A model-free predictive control method for electrolytic capacitor-less Vienna rectifier

Abstract

The application discloses a kind of electrolytic capacitor-free Vienna rectifier model-free predictive control methods, belong to rectifier control field, the electrolytic capacitor-free Vienna rectifier model-free predictive control method described in the application includes the following steps: step S1, constructs the hyperlocal model of Vienna rectifier, and based on hyperlocal model design preset time sliding mode disturbance observer, compared with prior art, the beneficial effects of the application are: the present application constructs hyperlocal model and designs preset time sliding mode disturbance observer, realizes the preset time accurate observation of total disturbance caused by parameter mismatch and current coupling, overcomes the convergence time uncertainty and chattering problem of traditional observer;According to the observed disturbance, the hybrid duty calculation method is used to compensate the grid-side current and the midpoint voltage, accurately synthesize voltage vector, reduce the prediction error, so as to suppress the midpoint voltage fluctuation and grid-side current harmonic under parameter mismatch, improve the system robustness and control accuracy.

Description

Technical Field

[0001] This invention belongs to the field of rectifier control, and particularly relates to a model-free predictive control method for a capacitor-free Vienna rectifier. Background Technology

[0002] Vienna rectifiers, as a high-performance three-level rectifier topology, offer advantages such as low grid-side current harmonic content, high power density, and no bridge arm shoot-through risk. They are widely used in electric vehicle fast charging systems, data center DC power supplies, and commercial air conditioning power supplies. To further improve system power density and extend device lifespan, Vienna rectifiers typically use smaller film capacitors instead of large-capacity electrolytic capacitors on the DC side. However, reducing capacitor value significantly exacerbates midpoint voltage fluctuations, increasing voltage stress on switching devices and film capacitors, which is detrimental to system reliability and long-term stable operation. Existing control strategies for suppressing midpoint voltage fluctuations and improving grid-side power quality mainly include voltage and current dual closed-loop control, direct power control, and midpoint voltage hysteresis control.

[0003] Furthermore, the grid-side inductance and DC-side thin-film capacitor are key parameters in the modeling and control of the Vienna rectifier, and their identification accuracy directly affects the control performance of the grid-side current and midpoint voltage. Especially in model predictive control, when inductance or capacitor parameters are mismatched, the prediction accuracy of the midpoint voltage and grid-side current will significantly decrease, leading to deterioration in control performance. To address the uncertainty caused by parameter mismatch, traditional methods often employ extended state observers or sliding mode controllers to estimate and compensate for equivalent disturbances. However, existing observers often suffer from inconsistent convergence times and are prone to chattering in estimation errors, potentially resulting in inaccurate disturbance observations and difficulty in consistently ensuring system performance under parameter mismatch conditions, necessitating improvements. Summary of the Invention

[0004] Therefore, it is necessary to provide a model-free predictive control method for capacitor-free Vienna rectifiers to address the above-mentioned problems.

[0005] The present invention is implemented as follows: a model-free predictive control method for an electrolytic capacitor-free Vienna rectifier includes the following steps: Step S1: Construct a hyperlocal model of the Vienna rectifier, and design a preset time sliding mode disturbance observer based on the hyperlocal model. The sliding mode disturbance observer observes the total disturbance caused by parameter mismatch and grid-side current cross-coupling within a preset time to achieve model-free predictive control. Step S2: Based on the observed total disturbance, the prediction error of the midpoint voltage and grid-side current is reduced and the accuracy of model-free predictive control is improved by using a hybrid duty cycle calculation method that combines offline pre-calculation and online geometric solution.

[0006] In one embodiment, the present invention provides a model-free predictive control method for a capacitor-free Vienna rectifier. In step S1, a hyperlocal model of the capacitor-free Vienna rectifier is constructed, with the expression: ; In the formula, ẏ(t) is the controller output derivative, u(t) is the controller input, F(t) is the total disturbance, α is the control input adjustment coefficient, and t is the system running time.

[0007] In one embodiment, the present invention provides a model-free predictive control method for a capacitor-free Vienna rectifier, wherein in step S1, the time-domain model expression of the capacitor-free Vienna rectifier is: ; In the formula, i d For the d-axis network side current, i q For the q-axis network side current, u gd For the d-axis grid-side voltage, u gq U is the q-axis network-side voltage. dn The d-axis output voltage, u qn ω is the output voltage along the dq axis, ω is the electrical frequency, and ∆u is the voltage fluctuation at the midpoint. These are the model parameters for the inductance value. These are the model parameters for the capacitance value. These are the model parameters for the equivalent resistance, i. np The current is at the midpoint. Based on the hyperlocal and time-domain models of the capacitorless Vienna rectifier, the expression for the pre-defined time sliding mode perturbation observer structure is as follows: ; In the formula, Let F̂ be the output derivative of the sliding mode disturbance observer, g be the observed value of the disturbance term, l be the disturbance estimation gain, u be the control input, and α be the control input adjustment coefficient. is the derivative of the observed disturbance quantity.

[0008] In one embodiment, the present invention provides a model-free predictive control method for an electrolytic capacitor-free Vienna rectifier. In step S1, the total disturbance caused by the coupling of parameter disturbance and grid-side current is observed according to a preset time sliding mode disturbance observer. The difference between the actual value and the observed value is defined as the observation error, and the expression for this error and its derivative is as follows: ; In the formula, e is the error matrix between current and midpoint voltage. The derivative of the error matrix, The error matrix represents the perturbation values. Let be the derivative of the perturbation error matrix. dq For dq axis current, These are the observed values ​​of the dq-axis current. This represents the midpoint voltage fluctuation value. These are the observed values ​​of the midpoint voltage fluctuation. The sliding film control law for grid-side current. For the sliding control law of the midpoint voltage, This represents the dq-axis current disturbance. The observed values ​​of the dq-axis current disturbance are... This represents the voltage fluctuation disturbance value at the midpoint. The observed value is the voltage fluctuation disturbance value at the midpoint.

[0009] To ensure that the observation disturbance error converges to zero within a preset time, the sliding mode control law in the above disturbance observer should be designed as follows: ; In the formula, e(t) represents the observation disturbance error, and T c ρ is the upper bound of the error observation time, ρ is the asymptotic factor of the sliding mode control law, sgn(e(t)) is the sign function, and k is the contraction coefficient.

[0010] In one embodiment, the present invention provides a model-free predictive control method for an electrolytic capacitor-free Vienna rectifier, wherein step S2 includes: Step S21: Extract the total disturbance observed by the preset time sliding mode disturbance observer, compensate the grid-side current and midpoint voltage, and eliminate the interference of parameter mismatch and current coupling on state prediction. Step S22: Combine the compensated grid-side current and midpoint voltage to construct a cost function for model-free predictive control. Select the three voltage vectors that are closest to the reference voltage vector through the cost function to form a voltage vector sequence, so as to reduce the prediction error introduced by the voltage vector selection process. Step S23: A hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is adopted to solve the precise duty cycle of the three selected voltage vectors, thereby achieving accurate synthesis of the reference voltage vector, reducing the prediction error of the midpoint voltage and grid-side current, and improving the accuracy of model-free predictive control.

[0011] In one embodiment, the present invention provides a model-free predictive control method for a capacitor-free Vienna rectifier. In step S21, the grid-side current and the midpoint voltage are compensated, and the compensated expression is: .

[0012] In the formula, Let k be the observed value of the d-axis current at time k. Let k be the observed value of the q-axis current at time k. The observed value of the d-axis current at time k+1 is... Let be the observed value of the q-axis current at time k+1. Let k be the observed value of the midpoint voltage at time k. Let T be the observed value of the midpoint voltage at time k+1. s To control the cycle, u Ld (k) represents the d-axis inductance voltage at time k, u Lq (k) represents the q-axis inductor voltage at time k. The observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the midpoint voltage at time k. Let be the sliding control law for the d-axis current at time k. Let be the sliding film control law for the q-axis current at time k. The sliding control law for the midpoint voltage at time k; In one embodiment, the present invention provides a model-free predictive control method for a capacitor-free Vienna rectifier. In step S22, model-free predictive control is used to control the capacitor-free Vienna rectifier by combining the expressions for the compensated grid-side current and the midpoint voltage. The cost function expression for model-free predictive control is: ; In the formula, i dqref (k+1) is the reference value of the dq-axis current at time k+1, Δu ref (k+1) is the reference value of the midpoint voltage at time k+1. The observed value of the d-axis current at time k+1 is... Let λ be the observed value of the q-axis current at time k+1, and λ be the weighting factor. Based on the cost function, the three voltage vectors closest to the reference voltage vector are selected to form a voltage vector sequence for vector synthesis.

[0013] In one embodiment, the present invention provides a model-free predictive control method for an electrolytic capacitor-free Vienna rectifier. In step S23, a hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is used to solve for the precise duty cycle of the three selected voltage vectors. The expression for offline pre-calculation is: ; ; Among them, u α1 The α-axis voltage component of a vector, u α2 The α-axis voltage component is a 2-vector, u α3 For the α-axis voltage component of the 3-vector, uβ1 The β-axis voltage component of the vector, u β2 For the β-axis voltage component of the 2-vector, u β3 Let Δu1 be the β-axis voltage component of the 3-vector, Δu2 be the midpoint voltage increment of the 1-vector, Δu3 be the midpoint voltage increment of the 2-vector, n be the normal vector of a and b, and k be the voltage increment of the 1-vector. n Let be the norm of the normal vectors of a and b. It is a vector of 1. It is a 2-vector. There are 3 vectors, and c1, c2, Ω1 and Ω2 are coefficients calculated offline.

[0014] The expression for online geometric solution in the hybrid duty cycle calculation method is: ; ; Where d1 is the duty cycle of vector 1, d2 is the duty cycle of vector 2, and the duty cycle of the third vector is (1-d1-d2), u αβref For the reference voltage vector, ∆u ref d is the reference value for the midpoint voltage. c The duty cycle matrix is ​​λ, and the weighting factor in the cost function is λ. The switching transistor is directly driven according to the duty cycle of the three voltage vectors to achieve robust control of the capacitorless Vienna rectifier under parameter mismatch.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention constructs a hyperlocal model and designs a preset time sliding mode disturbance observer to achieve accurate observation of the total disturbance caused by parameter mismatch and current coupling at a preset time, overcoming the uncertainty of convergence time and chattering problems of traditional observers; based on the observed disturbance, a hybrid duty cycle calculation method is used to compensate for the grid-side current and midpoint voltage, accurately synthesize the voltage vector, and reduce prediction errors, thereby synergistically suppressing midpoint voltage fluctuations and grid-side current harmonics under parameter mismatch, and improving system robustness and control accuracy. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the circuit structure of the capacitorless Vienna rectifier provided in an embodiment of the present invention.

[0017] Figure 2 This is a control block diagram of a model-free predictive control method for an electrolytic capacitor-free Vienna rectifier provided in an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram illustrating the relationship between input current error and equivalent resistance and inductance mismatch, provided for an embodiment of the present invention.

[0019] Figure 4This is a schematic diagram illustrating the relationship between disturbance error, initial error value, and running time, provided in an embodiment of the present invention.

[0020] Figure 5 This is a schematic diagram illustrating the influence of different preset time upper bounds Tc and arrival law exponent ρ on poles, as provided in embodiments of the present invention.

[0021] Figure 6 The experimental waveforms of the traditional strategy and the proposed strategy are compared when the inductor parameters are mismatched (L=1mH) according to the embodiments of the present invention.

[0022] Figure 7 The experimental waveforms of the traditional strategy and the proposed strategy are compared when the inductor parameters are mismatched (L=5mH) according to the embodiments of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0024] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0025] In one embodiment, such as Figure 2 As shown, a model-free predictive control method for a capacitor-free Vienna rectifier includes the following steps: Step S1: Construct a hyperlocal model of the Vienna rectifier, and design a preset time sliding mode disturbance observer based on the hyperlocal model. The sliding mode disturbance observer observes the total disturbance caused by parameter mismatch and grid-side current cross-coupling within a preset time to achieve model-free predictive control. Step S2: Based on the observed total disturbance, the prediction error of the midpoint voltage and grid-side current is reduced and the accuracy of model-free predictive control is improved by using a hybrid duty cycle calculation method that combines offline pre-calculation and online geometric solution.

[0026] Traditional model predictive control (MPC) requires precise mathematical formulas for the controlled object (such as a motor or rectifier), but in reality, these parameters (such as inductance and capacitance) can change or become mismatched, leading to a decrease in control performance. Model-free predictive control attempts to break free from this constraint.

[0027] Online-updated hyperlocal models do not aim to build a globally accurate physical model. Instead, they use the system's input and output data to build a very simple hyperlocal model that only describes the behavior at the current instant within each extremely short control cycle.

[0028] On this continuously updated simplified model (hyperlocal model), the two major steps of model predictive control, "predicting future states and rolling optimization", are performed to calculate the optimal control quantity.

[0029] Please see Figure 1 The specific structure of the capacitorless Vienna rectifier is shown in the figure; in the figure, M is the midpoint of the three-phase power grid voltage, and L... s For inductance, R s D is the equivalent resistance on the grid side. ap D an D bp D bn D cp D cn The six diodes that form the rectifier bridge; S a1 S a2 S b1 S b2 S c1 S c2 For three-phase switching transistors, D a1 D a2 D b1 D b2 D c1 D c2 For the diode associated with the switching transistor, i sa i sb and i sc For grid-side inductor current, i p For the DC bus current, C p For the upper DC bus capacitor, i cp i is the current of the upper DC bus capacitor. np Let i be the midpoint current. cn i is the current of the lower DC bus capacitor. o Let i be the current flowing through the midpoint of the load. dc C is the current flowing through the load. n For the lower DC bus capacitor, i n Let P be the DC bus current, O be the DC bus top, O be the DC bus capacitor midpoint, and N be the DC bus low end.

[0030] Please see Figure 2 In the figure, i abc and i dq Let u be the grid-side current in the abc and dq coordinate systems, respectively. PO and u ONThe voltage at terminals PO and ON is u. dcref i is the DC bus voltage setpoint. dref Here, d is the given value for the d-axis current, ∆u is the midpoint voltage fluctuation, l is the sliding mode gain, and T is the given value for the d-axis current. c To set a preset time limit, Let g be the disturbance quantity, g be the sliding mode equation, and J be the variable. pro Let v be the cost function. 123 For the three selected voltage vectors, d a d b and d c Let α be the duty cycle of the three-phase switch and α be the control input coefficient. This strategy mainly consists of two parts: the construction of a preset-time sliding mode observer and a model-free predictive control strategy based on hybrid duty cycle calculation. First, a preset-time sliding mode disturbance observer is constructed based on the hyperlocal model of the Vienna rectifier to observe the total disturbance caused by parameter mismatch and dq-axis current coupling, ensuring stable convergence according to the preset time upper bound. Subsequently, discrete small-signal modeling is performed on PT-SMDO (a sliding mode disturbance observer control algorithm) to analyze the impact of the preset time upper bound and disturbance estimation gain on the steady-state and dynamic performance of the observer, obtaining the design criteria and accurate range of the main parameters. Finally, to improve the accuracy of model-free predictive control, the PT-SMDO observation results are extracted to compensate for the state prediction equation, and a hybrid duty cycle calculation strategy combining offline pre-calculation and online geometric solution is used to solve for the optimal duty cycle of the voltage vector sequence, thereby reducing the prediction error of the grid-side current and the midpoint voltage. When parameter mismatch occurs, compared with traditional strategies, the proposed strategy can further achieve synergistic suppression of grid-side current harmonics and midpoint voltage fluctuations under parameter mismatch.

[0031] Figure 3 The graph shows the relationship between input current error and equivalent resistance and inductance mismatch. It can be seen that the current error increases sharply with the increase of inductance mismatch and equivalent resistance mismatch.

[0032] Figure 4 The relationship between the disturbance error, the initial error value, and the running time is shown. It can be seen that the proposed observer can converge accurately within the preset time.

[0033] Figure 5 This figure illustrates the influence of different preset time upper bounds Tc and the arrival rate exponent ρ on the poles. The stable range of the preset time arrival rate exponent can be obtained from this figure.

[0034] In this embodiment, please refer to Figure 2 In step S1, a hyperlocal model of the capacitor-free Vienna rectifier is constructed, and its expression is: ; In the formula, ẏ(t) is the controller output derivative, u(t) is the controller input, F(t) is the total disturbance, α is the control input adjustment coefficient, and t is the system running time.

[0035] In step S1, the time-domain model expression of the capacitorless Vienna rectifier is: ; In the formula, i d For the d-axis network side current, i q For the q-axis network side current, u gd For the d-axis grid-side voltage, u gq U is the q-axis network-side voltage. dn The d-axis output voltage, u qn ω is the output voltage along the dq axis, ω is the electrical frequency, and ∆u is the voltage fluctuation at the midpoint. These are the model parameters for the inductance value. These are the model parameters for the capacitance value. These are the model parameters for the equivalent resistance, i. np The current is at the midpoint. Based on the hyperlocal and time-domain models of the capacitorless Vienna rectifier, the expression for the pre-defined time sliding mode perturbation observer structure is as follows: ; In the formula, Let F̂ be the output derivative of the sliding mode disturbance observer, g be the observed value of the disturbance term, l be the disturbance estimation gain, u be the control input, and α be the control input adjustment coefficient. is the derivative of the observed disturbance quantity.

[0036] In step S1, the total disturbance caused by the coupling between the parameter disturbance and the grid-side current is observed according to the preset time sliding mode disturbance observer. The difference between the actual value and the observed value is defined as the observation error, and the expression for this error and its derivative is as follows: ; In the formula, e is the error matrix between current and midpoint voltage. The derivative of the error matrix, The error matrix represents the perturbation values. Let be the derivative of the perturbation error matrix. dq For dq axis current, These are the observed values ​​of the dq-axis current. This represents the midpoint voltage fluctuation value. These are the observed values ​​of the midpoint voltage fluctuation. The sliding film control law for grid-side current. For the sliding control law of the midpoint voltage, This represents the dq-axis current disturbance. The observed values ​​of the dq-axis current disturbance are... This represents the voltage fluctuation disturbance value at the midpoint. The observed value is the voltage fluctuation disturbance value at the midpoint.

[0037] To ensure that the observation disturbance error converges to zero within a preset time, the sliding mode control law g in the above disturbance observer should be designed as follows: ; In the formula, e(t) represents the observation disturbance error, and T c ρ is the upper bound of the error observation time, ρ is the asymptotic factor of the sliding mode control law, sgn(e(t)) is the sign function, and k is the contraction coefficient.

[0038] In step S1, a diagonally composite Lyapunov function is constructed to prove the stability and convergence time of the preset time-sliding mode perturbation observer. The convergence time and the stable range of the sliding mode perturbation gain in the preset time-sliding perturbation observer are obtained through theoretical analysis.

[0039] When the inductor and capacitor parameters are mismatched, the convergence time and stability of the observer have a significant impact on the disturbance error. Therefore, it is necessary to ensure that the PT-SMDO converges stably and strictly according to the preset upper time bound. First, the time convergence of the observer is proven by constructing the function S(e) and its derivative dS / dt as follows: ; In the formula, e is the error matrix between current and midpoint voltage. Let T be the error matrix of the disturbance values. c ρ is the upper time bound of the sliding mode observer, ρ is the asymptotic factor of the sliding mode control law, and sgn(e) is the sign function.

[0040] Disturbances caused by mismatch in inductance and capacitance parameters are generally slowly varying disturbances, therefore satisfying ,in This is the upper bound of the derivative. To solve for the convergence time of PT-SMDO, the perturbation error e needs to be considered. F The derivative is calculated over a range, and it satisfies the following inequality: ; In the formula, The upper bound of the derivative, Let be the error matrix of the disturbance value, l be the disturbance estimation gain, and g be the sliding control law.

[0041] Therefore, by combining the above equation, we can construct a diagonally dominant composite Lyapunov function and solve for its derivative: ; In the formula, W(t) is the diagonally composite Lyapunov function, and Ẇ is its derivative. e is the current and midpoint voltage error matrix. Let T be the error matrix of the perturbation values, l be the perturbation estimation gain, and T be the error matrix of the perturbation values. c Let ρ be the upper bound of the error observation time, and ρ be the asymptotic factor of the sliding mode control law. This is the upper bound of the perturbation derivative. When the sliding mode gain satisfies... At that time, W(t) has at least 1 / T c The linear rate decreases monotonically, and at t=T c When the time drops to 0, the observer converges stably within the preset time.

[0042] In this embodiment, please refer to Figure 2 A model-free predictive control method for a capacitor-free Vienna rectifier, step S2 includes: Step S21: Extract the total disturbance observed by the preset time sliding mode disturbance observer, compensate the grid-side current and midpoint voltage, and eliminate the interference of parameter mismatch and current coupling on state prediction. Step S22: Combine the compensated grid-side current and midpoint voltage to construct a cost function for model-free predictive control. Select the three voltage vectors that are closest to the reference voltage vector through the cost function to form a voltage vector sequence, so as to reduce the prediction error introduced by the voltage vector selection process. Step S23: A hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is adopted to solve the precise duty cycle of the three selected voltage vectors, thereby achieving accurate synthesis of the reference voltage vector, reducing the prediction error of the midpoint voltage and grid-side current, and improving the accuracy of model-free predictive control.

[0043] In step S21, the grid-side current and midpoint voltage are compensated, and the compensated expression is: .

[0044] In the formula, Let k be the observed value of the d-axis current at time k. Let k be the observed value of the q-axis current at time k. The observed value of the d-axis current at time k+1 is... Let be the observed value of the q-axis current at time k+1. Let k be the observed value of the midpoint voltage at time k. Let T be the observed value of the midpoint voltage at time k+1. s To control the cycle, u Ld (k) represents the d-axis inductance voltage at time k, u Lq (k) represents the q-axis inductor voltage at time k. The observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the midpoint voltage at time k. Let be the sliding control law for the d-axis current at time k. Let be the sliding film control law for the q-axis current at time k. The sliding control law for the midpoint voltage at time k; In step S22, model-free predictive control is used to control the capacitorless Vienna rectifier, based on the expressions for the compensated grid-side current and midpoint voltage. The cost function J of the model-free predictive control is... pro The expression is: ; In the formula, i dqref (k+1) is the reference value of the dq-axis current at time k+1, Δu ref (k+1) is the reference value of the midpoint voltage at time k+1. The observed value of the d-axis current at time k+1 is... Let λ be the observed value of the q-axis current at time k+1, and λ be the weighting factor. Based on the cost function, the three voltage vectors closest to the reference voltage vector are selected to form a voltage vector sequence for vector synthesis.

[0045] In step S23, a hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is used to solve for the precise duty cycle of the three selected voltage vectors. The expression for offline pre-calculation is as follows: ; ; Among them, u α1 The α-axis voltage component of a vector, u α2 The α-axis voltage component is a 2-vector, u α3 For the α-axis voltage component of the 3-vector, u β1 The β-axis voltage component of the vector, u β2 For the β-axis voltage component of the 2-vector, u β3 Let Δu1 be the β-axis voltage component of the 3-vector, Δu2 be the midpoint voltage increment of the 1-vector, Δu3 be the midpoint voltage increment of the 2-vector, n be the normal vector of a and b, and k be the voltage increment of the 1-vector. n Let be the norm of the normal vectors of a and b. It is a vector of 1. It is a 2-vector. There are 3 vectors, and c1, c2, Ω1 and Ω2 are coefficients calculated offline.

[0046] The expression for online geometric solution in the hybrid duty cycle calculation method is: ; ; Where d1 is the duty cycle of vector 1, d2 is the duty cycle of vector 2, and the duty cycle of the third vector is (1-d1-d2), u αβref For the reference voltage vector, ∆u ref d is the reference value for the midpoint voltage. c The duty cycle matrix is ​​λ, and the weighting factor in the cost function is λ. The switching transistor is directly driven according to the duty cycle of the three voltage vectors to achieve robust control of the capacitorless Vienna rectifier under parameter mismatch.

[0047] For a specific example, the effectiveness of the model-free predictive control method for the capacitorless Vienna rectifier based on a preset time sliding mode disturbance observer was verified on a capacitorless Vienna rectifier platform. The experimental platform parameters were set as follows: grid frequency 50Hz, input rated inductance 3mH, switching frequency 50kHz, and DC-side capacitor rated capacitance 10μF.

[0048] from Figure 6 and Figure 7 As can be seen, under the traditional strategy, when the inductor model parameter is 1mH, the voltage vector selection is affected by the inductor parameter mismatch, resulting in severe midpoint voltage fluctuations. When the inductor model parameter is 5mH, the current change rate error of the voltage vector is large, leading to large fluctuations in the d-axis current and DC bus voltage, increasing the regulation pressure on the voltage outer loop. The maximum midpoint voltage fluctuations using the traditional strategy are 46.2V and 32.4V, respectively, with grid-side current THDs of 3.7% and 4.1%, respectively. Using the strategy proposed in this invention, PT-SMDO can guarantee deadbeat tracking of the d-axis current under inductor parameter mismatch, verifying the accuracy of the proposed observer. Furthermore, compensation of the state prediction equation based on disturbance observation can suppress d-axis current fluctuations. Under different inductor parameter mismatches, the maximum midpoint voltage fluctuations using the proposed strategy are 16.5V and 21.3V, respectively, with grid-side current THDs of 2.5% and 3.4%, respectively. The d-axis current fluctuations are reduced by 40.0% and 66.4% compared to the traditional strategy.

[0049] This invention constructs a hyperlocal model and designs a preset-time sliding mode disturbance observer to achieve accurate observation of the total disturbance caused by parameter mismatch and current coupling at a preset time, overcoming the uncertainty of convergence time and chattering problems of traditional observers. Based on the observed disturbance, a hybrid duty cycle calculation method is used to compensate for the grid-side current and midpoint voltage, accurately synthesizing the voltage vector and reducing prediction errors. Thus, under parameter mismatch, the midpoint voltage fluctuation and grid-side current harmonics are synergistically suppressed, improving the system robustness and control accuracy.

[0050] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0051] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0052] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0054] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A model-free predictive control method for an electrolytic capacitor-free Vienna rectifier, characterized in that, The model-free predictive control method for the capacitor-free Vienna rectifier includes the following steps: Step S1: Construct a hyperlocal model of the Vienna rectifier, and design a preset time sliding mode disturbance observer based on the hyperlocal model. The sliding mode disturbance observer observes the total disturbance caused by parameter mismatch and grid-side current cross-coupling within a preset time to achieve model-free predictive control. Step S2: Based on the observed total disturbance, the prediction error of the midpoint voltage and grid-side current is reduced and the accuracy of model-free predictive control is improved by using a hybrid duty cycle calculation method that combines offline pre-calculation and online geometric solution.

2. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 1, characterized in that, In step S1, a hyperlocal model of the capacitor-free Vienna rectifier is constructed, with the following expression: ； In the formula, ẏ(t) is the controller output derivative, u(t) is the controller input, F(t) is the total disturbance, α is the control input adjustment coefficient, and t is the system running time.

3. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 2, characterized in that, In step S1, the time-domain model expression of the capacitorless Vienna rectifier is: ； In the formula, i d For the d-axis network side current, i q For the q-axis network side current, u gd For the d-axis grid-side voltage, u gq U is the q-axis network-side voltage. dn The d-axis output voltage, u qn ω is the output voltage along the dq axis, ω is the electrical frequency, and ∆u is the voltage fluctuation at the midpoint. These are the model parameters for the inductance value. These are the model parameters for the capacitance value. These are the model parameters for the equivalent resistance, i. np The current is at the midpoint. Based on the hyperlocal and time-domain models of the capacitorless Vienna rectifier, the expression for the pre-defined time sliding mode perturbation observer structure is as follows: ； In the formula, Let F̂ be the output derivative of the sliding mode disturbance observer, g be the observed value of the disturbance term, l be the disturbance estimation gain, u be the control input, and α be the control input adjustment coefficient. is the derivative of the observed disturbance quantity.

4. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 1, characterized in that, In step S1, the total disturbance caused by the coupling between the parameter disturbance and the grid-side current is observed according to the preset time sliding mode disturbance observer. The difference between the actual value and the observed value is defined as the observation error, and the expression for this error and its derivative is as follows: ； In the formula, e is the error matrix between current and midpoint voltage. The derivative of the error matrix, The error matrix represents the perturbation values. i is the derivative of the perturbation error matrix; dq For dq axis current, These are the observed values ​​of the dq-axis current. This represents the midpoint voltage fluctuation value. These are the observed values ​​of the midpoint voltage fluctuation. The sliding film control law for grid-side current. For the sliding control law of the midpoint voltage, This represents the dq-axis current disturbance. The observed values ​​of the dq-axis current disturbance are... This represents the voltage fluctuation disturbance value at the midpoint. The observed value of the midpoint voltage fluctuation disturbance; To ensure that the observation disturbance error converges to zero within a preset time, the sliding mode control law in the above disturbance observer should be designed as follows: ； In the formula, e(t) is the observation disturbance error, and T c ρ is the upper bound of the error observation time, ρ is the asymptotic factor of the sliding mode control law, sgn(e(t)) is the sign function, and k is the contraction coefficient.

5. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to any one of claims 1 to 4, characterized in that, Step S2 includes: Step S21: Extract the total disturbance observed by the preset time sliding mode disturbance observer, compensate the grid-side current and midpoint voltage, and eliminate the interference of parameter mismatch and current coupling on state prediction. Step S22: Combine the compensated grid-side current and midpoint voltage to construct a cost function for model-free predictive control. Select the three voltage vectors that are closest to the reference voltage vector through the cost function to form a voltage vector sequence, so as to reduce the prediction error introduced by the voltage vector selection process. Step S23: A hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is adopted to solve the precise duty cycle of the three selected voltage vectors, thereby achieving accurate synthesis of the reference voltage vector, reducing the prediction error of the midpoint voltage and grid-side current, and improving the accuracy of model-free predictive control.

6. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 5, characterized in that, In step S21, the grid-side current and midpoint voltage are compensated, and the compensated expression is: ； In the formula, Let k be the observed value of the d-axis current at time k. Let k be the observed value of the q-axis current at time k. The observed value of the d-axis current at time k+1 is... Let be the observed value of the q-axis current at time k+1. Let k be the observed value of the midpoint voltage at time k. Let T be the observed value of the midpoint voltage at time k+1. s To control the cycle, u Ld (k) represents the d-axis inductance voltage at time k, u Lq (k) represents the q-axis inductor voltage at time k. The observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the d-axis current at time k. Let k be the observed value of the disturbance of the midpoint voltage at time k. Let be the sliding control law for the d-axis current at time k. Let be the sliding film control law for the q-axis current at time k. Let k be the sliding control law for the midpoint voltage at time k.

7. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 5, characterized in that, In step S22, model-free predictive control is used to control the capacitorless Vienna rectifier, based on the expressions for the compensated grid-side current and midpoint voltage. The cost function expression for model-free predictive control is as follows: ； In the formula, i dqref (k+1) is the reference value of the dq-axis current at time k+1, Δu ref (k+1) is the reference value of the midpoint voltage at time k+1. The observed value of the d-axis current at time k+1 is... Let λ be the observed value of the q-axis current at time k+1, and λ be the weighting factor. Based on the cost function, the three voltage vectors closest to the reference voltage vector are selected to form a voltage vector sequence for vector synthesis.

8. The model-free predictive control method for an electrolytic capacitor-free Vienna rectifier according to claim 5, characterized in that, In step S23, a hybrid duty cycle calculation method combining offline pre-calculation and online geometric solution is used to solve for the precise duty cycle of the three selected voltage vectors. The expression for offline pre-calculation is as follows: ； ； Among them, u α1 The α-axis voltage component of a vector, u α2 The α-axis voltage component is a 2-vector, u α3 For the α-axis voltage component of the 3-vector, u β1 The β-axis voltage component of the vector, u β2 For the β-axis voltage component of the 2-vector, u β3 Let Δu1 be the β-axis voltage component of the 3-vector, Δu2 be the midpoint voltage increment of the 1-vector, Δu3 be the midpoint voltage increment of the 2-vector, n be the normal vector of a and b, and k be the voltage increment of the 1-vector. n Let be the norm of the normal vectors of a and b. It is a vector of 1. It is a 2-vector. For a 3-vector system, c1, c2, Ω1, and Ω2 are coefficients calculated offline; The expression for online geometric solution in the hybrid duty cycle calculation method is: ； ； Where d1 is the duty cycle of vector 1, d2 is the duty cycle of vector 2, and the duty cycle of the third vector is (1-d1-d2), u αβref For the reference voltage vector, ∆u ref d is the reference value for the midpoint voltage. c The duty cycle matrix is ​​λ, and the weighting factor in the cost function is λ. The switching transistor is directly driven according to the duty cycle of the three voltage vectors to achieve robust control of the capacitorless Vienna rectifier under parameter mismatch.

Images