A high-power LLC resonant converter hybrid control smooth switching method and system

Abstract

The application discloses a high-power LLC resonant converter hybrid control smooth switching method and system, and belongs to the technical field of power electronic converter control. The method comprises the following steps: collecting input voltage, output voltage and load current, calculating equivalent voltage gain and determining control mode; predicting target control quantity when mode switching is detected; taking frequency conversion and phase shift active disturbance rejection controller as two subsystems of a switching system, constructing Lyapunov function, determining state mapping relationship according to energy non-increasing condition, using gain ratio conversion, taking the predicted control quantity as a correction term to write the initial state of the target extended state observer; setting an event trigger condition, taking the mode switching edge as a trigger source, temporarily increasing the bandwidth of the observer and smoothly reducing it; calculating the control quantity and generating a PWM signal. The application can improve the switching transient process, and takes into account the rapidity and noise immunity.

Description

Technical Field

[0001] This invention relates to the field of power electronic converter control technology, specifically to a hybrid control smooth switching method and system for a high-power LLC resonant converter. Background Technology

[0002] LLC resonant converters are used in wide input / wide output applications such as DC charging for electric vehicles, energy storage systems, and high-voltage DC power supplies for data centers because they can achieve zero-voltage turn-on of the primary-side switch and zero-current turn-off of the secondary-side rectifier diode. To meet the requirements of wide-range voltage regulation and efficiency under all operating conditions, frequency conversion control and phase-shift control are typically combined. Frequency conversion control is suitable for boost voltage and regulation near the resonant point, while phase-shift control is suitable for buck voltage and light-load conditions. When switching between the two modes, the control channel gain and the physical meaning of the control variables change abruptly, which can easily cause output voltage drops or overshoots, resonant current surges, and weakened soft-switching characteristics.

[0003] Existing smooth switching schemes are mostly based on proportional-integral controllers or their improved algorithms. Some schemes introduce intermediate variables to unify the model, but the nonlinear changes at the mode boundaries are significant, and single or fixed parameters are difficult to take into account the dynamic characteristics of both modes. Some schemes rely on complex online parameter adjustments, resulting in large computational delays and difficulty in adapting to switching transients. In addition, although active disturbance rejection control has been introduced, it has the ability to resist disturbances, but the extended state observer state is usually cleared or simply assigned a value during switching, without considering the impact of sudden changes in control channel gain on the physical meaning of the state, which can easily lead to an increase in observation error energy; and fixed observer bandwidth is difficult to balance steady-state high-frequency noise suppression and rapid transient tracking.

[0004] Therefore, it is essential to design a hybrid control smooth switching method and system for high-power LLC resonant converters. Summary of the Invention

[0005] The purpose of this invention is to provide a hybrid control smooth switching method and system for high-power LLC resonant converters, which can improve the switching transient process and balance speed and noise immunity.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A hybrid control smooth switching method for a high-power LLC resonant converter includes: Step S1: Real-time acquisition of input voltage, output voltage and load current of LLC main circuit, calculation of current equivalent voltage gain, and determination of current control mode based on comparison of equivalent voltage gain with preset threshold. Step S2: When a mode switching signal is detected, predict the initial control quantity in the target control mode according to the fundamental approximation model or a pre-calibrated offline lookup table; Step S3: Treat the frequency converter active disturbance rejection controller and the phase-shift active disturbance rejection controller as two subsystems of the ADRC hybrid control module, construct a Lyapunov function, determine the state mapping relationship based on the condition that the system energy does not increase at the moment of switching, use the ratio between the control channel gains before and after switching to perform energy equivalent conversion on the extended state observer state, and incorporate the predicted control quantity in step S2 as a correction term into the state mapping relationship, calculate the initial state of the extended state observer of the target active disturbance rejection controller based on the state mapping relationship, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port under the switching trigger; Step S4: Set the event triggering condition based on the observation error, and use the mode switching edge signal as the trigger source. When the event triggering condition is met or the mode switching edge signal is detected, increase the observer bandwidth of the extended state observer to a transient high value. After triggering, smoothly reduce the observer bandwidth to a steady-state low value according to the preset decay law. Step S5: In each control cycle, the activated active disturbance rejection controller calculates the final control quantity based on the updated extended state observer state and the linear state error feedback control law. The frequency conversion active disturbance rejection controller outputs the frequency control quantity, the phase-shifting active disturbance rejection controller outputs the phase-shifting control quantity, and the PWM generator generates the PWM signal to drive the LLC main circuit switching transistor based on the final control quantity, and selects the corresponding PWM signal to output according to the current control mode.

[0007] Furthermore, in step S1, the formula for calculating the current equivalent voltage gain is as follows: ; In the formula, Input voltage, For output voltage, This represents the turns ratio of the primary and secondary sides of the transformer.

[0008] Further, in step S1, the current control mode is determined based on the comparison result between the equivalent voltage gain and the preset threshold, specifically as follows: If the equivalent voltage gain When the value exceeds the preset threshold, the system operates in frequency conversion control mode; If the equivalent voltage gain When the threshold value is less than the preset lower limit, the system operates in phase-shift control mode; If the equivalent voltage gain The current mode remains unchanged between the upper and lower limits of the preset threshold.

[0009] Furthermore, in step S2, if switching from frequency conversion control mode to phase shift control mode, the fixed switching frequency is the resonant frequency. The predicted phase shift angle is: ; In the formula, The equivalent gain in the frequency converter control mode before switching; If switching from phase-shift control mode to frequency conversion control mode, the fixed phase shift angle is 0, and the gain is determined based on the equivalent gain corresponding to the current phase shift angle. The target switching frequency can be obtained by looking up a table in reverse. .

[0010] Further, in step S4, the function for the event triggering condition is: ; when Alternatively, when a mode switching signal is detected, an event is triggered, which jumps the ESO's observer bandwidth to its maximum value.

[0011] Furthermore, in step S4, the preset decay law is exponential decay.

[0012] This invention also provides a hybrid control smooth switching system for a high-power LLC resonant converter, applied to the aforementioned hybrid control smooth switching method for a high-power LLC resonant converter. The system includes: an LLC main circuit module, a signal sampling module, a mode determination module, an ADRC hybrid control module, an extended state observer state mapping and initialization module, and an event-triggered variable bandwidth control module. The LLC main circuit module is connected to the signal sampling module and the mode determination module. The signal sampling module, the mode determination module, the extended state observer state mapping and initialization module, and the event-triggered variable bandwidth control module are all connected to the ADRC hybrid control module. The signal sampling module is used to collect the input voltage, output voltage and load current of the LLC main circuit module in real time; The mode determination module is used to calculate the current equivalent voltage gain based on the acquired signal and output a mode selection signal through a hysteresis comparator. The ADRC hybrid control module includes a frequency converter active disturbance rejection controller and a phase-shift active disturbance rejection controller, which perform state control based on the reference voltage and the output voltage, respectively. The frequency converter active disturbance rejection controller outputs a frequency control quantity, and the phase-shift active disturbance rejection controller outputs a phase-shift control quantity. The extended state observer state mapping and initialization module is used to receive the mode selection signal and the current state of the controller, calculate the initial state of the target controller based on the Lyapunov stability theory, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port. The event-triggered variable bandwidth control module is used to dynamically adjust the observer bandwidth of the extended state observer based on the observation error and the mode switching edge signal.

[0013] Furthermore, the ADRC hybrid control module also includes a frequency conversion PWM generator and a phase-shift PWM generator. The frequency conversion PWM generator is connected to the frequency conversion active disturbance rejection controller, and the phase-shift PWM generator is connected to the phase-shift active disturbance rejection controller. The frequency conversion PWM generator and the phase-shift PWM generator are used to generate PWM signals to drive the switching transistors of the LLC main circuit according to the control quantity output by the ADRC hybrid control module, and output the corresponding PWM signals according to the mode selection signal.

[0014] In summary, the present invention has at least one of the following beneficial technical effects: 1. By using the state mapping mechanism based on the Lyapunov energy-increase condition, the switching of the dual active disturbance rejection controller is modeled as a switching system. The energy equivalent conversion is performed using the control channel gain ratio, which can suppress the trend of increased energy of the extended state observer error during switching and improve the stability of the switching process.

[0015] 2. By co-initializing the predictive control quantity with the state mapping, the predictive control quantity obtained from the fundamental approximation model or offline lookup table is integrated into the initial state of the extended state observer, making the control quantity more continuous after switching, shortening the transient convergence process, and reducing the output voltage deviation.

[0016] 3. By using an event-triggered variable bandwidth mechanism, the observer bandwidth is actively increased at the mode switching edge and then smoothly restored to a low bandwidth according to a preset rule, taking into account both transient fast response and steady-state high-frequency noise suppression.

[0017] 4. The dual active disturbance rejection architecture works in conjunction with the state mapping and variable bandwidth modules, reducing the dependence on accurate models and improving the adaptability and robustness of the hybrid control strategy in high-power applications with wide input / wide output. Attached Figure Description

[0018] Figure 1 A block diagram of a high-power LLC resonant converter hybrid control smooth switching system provided in an embodiment of the present invention; Figure 2 The diagram shows the internal structure and signal flow of an active disturbance rejection controller based on state mapping and event triggering. Figure 3 A schematic diagram of the hybrid control smooth switching method for a high-power LLC resonant converter provided in an embodiment of the present invention; Figure 4 This diagram illustrates a comparison of the output voltage during mode switching between traditional proportional-integral control, unmodified active disturbance rejection control, and the improved active disturbance rejection control of this invention. Detailed Implementation

[0019] 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. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0020] like Figure 1 and Figure 2 As shown, this invention provides a high-power LLC resonant converter hybrid control smooth switching system, including: an LLC main circuit module, a signal sampling module, a mode determination module, an ADRC hybrid control module, an extended state observer (ESO) state mapping and initialization module, and an event-triggered variable bandwidth control module. The LLC main circuit module is connected to the signal sampling module and the mode determination module, and the signal sampling module, the mode determination module, the extended state observer state mapping and initialization module, and the event-triggered variable bandwidth control module are all connected to the ADRC hybrid control module. The signal sampling module is used to collect the input voltage, output voltage and load current of the LLC main circuit module in real time; The mode determination module is used to calculate the current equivalent voltage gain based on the acquired signal and output a mode selection signal through a hysteresis comparator. The ADRC hybrid control module includes a frequency converter active disturbance rejection controller (FM-ADRC) and a phase-shift active disturbance rejection controller (PS-ADRC), which perform state control based on the reference voltage and the output voltage, respectively. The frequency converter active disturbance rejection controller outputs a frequency control quantity, and the phase-shift active disturbance rejection controller outputs a phase-shift control quantity. The extended state observer state mapping and initialization module is used to receive the mode selection signal and the current state of the controller, calculate the initial state of the target controller based on the Lyapunov stability theory, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port. The event-triggered variable bandwidth control module is used to dynamically adjust the observer bandwidth of the extended state observer based on the observation error and the mode switching edge signal.

[0021] The ADRC hybrid control module further includes a frequency conversion PWM generator and a phase-shift PWM generator. The frequency conversion PWM generator is connected to the frequency conversion active disturbance rejection controller, and the phase-shift PWM generator is connected to the phase-shift active disturbance rejection controller. The frequency conversion PWM generator and the phase-shift PWM generator are used to generate PWM signals to drive the switching transistors of the LLC main circuit according to the control quantity output by the ADRC hybrid control module, and output the corresponding PWM signals according to the mode selection signal.

[0022] Compared to the conventional dual-independent ADRC architecture in the prior art, the system described in this invention provides a physical basis for smooth switching at the hardware level by adding an ESO state mapping and initialization module and an event-triggered variable bandwidth control module. At the moment of switching, the mapping module calculates the optimal initial state of the target ESO using Lyapunov energy constraints and gain ratio calculations, and forces loading it through a specially added external state injection port; the variable bandwidth module dynamically increases the observer bandwidth under the triggering of the switching event, and restores the low bandwidth exponentially after the switch. This combination of "active state transition" and "active bandwidth scheduling" is not present in the existing simple parallel architecture of dual ADRC, and is the key to overcoming the aforementioned technical barriers. The algorithmic aspects working in conjunction with this are described in detail in the subsequent methods section of this invention.

[0023] like Figure 3 As shown, the present invention also provides a hybrid control smooth switching method for a high-power LLC resonant converter, applied to the above-mentioned system, comprising: Step S1: Real-time acquisition of input voltage, output voltage and load current of LLC main circuit, calculation of current equivalent voltage gain, and determination of current control mode based on comparison of equivalent voltage gain with preset threshold. Step S2: When a mode switching signal is detected, predict the initial control quantity in the target control mode according to the fundamental approximation model or a pre-calibrated offline lookup table; Step S3: Treat the frequency converter active disturbance rejection controller and the phase-shift active disturbance rejection controller as two subsystems of the ADRC hybrid control module, construct a Lyapunov function, determine the state mapping relationship based on the condition that the system energy does not increase at the moment of switching, use the ratio between the control channel gains before and after switching to perform energy equivalent conversion on the extended state observer state, and incorporate the predicted control quantity in step S2 as a correction term into the state mapping relationship, calculate the initial state of the extended state observer of the target active disturbance rejection controller based on the state mapping relationship, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port under the switching trigger; Step S4: Set the event triggering condition based on the observation error, and use the mode switching edge signal as the trigger source. When the event triggering condition is met or the mode switching edge signal is detected, increase the observer bandwidth of the extended state observer to a transient high value. After triggering, smoothly reduce the observer bandwidth to a steady-state low value according to the preset decay law. Step S5: In each control cycle, the activated active disturbance rejection controller calculates the final control quantity based on the updated extended state observer state and the linear state error feedback control law. The frequency conversion active disturbance rejection controller outputs the frequency control quantity, the phase-shifting active disturbance rejection controller outputs the phase-shifting control quantity, and the PWM generator generates the PWM signal to drive the LLC main circuit switching transistor based on the final control quantity, and selects the corresponding PWM signal to output according to the current control mode.

[0024] Next, each step will be explained in detail: Step S1: Signal Acquisition and Mode Determination. Real-time acquisition of input voltage. Output voltage and load current Calculate the current equivalent voltage gain: ; in, This is the turns ratio of the primary and secondary sides of the transformer. When... When the value exceeds the preset upper limit threshold (1.05), the system operates in frequency conversion control mode; when When the value is less than the preset lower threshold (0.95), the system operates in phase-shift control mode; when the value is between the upper and lower thresholds, the current mode remains unchanged to form a hysteresis interval and prevent frequent mode jitter.

[0025] Step S2: Target Control Quantity Prediction. When the mode determination module outputs a mode switching signal, the initial control quantity under the target control mode is calculated based on the fundamental frequency approximation (FHA) model or a pre-calibrated offline lookup table. If switching from frequency conversion to phase shifting, the fixed switching frequency is the resonant frequency. Predict the phase shift angle: ; in The equivalent gain in the previous inverter mode; if switching from phase shift to inverter mode, with the phase shift angle fixed at 0, the equivalent gain is based on the current phase shift angle. The target switching frequency can be obtained by looking up a table in reverse. ; The predictive control quantity obtained in this step All of these are included in the ESO initialization as correction terms of the state mapping matrix in step S3. In contrast, existing technologies often employ predictive feedforward, typically... It is directly superimposed on the control output terminal to form The combination of these factors does not change the initial state of the ESO within the ADRC; the ESO still requires tens of cycles to converge, resulting in a significant error in the output control quantity at the moment of switching. This invention does not directly output the predicted value, but rather transforms it into the initial total disturbance of the ESO. The correction amount allows ADRC to naturally produce a control action consistent with the predicted value.

[0026] Step S3: ESO state mapping and initialization based on Lyapunov stability (combined with...) Figure 2 ). Figure 2 This is a diagram showing the internal structure and signal flow of the ADRC based on state mapping and event triggering. The ADRC includes a tracking differentiator (TD), an extended state observer (ESO), and a linear state error feedback control law (LSEF). The ESO contains three state variables: (Output estimation) (Output differential estimate) and (Total perturbation estimation). At the instant of switching, construct the quadratic Lyapunov function of the ESO error system and design the state mapping matrix that satisfies the energy non-increase condition. The initial state of the target controller ESO is calculated using this mapping matrix. , , The target ADRC is written through the external state injection port to achieve zero-order continuity of the control quantity and attenuation of observation error energy.

[0027] Existing conventional ADRC systems typically only assign a simple value to or clear the ESO state during mode switching, for example, by directly setting... Or it can reuse the old value. This type of method ignores... Mutation pairs A fundamental change in physical meaning makes it impossible to guarantee the perturbation compensation term. The continuity of the system is lacking, and it is completely devoid of theoretical constraints. In contrast, the originality of step S3 of this invention lies in: for the first time modeling the dual ADRC switching as a switching system, deriving the inequality constraints of the state mapping matrix using Lyapunov stability theory, and... The ratio is used as the core element of the mapping matrix for energy equivalence conversion, while the predictive control quantity is converted through the correction term. Integrating into the ESO, a three-in-one initialization mechanism of "theoretical constraints + physical conversion + prediction fusion" is formed. This mechanism is not taught in existing literature, nor can it be naturally derived from conventional experience, fundamentally solving the technical barrier of transient divergence during conventional ADRC switching. The specific process of step 3 will be elaborated in detail below: 1. Consider FM-ADRC and PS-ADRC as two subsystems of the switching system. First, establish the error dynamic equation of the subsystem ESO: ; Constructing a quadratic Lyapunov function: ; To ensure system stability during the switching process, the condition of no energy increase must be met: ; This leads to the state mapping matrix. Matrix inequality constraints that must be satisfied: ; 2. Specialize abstract matrix inequalities into mapping matrices with explicit physical meaning: ; diagonal elements Solved Caused by mutation Energy equivalence conversion problem, ensuring disturbance compensation amount The physical continuity. This innovatively introduces the predictive control quantity in step S2. As a feedforward correction, the physical model of LLC is deeply integrated with the state transition of ADRC.

[0028] 3. Based on the mapping matrix, the specific initial state value of the target ESO is calculated, and it is written into the ESO through a specially added external state injection port and trigger pulse in the hardware, thereby realizing the above-mentioned theoretical optimal migration at the physical level.

[0029] Through the above three steps, this invention successfully overcomes the technical barriers of conventional ADRC, which suffers from "no theoretical constraints, no physical conversion, and no predictive fusion" during mode switching.

[0030] Step S4: Event-triggered variable bandwidth ESO control. An event-triggered evaluator is designed to address high-frequency switching noise under high-power conditions. The event-triggered condition function is defined as follows: ; In step S4, the function for the event triggering condition is: ; when Or, when a mode switching signal is detected, an event is triggered, increasing the ESO's observer bandwidth. Jump to maximum value After triggering, the observer bandwidth decays exponentially to smoothly transition to the base low bandwidth. ; Traditional ESO bandwidth Once set, it becomes a constant. A low value fails to quickly track switching disturbances, while a high value amplifies high-frequency noise; both are mutually exclusive. Existing technologies, even employing simple variable bandwidth strategies based on error amplitude, only passively adjust after the error has become significant, resulting in response lag, and do not use the mode switching edge as an active trigger source. The originality of step S4 in this invention lies in: for the first time, using the mode switching signal Edge as the highest priority trigger event, achieving a zero-delay bandwidth leap, and designing an exponential decay law guaranteed by ISS theory, enabling the bandwidth to smoothly return to a low steady-state value without secondary impact. This mechanism theoretically decouples the contradiction between "fast transient response" and "strong steady-state noise resistance," representing a crucial step in overcoming the aforementioned technical barriers through event-driven control and time-varying observer theory.

[0031] The present invention will now elaborate on step S4 above, which specifically includes the following: 1. Design a composite event triggering mechanism and define the event triggering function as follows: ; For the first time, the mode switching edge signal is used as an independent, highest-priority trigger source. When the edge signal arrives, the ESO bandwidth immediately jumps to [a higher level] without waiting for error accumulation. It achieves zero-delay transient response.

[0032] 2. The design incorporates an exponential decay law guaranteed by ISS theory. After triggering, the observer bandwidth does not remain constant with a step change, but rather decays smoothly according to an exponential differential equation, as follows: ; Furthermore, by rigorously deriving the upper bound of the estimation error under time-varying bandwidth using the input-state stability (ISS) theory, the stability during the decay process is mathematically proven, ensuring that the system can smoothly and without secondary oscillations return to a highly noise-resistant low-bandwidth steady state.

[0033] 3. Hardware-based independent evaluator module In terms of hardware architecture, this mechanism is implemented as a standalone "event-triggered variable bandwidth estimator" (as described in the patent). Figure 2 It outputs dynamic bandwidth in real time and dynamically refreshes ESO gain.

[0034] Through the above scheme, the present invention theoretically decouples the contradiction between "transient speed" and "steady-state noise immunity" that is irreconcilable in traditional ADRC.

[0035] Step S5: Control output and PWM generation (combined) Figure 3 and Figure 4 ). Figure 3This is a flowchart illustrating the smooth switching logic and state mapping execution for hybrid control. Within each control cycle, ADRC calculates the control input based on the updated ESO state. Figure 4 The comparison charts showing the smooth switching effects under different control strategies demonstrate the significant advantages of the method of this invention over traditional PI and unmodified ADRC in suppressing voltage drops and shortening recovery time.

[0036] The present invention also provides an embodiment, taking an LLC resonant converter of a high-power DC charger as an example, to elaborate on the above method in detail: High-power circuit parameters and basic control parameters settings: In this embodiment, the LLC main circuit parameters are as follows: resonant inductor resonant capacitor Magnetizing inductor primary-to-secondary turns ratio of transformer Rated load resistance .

[0037] The resonant frequency was calculated to be: Characteristic impedance Equivalent AC load Quality factor Inductance ratio The system input voltage range is 400V-600V, and the output voltage range is 0V-6kV.

[0038] S1: Real-time acquisition of input voltage Output voltage and load current Calculate the current equivalent voltage gain: ; in This is the turns ratio of the primary and secondary sides of the transformer. When... When the value exceeds the preset upper limit threshold (1.05), the system operates in frequency conversion control mode; when When the value is less than the preset lower threshold (0.95), the system operates in phase-shift control mode; when the value is between the upper and lower thresholds, the current mode remains unchanged to form a hysteresis interval and prevent frequent mode jitter.

[0039] S2: After detecting the switching edge pulse signal, the system needs to quickly calculate the initial control quantity (i.e., the predicted phase shift angle or the predicted switching frequency) in the target control mode to ensure the continuity of the output voltage at the moment of switching. This embodiment provides two alternative prediction implementation methods: the fundamental frequency approximation (FHA) analytical method and the offline lookup table method based on simulation data pre-calibration.

[0040] (1) Prediction based on the fundamental frequency approximation (FHA) analytical model Prediction during the switch from frequency conversion mode to phase shift mode: When switching to phase shift control mode, the fixed switching frequency is the resonant frequency. According to the FHA model, the equivalent voltage gain in phase-shift control mode. With phase angle The relationship is approximated as: ; Assuming the equivalent gain is conserved before and after the switch, calculate the current equivalent gain. Then, substituting the values, we obtain the predicted target phase shift angle as: ; Considering gain Measurement noise, if If, then let; Then let .

[0041] Prediction of the transition from phase-shifting mode to frequency conversion mode: When switching from phase-shifting mode to frequency conversion mode, the phase shift angle is fixed at 0. First, based on the current phase shift angle... Calculate the current equivalent gain: ; Subsequently, according to the FHA frequency-gain equation: ; make The predicted normalized frequencies were obtained by solving the problem offline using numerical methods (Newton's iteration method). Thus, the predicted target switching frequency is obtained. .

[0042] in, To normalize the switching frequency, This is the actual switching frequency.

[0043] (2) Prediction based on pre-calibrated offline data lookup table Considering the FHA model under extremely light load conditions (in this embodiment) Due to inherent errors, this embodiment preferably employs an offline lookup table method based on pre-calibrated simulation data. During offline calibration, a steady-state operating point scan of the LLC resonant converter is performed in a high-fidelity simulation environment (Matlab / Simulink): keeping the load constant, the input voltage and switching frequency are changed, and the output voltage at each steady-state point is recorded to generate a "frequency-voltage" relationship table in the frequency conversion mode; the switching frequency is fixed at... The phase shift angle is changed to generate a "phase shift angle-voltage" relationship table in phase shift mode. This data table is then reorganized into a two-dimensional lookup table that directly retrieves the target control quantity from the input and output voltages, and written into the digital controller's storage unit. During online prediction, the digital controller collects the current... and It can be obtained directly through linear interpolation. or .

[0044] Regardless of which of the above implementation methods is used, the resulting predictive control quantity All of these will be used as key feedforward corrections in step S3 for the calculation of the ESO state mapping.

[0045] S3: Deep derivation of ESO state mapping based on Lyapunov stability. In the dual ADRC architecture, the frequency conversion ADRC (FM-ADRC) and phase shift ADRC (PS-ADRC) each have independent extended state observers (ESOs). Figure 2 As shown, each ESO contains three state variables: (Output estimation) (Output differential estimate) and (Total perturbation estimation). Define the estimation error vector of ESO. for: ; in, For ESO state vectors, This represents a true expansion state. For system output, This represents the total system disturbance.

[0046] The dynamic equation for ESO error in continuous time is: ; in, To represent the closed-loop error system matrix, let A be the system matrix: ; C is the output matrix: ; E is the perturbation input matrix: ; L is the observer gain vector: ; ; in, , , These represent the first, second, and third gain coefficients of the ESO, respectively. This represents the observer bandwidth of the ESO.

[0047] Model the dual ADRC system as a switching system. Switching signals At the switching time Due to the control channel gain The mutation (in this embodiment, the critical gain identified by FM-ADRC) Critical gain of PS-ADRC When the ESO state transitions, the mapping relationship is as follows: ; in, This indicates the time after switching (Mode 2, phase-shifting mode) The ESO estimation error vector, This indicates the time before switching (Mode 1, Variable Frequency Mode) The ESO estimation error vector of the previous instant. To represent the state mapping matrix from mode 1 to mode 2, Indicates the switching time The moment before the limit.

[0048] To ensure the global asymptotic stability of the switching system, a quadratic Lyapunov function is constructed: ; in, Representation pattern Lyapunov energy function under these conditions; Representation pattern A positive definite symmetric matrix; Indicates the index of the control mode. 1 represents frequency conversion mode, and 2 represents phase shift mode. Matrix And it satisfies the Lyapunov equation: ; in, Representation pattern The closed-loop error system matrix is ​​as follows; Representation pattern Given a positive definite symmetric matrix, the energy constant condition must be satisfied at the instant of switching: ; in, This represents the Lyapunov function value of Mode 2 at the time of switching. This represents the Lyapunov function value of Mode 1 just before the switch. Substituting this into the state mapping relationship, we obtain the matrix inequality constraints: ; in, Represents a positive definite symmetric matrix under mode 2; Represents a positive definite symmetric matrix under mode 1.

[0049] Based on this constraint, the present invention designs a state mapping matrix. The specific form is as follows: ; in, This represents the critical gain parameter of ADRC under variable frequency control mode; This represents the critical gain parameter of ADRC in phase-shift control mode; To predict the control quantity based on step S2 Correction items.

[0050] Based on the mapping matrix above, the initial state of the target ESO is calculated as follows: ; in, , This represents the theoretical phase shift angle calculated based on the FHA model. When using the offline lookup table method for prediction... This can be further reduced to zero. By writing these initial values ​​into the target ADRC's ESO through an external state injection port under pulse triggering, the continuity of the control quantity at the moment of switching can be achieved, and the inherent limitations of the control can be eliminated. The energy surge in observation error caused by a sudden change.

[0051] Conversely, when switching back from phase-shifting mode to frequency conversion mode, the state mapping matrix Γ2→1 is designed similarly, only requiring the diagonal elements to be replaced with... And adjust the prediction correction terms accordingly.

[0052] S4: To address the strong high-frequency switching noise characteristic of high-power LLC converters, this invention proposes an event-triggered variable bandwidth ESO mechanism. For example... Figure 2 As shown, the event-triggered variable bandwidth estimator receives the estimated values ​​of the output voltage and ESO in real time. Define the event trigger condition function: ; in, express The value of the event triggering condition function at any given moment; express The actual output voltage measurement value at that moment; express Estimated output value of ESO at time t; Indicates the relative threshold coefficient; This represents an absolute threshold. In this embodiment, a relative threshold is set. absolute threshold When a mode switching signal is detected, an event is triggered, and the ESO bandwidth jumps to [a certain value]. After triggering, the observer bandwidth decays exponentially to ensure a smooth transition. ; in, Indicates the attenuation rate coefficient; express ESO observer bandwidth at time t, This is the basic low bandwidth under steady-state conditions. (This embodiment...) , .

[0053] According to the input-to-state stability (ISS) theory of linear time-varying systems, the error of the time-varying bandwidth ESO satisfies: ; in, express The norm of the ESO estimation error vector at time step; Represents positive constants related to the initial state of the system; Indicates the initial time. The norm of the ESO estimation error vector; Indicates the bandwidth of the observer The relevant transient convergence rate function; Represents the integral variable; This represents the upper bound constant of the derivative of the total disturbance of the system; Indicates the corresponding basic low bandwidth The minimum convergence rate at which the time is reached.

[0054] The event triggers an instant boost It can significantly increase the transient convergence rate. This accelerates error decay; the subsequent exponential decay ensures that the system eventually returns to a low-bandwidth steady state with high noise immunity, thus achieving the theoretical decoupling of transient speed and steady-state noise immunity.

[0055] Event-triggered variable bandwidth estimator output dynamic bandwidth The data is fed into the ESO gain generation unit to calculate the observer gain in real time. This allows for dynamic adjustment of the ESO's tracking performance.

[0056] S5: As Figure 1 and Figure 3 As shown, within each control cycle, the activated ADRC calculates the final control quantity based on the updated ESO state and the linear state error feedback control law. Taking the phase-shifting ADRC as an example, its control law is as follows: ; ; in, For reference output voltage, , , This is the controller bandwidth (60 rad / s in this embodiment). The control law form is the same for the frequency converter ADRC, only... The parameters are different.

[0057] Two PWM generators based on the final control quantity (Switching frequency in variable frequency mode) or phase shift angle in phase shift mode It generates PWM pulses to drive the switching transistors of the LLC main circuit, and then selects the specific PWM according to the current mode to send to the main circuit, thus completing the smooth switching and closed-loop regulation of the entire hybrid control.

[0058] Simulation comparison verification and effect analysis (combined) Figure 4 To verify the effectiveness of this invention, a detailed high-power LLC simulation model was built in Simulink, and simulations were performed. Comparative test of switching from frequency conversion mode (input 450V, output 3kV) to phase shift mode (input 550V, output 3kV). Figure 4 The output voltage waveforms under three control strategies are shown: Traditional PI control: Traditional PI controllers do not rely on an observer and can directly respond to error signals at the moment of switching, thus their initial adjustment speed is relatively fast. However, because PI controllers lack the ability to feedforward compensation for sudden changes in control channel gain and the ability to estimate total disturbances, they cannot predict the target control quantity at the moment of switching, resulting in a momentary drop in output voltage of up to 680V (approximately 23%). Unmodified conventional ADRC control: Conventional ADRC estimates the total disturbance in real time through ESO, which can compensate for the dynamic changes caused by mode switching to some extent. Therefore, compared with PI control, the voltage drop at the moment of switching is significantly reduced, with the maximum measured deviation being approximately 270V (about 9%). However, since conventional ADRC only assigns or clears the ESO state during switching without considering the impact of sudden changes in control gain on the observation error energy, it requires additional convergence time. This results in a relatively slow recovery process, with an overall recovery time of approximately 0.26s, slightly longer than that of the PI controller.

[0059] The improved ADRC control of this invention achieves a maximum output voltage deviation of only 10V (0.3%) at the moment of switching, realizing a smooth transition with almost no disturbance and reducing the recovery time to less than 0.05s.

[0060] As can be seen from the above specific embodiments and comparative data, the present invention effectively solves the transient oscillation problem of high-power LLC resonant converters during hybrid control switching, and has significant technological progress and engineering application value.

[0061] Embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0062] Contents not described in detail in this specification are prior art known to those skilled in the art. It is hereby indicated that the above description is intended to help those skilled in the art understand this invention, but does not limit the scope of protection of this invention. Any equivalent substitutions, modifications, improvements, or simplifications of the above descriptions that do not depart from the essential content of this invention fall within the scope of protection of this invention.

Claims

1. A hybrid control smooth switching method for a high-power LLC resonant converter, characterized in that, include: Step S1: Real-time acquisition of input voltage, output voltage and load current of LLC main circuit, calculation of current equivalent voltage gain, and determination of current control mode based on comparison of equivalent voltage gain with preset threshold. Step S2: When a mode switching signal is detected, predict the initial control quantity in the target control mode according to the fundamental approximation model or a pre-calibrated offline lookup table; Step S3: Treat the frequency converter active disturbance rejection controller and the phase-shift active disturbance rejection controller as two subsystems of the ADRC hybrid control module, construct a Lyapunov function, determine the state mapping relationship based on the condition that the system energy does not increase at the moment of switching, use the ratio between the control channel gains before and after switching to perform energy equivalent conversion on the extended state observer state, and incorporate the predicted control quantity in step S2 as a correction term into the state mapping relationship, calculate the initial state of the extended state observer of the target active disturbance rejection controller based on the state mapping relationship, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port under the switching trigger; Step S4: Set the event triggering condition based on the observation error, and use the mode switching edge signal as the trigger source. When the event triggering condition is met or the mode switching edge signal is detected, increase the observer bandwidth of the extended state observer to a transient high value. After triggering, smoothly reduce the observer bandwidth to a steady-state low value according to the preset decay law. Step S5: In each control cycle, the activated active disturbance rejection controller calculates the final control quantity based on the updated extended state observer state and the linear state error feedback control law. The frequency conversion active disturbance rejection controller outputs the frequency control quantity, the phase-shifting active disturbance rejection controller outputs the phase-shifting control quantity, and the PWM generator generates the PWM signal to drive the LLC main circuit switching transistor based on the final control quantity, and selects the corresponding PWM signal to output according to the current control mode.

2. The hybrid control smooth switching method for high-power LLC resonant converters according to claim 1, characterized in that, In step S1, the formula for calculating the current equivalent voltage gain is as follows: ； In the formula, Input voltage, For output voltage, This represents the turns ratio of the primary and secondary sides of the transformer.

3. The hybrid control smooth switching method for high-power LLC resonant converters according to claim 2, characterized in that, In step S1, the current control mode is determined based on the comparison result between the equivalent voltage gain and the preset threshold, specifically as follows: If the equivalent voltage gain When the value exceeds the preset threshold, the system operates in frequency conversion control mode; If the equivalent voltage gain When the threshold value is less than the preset lower limit, the system operates in phase-shift control mode; If the equivalent voltage gain The current mode remains unchanged between the upper and lower limits of the preset threshold.

4. The hybrid control smooth switching method for high-power LLC resonant converters according to claim 3, characterized in that, In step S2, if switching from frequency conversion control mode to phase shift control mode, the fixed switching frequency is the resonant frequency. The predicted phase shift angle is: ； In the formula, The equivalent gain in the frequency converter control mode before switching; If switching from phase-shift control mode to frequency conversion control mode, the fixed phase shift angle is 0, and the gain is determined based on the equivalent gain corresponding to the current phase shift angle. The target switching frequency can be obtained by looking up a table in reverse. .

5. The hybrid control smooth switching method for a high-power LLC resonant converter according to claim 4, characterized in that, In step S4, the function for the event triggering condition is: ； when Alternatively, when a mode switching signal is detected, an event is triggered, which jumps the ESO's observer bandwidth to its maximum value.

6. The hybrid control smooth switching method for a high-power LLC resonant converter according to claim 5, characterized in that, In step S4, the preset decay law is exponential decay.

7. A hybrid control smooth switching system for a high-power LLC resonant converter, applied to the hybrid control smooth switching method for a high-power LLC resonant converter as described in any one of claims 1-6, characterized in that, include: The LLC main circuit module comprises a signal sampling module, a mode judgment module, an ADRC hybrid control module, an extended state observer state mapping and initialization module, and an event-triggered variable bandwidth control module. The LLC main circuit module is connected to the signal sampling module and the mode judgment module. The signal sampling module, the mode judgment module, the extended state observer state mapping and initialization module, and the event-triggered variable bandwidth control module are all connected to the ADRC hybrid control module. The signal sampling module is used to collect the input voltage, output voltage and load current of the LLC main circuit module in real time; The mode determination module is used to calculate the current equivalent voltage gain based on the acquired signal and output a mode selection signal through a hysteresis comparator. The ADRC hybrid control module includes a frequency converter active disturbance rejection controller and a phase-shift active disturbance rejection controller, which perform state control based on the reference voltage and the output voltage, respectively. The frequency converter active disturbance rejection controller outputs a frequency control quantity, and the phase-shift active disturbance rejection controller outputs a phase-shift control quantity. The extended state observer state mapping and initialization module is used to receive the mode selection signal and the current state of the controller, calculate the initial state of the target controller based on the Lyapunov stability theory, and write it into the extended state observer of the target active disturbance rejection controller through the external state injection port. The event-triggered variable bandwidth control module is used to dynamically adjust the observer bandwidth of the extended state observer based on the observation error and the mode switching edge signal.

8. The hybrid control smooth switching method for a high-power LLC resonant converter according to claim 7, characterized in that, The ADRC hybrid control module further includes a frequency conversion PWM generator and a phase-shift PWM generator. The frequency conversion PWM generator is connected to the frequency conversion active disturbance rejection controller, and the phase-shift PWM generator is connected to the phase-shift active disturbance rejection controller. The frequency conversion PWM generator and the phase-shift PWM generator are used to generate PWM signals to drive the switching transistors of the LLC main circuit according to the control quantity output by the ADRC hybrid control module, and output the corresponding PWM signals according to the mode selection signal.

Images