows: Figure 8 b. When t is 0-100ms, the control system adopts the traditional MPC and the control strategy proposed in document 2 respectively; starting from t being 100ms, the control system switches to the control strategy of the present invention.

[0162] When the model parameters are mismatched, Figure 8 It can be seen from a that the grid-connected current waveform of the traditional MPC strategy has collapsed, and the THD value has reached 10.17%, while the current THD value under the control strategy of the present invention is only 1.13%; Figure 8 b It can be seen that the current harmonic content THD values ​​of the control strategy proposed in document 2 and the control strategy described in the present invention are 3.73% and 1.12% respectively, but the control strategy described in the present invention has better steady-state performance, which further verifies its effectiveness.

[0163] In order to verify the dynamic performance of the proposed control strategy, a comparative analysis is performed through dynamic simulation of the control strategy of the present invention and the control strategy proposed in Reference 2. Fig. 9 It can be seen that when t is the 100th ms, the reference current suddenly changes from 10A to 15A, the dynamic response time of the KF-based MFPC strategy is 1.44ms, and the dynamic response time of the control strategy described in the present invention is 0.83ms, which indicates that the control strategy described in the present invention has better dynamic response performance.

[0164] The following is an embodiment of a non-weighted adaptive grid-connected inverter model-free predictive control device of the present invention, which can be used to execute an embodiment of a non-weighted adaptive grid-connected inverter model-free predictive control method of the present invention. For details not disclosed in the embodiment of the non-weighted adaptive grid-connected inverter model-free predictive control device of the present invention, please refer to the embodiment of the non-weighted adaptive grid-connected inverter model-free predictive control method of the present invention.

[0165] Reference Fig.10 As shown, in one embodiment, a weight-free adaptive grid-connected inverter model-free predictive control device is provided, and the device includes:

[0166] A mathematical model module is used to obtain the three-phase voltage and current at the output side of the grid-connected inverter and establish a mathematical model of the three-level grid-connected inverter according to Kirchhoff's law;

[0167] A coordinate system transformation module is used to perform Clark transformation on the three-phase voltage and current to obtain the mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system;

[0168] A current reference value module is used to obtain a current reference value by passing the three-phase grid voltage through a phase-locked loop;

[0169] The unscented Kalman filter according to any one of claims 1 to 3 is used to obtain an estimated identification value based on the grid-side current and voltage of the grid-connected inverter at time k;

[0170] A parameter adaptive calculation module is used to obtain a weightless adaptive optimal value of a proportional coefficient α through a weightless adaptive algorithm according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value;

[0171] The model-free prediction module is used to estimate the identification value, the weight-free adaptive optimal value of the proportional coefficient α, and the grid voltage u in the stationary coordinate system. αβ The output current of the grid-connected inverter at time k+1 obtained by model-free prediction;

[0172] The optimal switching state selection module is used to substitute the output current and current reference value of the grid inverter at time k+1 into the unweighted cost function to find the optimal value, select the optimal voltage vector to control the switching state of the grid-connected inverter at the next moment, and realize model-free predictive control of the grid-connected inverter under the condition of system parameter mismatch.

[0173] The various functional modules in the embodiments of the present invention may be integrated into one processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0174] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this patent can make some changes or modify the technical contents suggested above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the invention without departing from the content of the technical solution of the present invention still fall within the scope of the solution of the present invention.

Claims

1. An unscented Kalman filter, characterized in that The unscented Kalman filter comprises: Adaptive Sigma point generation module, used to estimate the current state and capture the characteristics of state distribution; The prediction module is used to calculate the predicted values ​​of the system state quantity and the observed quantity according to the generated Sigma points; Dynamic adjustment module, used to dynamically adjust the mean and covariance according to the predicted value, and incorporate historical errors into the covariance update; The enhanced noise model module is used to obtain a noise model suitable for the actual state through the autoregressive model and multiplicative noise theory modeling, and then use the enhanced noise model to adjust the noise covariance and update the process noise and observation noise; Weighted observable, used to determine the observability of the system.

2. An unscented Kalman filter according to claim 1, characterized in that: The Sigma point generation formula is: In the formula, i = 1 ~ n, L = n + 1 ~ 2n, n is the dimension of the state quantity; The noise model is: In the formula, a i 、b i are the model coefficients, p is the lag order; y(t) and u(t) are enhanced noise models; The updated process noise and observation noise are: In the formula, φ and are the adjustment coefficients for process noise and observation noise, Q and R are the process noise covariance and measurement noise covariance, respectively; and are the state estimate and the measurement estimate respectively; X pred and Z pred are the initial values ​​of the state quantity and the measurement quantity respectively.

3. An unscented Kalman filter according to claim 1, characterized in that: Weighted Observable O w for: In the formula, w i is the weighting coefficient; Through singular value decomposition calculation, O can be quickly determined w The rank of is: O w =UΣV T In the formula, O w is the weighted observable matrix, U and V are the left and right singular vectors in the singular value decomposition method respectively; Count the number of non-zero singular values ​​of Σ to determine O w rank.

4. A weight-free adaptive model-free predictive control method for a grid-connected inverter, characterized in that: The method converts the total disturbance F αβ The state observation is performed by an unscented Kalman filter, and the parameter identification of the proportional coefficient α in the hyperlocal model is performed by an unweighted adaptive algorithm; the method comprises: Obtain the three-phase voltage and current on the output side of the grid-connected inverter, and establish a mathematical model of the three-level grid-connected inverter based on Kirchhoff's law; Clark transformation is performed on the three-phase voltage and current to obtain the mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system, and the three-phase grid voltage is passed through a phase-locked loop to obtain the current reference value; Substituting the grid-side current and voltage at time k as initial values ​​into the unscented Kalman filter described in any one of claims 1 to 3 to obtain an estimated identification value; According to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value, a weightless adaptive optimal value of the proportional coefficient α is obtained through a weightless adaptive algorithm; The output current of the grid-connected inverter at time k+1 is obtained by model-free prediction of the estimated identification value, the unweighted adaptive optimal value of the proportional coefficient α and the grid voltage in the two stationary coordinate systems; The output current and current reference value of the grid-connected inverter at time k+1 are substituted into the unweighted cost function to find the optimal value, and the optimal voltage vector is selected to control the switching state of the grid-connected inverter at the next moment, so as to realize model-free predictive control of the grid-connected inverter under the condition of system parameter mismatch.

5. The method for non-weighted adaptive model-free predictive control of a grid-connected inverter according to claim 1, characterized in that: The estimated identification value is: In the formula, i αβ (k), i αβ (k+1) are the output currents of the grid-connected inverter at time k and k+1 respectively, are the estimated identification values ​​of the grid-connected inverter at time k and k+1 respectively; A is the state transfer matrix, T is the sampling period, ω is the grid angular frequency; Z(k) is the observation equation; B is the noise driving matrix, α is the proportionality coefficient, usually set to α = 1 / L; H is the observation matrix, H = [1 0].

6. The weight-free adaptive grid-connected inverter model-free predictive control method according to claim 1, characterized in that: The method of obtaining the unweighted adaptive optimal value of the proportional coefficient α by an unweighted adaptive algorithm according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value includes: Define a control error and select the target signal to be tracked as the reference current i in the stationary coordinate system αβref (t), the error state equation is defined as: e(t)=i αβref (t)-i αβ (t) According to the above formula, the hyperlocal model Substituting it in, we get the error dynamic equation: The error dynamic equation is used as the basis of the adaptive law, and the adaptive law is established based on the gradient descent method. Its expression is: Where γ is the learning rate, which needs to be adjusted properly. Too large a value may lead to instability, while too small a value may result in slow convergence. k is the Δu balance gain coefficient. After the proportional coefficient α is discretized, a nonlinear dynamic model is introduced, and the control parameters are dynamically adjusted in the αβ axis direction through a nonlinear feedback mechanism. The specific adjustment is: a α (k+1)=a α (k)-Asgn(e β (k))γ|e β (k)| λ you α (k)T+ksgn(Δu)T α β (k+1)=α β (k)-Bsgn(e α (k))γ|e α (k)| λ u β (k)T+ksgn(Δu)T Where sgn(e(k)) represents the sign of the error to ensure that the update direction is consistent with the error; |e(k)| λ is the power of the absolute value of the error (usually λ>0); A and B are the error coupling coefficients respectively; The Lyapunov function is designed as a polynomial form, combining the changes in error and control parameters, specifically: In the formula, the initial value α0 is set to 50; Improve the capture of transient dynamic behavior, k is a positive weight coefficient used to adjust the error state The impact of is the derivative of the voltage fluctuation, which further provides information about the midpoint potential fluctuation; According to Lyapunov's stability principle, the error state e(t), and In the adjustment process, the convergence is consistent to zero, and the control parameter α fluctuates around α0. The system remains stable, ensuring that the error state and parameter α converge to the optimal value α opt .

7. The weight-free adaptive model-free predictive control method for grid-connected inverter according to claim 6, characterized in that: The learning rate γ is dynamically adjusted according to the deviation of the target performance indicator e(t), and its dynamic adjustment strategy is: if the current error e k Less than the error e at the previous moment k-1 , then increase the learning rate: γ=min(1.05γ,γ max ); if the current error increases, reduce the learning rate: γ=max(0.95γ,γ min ).

8. The weight-free adaptive model-free predictive control method for grid-connected inverter according to claim 7, characterized in that: The estimated identification value, the weightless adaptive optimal value of the proportional coefficient α and the grid voltage u αβ The output current of the grid-connected inverter at time k+1 obtained by model-free prediction is:

9. The non-weighted adaptive model-free predictive control method for grid-connected inverters according to claim 1, characterized in that: The unweighted cost function is: g=(i αref -i α (k+1)) 2 +(i βref -i β (k+1)) 2 In the formula, i αref 、i βref is the reference current in the stationary coordinate system, i α (k+1), i β (k+1) is the output current of the grid-connected inverter at time k+1.

10. A non-weighted adaptive model-free predictive control device for a grid-connected inverter, characterized in that: The device comprises: A mathematical model module is used to obtain the three-phase voltage and current at the output side of the grid-connected inverter and establish a mathematical model of the three-level grid-connected inverter according to Kirchhoff's law; A coordinate system transformation module is used to perform Clark transformation on the three-phase voltage and current to obtain the mathematical model of the three-level grid-connected inverter in the αβ-term stationary coordinate system; A current reference value module is used to obtain a current reference value by passing the three-phase grid voltage through a phase-locked loop; The unscented Kalman filter according to any one of claims 1 to 3 is used to obtain an estimated identification value based on the grid-side current and voltage of the grid-connected inverter at time k; A parameter adaptive calculation module is used to obtain a weightless adaptive optimal value of a proportional coefficient α through a weightless adaptive algorithm according to the obtained estimated identification value, the three-phase voltage and current at the output side of the grid-connected inverter, and the current reference value; A model-free prediction module is used to obtain the output current of the grid-connected inverter at time k+1 by model-free prediction of the estimated identification value, the weight-free adaptive optimal value of the proportional coefficient α and the grid voltage in the two stationary coordinate systems; The optimal switching state selection module is used to substitute the output current and current reference value of the grid inverter at time k+1 into the unweighted cost function to find the optimal value, select the optimal voltage vector to control the switching state of the grid-connected inverter at the next moment, and realize model-free predictive control of the grid-connected inverter under the condition of system parameter mismatch.

Images