DC bus ripple suppression method for totem-pole power factor correction circuits

By constructing a ripple estimation model and real-time correcting of the equivalent capacitance and load power estimates, the totem-pole power factor correction circuit achieves ripple suppression under both steady-state and dynamic operating conditions under high-speed dynamic loads of servers. This solves the problem of ripple suppression and fast dynamic response being mutually exclusive due to voltage loop bandwidth limitations, and improves the accuracy of steady-state ripple suppression and the speed of dynamic response.

CN123122227APending Publication Date: 2026-11-17XIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611296696.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In single-phase totem-pole power factor correction circuits, under high-speed dynamic loads in servers, the voltage loop bandwidth limitation leads to a problem where voltage ripple suppression and fast dynamic response cannot be simultaneously achieved.

Method used

By constructing a ripple estimation model, the estimated values ​​of equivalent capacitance and load power are corrected in real time, and time-sharing coordinated control is implemented to achieve ripple suppression under steady-state and dynamic operating conditions. A voltage divider and conditioning circuit is used to sample the DC bus voltage in real time, and the voltage sample value is obtained by an analog-to-digital converter. A ripple estimation model is constructed based on the power balance principle, and error filtering and iterative calibration are performed by sliding window averaging.

Benefits of technology

While ensuring dynamic response speed, it improves steady-state ripple suppression accuracy and actively suppresses 100Hz double power frequency ripple, solving the problem that ripple suppression and fast dynamic response cannot be achieved simultaneously under voltage loop bandwidth limitations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122227A_ABST
    Figure CN123122227A_ABST
Patent Text Reader

Abstract

The DC bus ripple suppression method for the totem-pole power factor correction circuit disclosed in this invention specifically includes the following steps: S1. Given a voltage reference value V dc_ref Collect DC bus voltage sampling value V dc S1. Obtain the actual steady-state ripple and construct a ripple estimation model ΔV; S2. Based on the DC bus voltage sampling value V dc S1. Determine the peak-to-peak value of the second harmonic ripple and the ripple error based on the actual steady-state ripple. Set a discrimination threshold ε based on the peak-to-peak value of the second harmonic ripple and determine the system operating condition based on the discrimination threshold ε. S2. Correct the equivalent capacitance C under steady-state conditions or the estimated load power under dynamic conditions based on the ripple estimation model ΔV to obtain the compensation amount. S3. Combine the compensation amount with the DC bus voltage sampling value V. dc The DC feedback voltage V is obtained by superposition. dcf The purpose of this invention is to provide a DC bus ripple suppression method for totem-pole power factor correction circuits, solving the problem of the inability to simultaneously achieve voltage ripple suppression and fast dynamic response in totem-pole PFC circuits.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the technical field of power electronic converter control methods, and relates to a DC bus ripple suppression method for a totem-pole power factor correction circuit. Background Technology

[0002] High-density server power supplies need to achieve high efficiency, high power density, and excellent dynamic voltage regulation performance simultaneously under single-phase AC power supply. A two-stage architecture consisting of a front-end totem-pole PFC (Power Factor Correction Circuit) and a rear-end LLC resonant converter has become the mainstream solution in this field due to its advantages of low input current harmonics, high efficiency, and compact structure. In this architecture, the stable operation of the rear-end LLC resonant converter highly depends on the stable DC bus voltage provided by the front-end totem-pole PFC. The inherent power ripple characteristics of the single-phase AC input of the totem-pole PFC will generate a voltage ripple at twice the grid frequency, such as 100Hz, on its DC output bus. If this ripple enters the voltage outer loop through the voltage feedback path, it will cause input current distortion and power factor deterioration. To block this coupling path, the voltage outer loop cutoff frequency is forced to be reduced to 10-20Hz in engineering practice. When the server load experiences a large step change within microseconds, the low-bandwidth voltage loop cannot adjust in time, and the bus voltage will drop significantly or overshoot, triggering undervoltage or overvoltage protection in severe cases.

[0003] To address the aforementioned issues, existing technologies employ open-loop compensation by estimating load power to accelerate response speed. However, the accuracy of this technology is highly dependent on the bus capacitance value and the model's accuracy. When parameters drift, a steady-state error is generated, leading to a degradation in the steady-state ripple suppression effect. Summary of the Invention

[0004] The purpose of this invention is to provide a DC bus ripple suppression method for totem-pole power factor correction circuits, solving the problem that voltage ripple suppression and fast dynamic response cannot be simultaneously achieved due to voltage loop bandwidth limitations in single-phase totem-pole power factor correction circuits under high-speed dynamic loads of servers.

[0005] The technical solution adopted in this invention is a DC bus ripple suppression method for a totem-pole power factor correction circuit, which specifically includes the following steps: S1. Given voltage reference value V dc_ref Collect DC bus voltage sampling value V dc Obtain the actual steady-state ripple And construct a ripple estimation model ΔV; S2. Based on the DC bus voltage sampling value V dc Actual steady-state ripple Determine the peak-to-peak value of the second harmonic ripple. and ripple error According to the peak-to-peak value of the second harmonic ripple Set a discrimination threshold ε, and determine the system operating condition based on the discrimination threshold ε: when When ≤ε, the system is considered to be in steady state. When the value is greater than ε, the system is determined to be in a dynamic operating condition. S3. Estimated equivalent capacitance C under steady-state conditions or load power under dynamic conditions based on the ripple estimation model ΔV. Correction, compensation amount obtained after correction ; S4. Adjust the compensation amount With DC bus voltage sampling value V dc The DC feedback voltage V is obtained by superposition. dcf Using DC feedback voltage V dcf Replacement DC bus voltage sampling value V dc To achieve ripple feedback.

[0006] The invention is further characterized by: The DC bus voltage is sampled in real time by a voltage divider and conditioning circuit, and the sampled DC bus voltage value V is obtained by an analog-to-digital converter. dc Actual steady-state ripple The calculation method is as follows: (1) Based on the power balance principle and the charge integral characteristics of equivalent capacitance, a ripple estimation model ΔV is constructed: (2) (3) (4) in, For grid power, This is an estimated load power value. This is the effective value of the grid voltage. This represents the current amplitude output by the voltage loop. This is the load current.

[0007] Second harmonic ripple peak-to-peak value It can be represented as: (5) Where f is the grid frequency, C is the equivalent capacitance of the DC bus, and V dc_ave This represents the average value of the DC bus voltage. Ripple error It can be represented as: (6) The discrimination threshold ε can be expressed as: ε=k (7) Where k is the proportionality coefficient, and the value of k ranges from 0.1 to 0.3.

[0008] Under steady-state conditions, the equivalent capacitance C is calibrated iteratively on a minute-by-minute basis. The specific process includes: S11. Define instantaneous ripple error : (8) S12. Using a fixed-benchmark split model, the equivalent capacitance actually involved in the model calculation is obtained. : (9) in, The equivalent capacitance is the factory-standard nominal value and remains fixed during the control cycle; This is the equivalent capacitance deviation correction amount; for Set a limit protection to prevent excessive correction of deviation due to abnormal operating conditions.

[0009] The instantaneous ripple error e is smoothed by mean filtering. slow Mean filtering is implemented using a sliding window averaging method, with a sliding window duration of 30–60 seconds; channel error e slow It can be represented as: (10) Based on channel error Correction amount for equivalent capacitance deviation Make corrections: (11) in, The equivalent capacitance correction ratio is adjusted in real time based on the nominal value of the equivalent capacitance, with a single correction step not exceeding 1% of the nominal value of the equivalent capacitance; the correction cycle is on the order of minutes.

[0010] when When >0, This reduces the equivalent capacitance. ;when When <0, This increases the equivalent capacitance. By measuring the equivalent capacitance To enable real-time updates It converges to zero.

[0011] Under dynamic operating conditions, the load power is estimated using the control cycle as the step size. Make corrections: (12) (13) (14) in, (k+1) represents the estimated load power for the next control cycle. (k) represents the estimated load power for the current control cycle. For load power increment, This is the load power correction factor. The range of values ​​is This is the DC feedback voltage.

[0012] when - If the value is greater than 0, the compensation amount is too large, so the estimated load power value should be increased. ;when - If the value is less than 0, the compensation is insufficient, so the estimated load power should be reduced. By estimating the load power To enable real-time updates It converges to zero.

[0013] Compensation amount High-precision matching with real-time AC ripple, compensation amount It can be represented as: (15) Among them, the estimated load power The initial value is determined by the grid power during the system startup phase. Determined, that is .

[0014] The beneficial effects of this invention are: By slowly calibrating the equivalent bus capacitance under steady-state conditions and quickly correcting the load power estimate under dynamic conditions, dual-time-scale time-sharing coordinated control of steady-state and dynamic conditions is achieved. This ensures the speed of dynamic response while improving the accuracy of steady-state ripple suppression, and coordinates the optimization of dynamic and steady-state performance. It actively suppresses 100Hz double power frequency ripple, solving the problem that ripple suppression and fast dynamic response cannot be simultaneously achieved under the limitation of voltage loop bandwidth. Attached Figure Description

[0015] Figure 1 This is a topology diagram of a traditional totem-pole PFC converter; Figure 2 It is a traditional totem pole PFC dual closed-loop control block diagram; Figure 3 This is a flowchart of the dynamic and static state judgment process of the system of the present invention; Figure 4 This is a flowchart of the capacitor value correction under steady-state conditions according to the present invention; Figure 5This is a flowchart of the load power correction under dynamic operating conditions according to the present invention; Figure 6 This is the block diagram of the totem pole PFC dual closed-loop control of the present invention. Detailed Implementation

[0016] The following detailed description is provided in conjunction with specific implementation methods.

[0017] This invention focuses on totem-pole PFC, i.e., totem-pole power factor correction circuit, and its system structure is as follows: Figure 1 As shown, the system includes the mains voltage V from the AC input source. g The circuit consists of an input inductor L, a totem-pole bridge arm Q, the equivalent capacitance C of the DC bus, and the downstream load R. The totem-pole bridge arm Q is composed of high-frequency arms Q1 and Q2 connected in parallel with low-frequency arms Q3 and Q4. The low-frequency arms employ a power frequency synchronous commutation mechanism, while the high-frequency arms rely on high-frequency switches to achieve power factor correction and energy conversion, transforming the AC input into a stable DC bus voltage to power the downstream load.

[0018] Input inductance L: Used to store and release energy.

[0019] High-frequency bridge arms Q1 and Q2 are responsible for high-frequency switching to control the shape of the input current and boost the voltage.

[0020] Low-frequency bridge arms Q3 and Q4: switch according to the positive and negative half-cycles of the grid voltage. Their main function is to provide a current loop and achieve synchronous rectification.

[0021] Working principle of totem pole PFC circuit: When the AC input voltage is positive: Low-frequency bridge arm operation: Low-frequency bridge arm Q4 remains continuously conducting to provide a grounding loop for the circuit; Q3 remains off.

[0022] High-frequency bridge arm operation: At this time, high-frequency bridge arms Q2 and Q1 are turned on complementaryly with high-frequency PWM signals, forming an equivalent Boost circuit.

[0023] When the high-frequency bridge arm Q2 is turned on, the current path is: AC input, inductor L, Q2, Q4, back to the input terminal. At this time, inductor L stores energy.

[0024] When the high-frequency bridge arm Q2 is turned off and Q1 is turned on, the energy stored in inductor L cannot change abruptly. The current path becomes: AC input, inductor L, Q1, output capacitor and load, Q4, back to the input. At this time, inductor L releases energy to charge the output capacitor and supply power to the load.

[0025] When the AC input voltage is negative: Power frequency bridge arm operation: Power frequency bridge arm Q3 remains continuously conducting to provide a current loop; Q4 remains off.

[0026] High-frequency bridge arm operation: At this time, high-frequency bridge arms Q1 and Q2 are also turned on by complementary high-frequency PWM signals, but their roles are reversed, forming another equivalent Boost circuit.

[0027] When the high-frequency bridge arm Q1 is turned on, the current path is: input terminal, Q3, Q1, inductor L, back to AC input. Inductor L stores energy in the reverse direction.

[0028] When the high-frequency bridge arm Q1 is turned off and Q2 is turned on, the inductor L releases energy, and the current path is: input terminal, Q3, output capacitor and load, Q2, inductor L, back to AC input.

[0029] Through the above process, the totem pole PFC can achieve power factor correction and voltage boosting functions throughout the entire AC voltage cycle.

[0030] The standard dual-closed-loop control process of totem-pole PFC with feedforward decoupling, consisting of an outer voltage loop and an inner current loop, is as follows: Figure 2 As shown, its core idea is: the outer loop stabilizes the voltage, the inner loop shapes the current, the feedforward eliminates interference, and the PLL locks the phase.

[0031] The totem pole PFC system adopts a dual closed-loop structure with an outer voltage loop and an inner current loop. The specific working process is as follows: 1. Outer loop voltage stabilization control: The outer voltage loop uses the voltage reference value V dc_ref For the target, the DC bus voltage sampling value V dc The feedback signal is subtracted, resulting in a voltage deviation signal ΔV. dc The signal is fed into the voltage loop PI regulator, which outputs the command current amplitude I. ref , representing the average power current amplitude required to maintain bus balance; 2. PLL, Phase Locking and Reference Generation: Acquires the instantaneous phase and frequency of the AC input voltage and tracks the grid voltage V in real time. g The current amplitude I is determined by the zero-crossing point and phase angle θ, and outputs a standard cosθ or sinθ. ref Command and grid voltage V g Multiplying the synchronous phase signal by cosθ or sinθ yields the result with respect to the grid voltage V. g In-phase instantaneous current loop given command i ref ; 3. Inner loop current shaping control: This allows the actual inductor current i to be shaped. L Fast and error-free tracking of the i generated by the outer loop ref , change i ref With the actual inductor current i L The difference, the current deviation signal Δi, is sent to the current loop PI regulator, and after passing through a digital delay stage. Then, after the decoupling process, D=(Vg -V L ) / v dc The output control quantity D and T are obtained. s For the digitally controlled sampling period, V L This is the inductor voltage.

[0032] Digital delay This is the biggest gap between the real system and the control theory simulation, which limits the current loop bandwidth. Sufficient phase margin needs to be reserved when designing PI parameters. The delay includes a 1-step calculation delay and a 0.5-step PWM update delay, which together result in a total delay: T. s +0.5T s =1.5T s ; Based on the traditional totem pole PFC, this invention achieves root cause suppression and stable filtering of bus second harmonic ripple by real-time correction of capacitor value and load power estimation value, thereby improving steady-state ripple suppression accuracy while ensuring dynamic response speed.

[0033] The DC bus ripple suppression method for totem-pole power factor correction circuits specifically includes the following steps: S1. Given voltage reference value V dc_ref Collect DC bus voltage sampling value V dc Obtain the actual steady-state ripple And construct a ripple estimation model ΔV; S2. Based on the DC bus voltage sampling value V dc Actual steady-state ripple Determine the peak-to-peak value of the second harmonic ripple. and ripple error According to the peak-to-peak value of the second harmonic ripple Set a discrimination threshold ε, and determine the system operating condition based on the discrimination threshold ε: when When ≤ε, the system is considered to be in steady state. When the value is greater than ε, the system is determined to be in a dynamic operating condition. S3. Estimated equivalent capacitance C under steady-state conditions or load power under dynamic conditions based on the ripple estimation model ΔV. Correction, compensation amount obtained after correction ; S4. Adjust the compensation amount With DC bus voltage sampling value V dc The DC feedback voltage V is obtained by superposition. dcf Using DC feedback voltage V dcf Replacement DC bus voltage sampling value V dc To achieve ripple feedback.

[0034] The DC bus voltage is sampled in real time by a voltage divider and conditioning circuit, and the sampled DC bus voltage value V is obtained by an analog-to-digital converter. dc Actual steady-state ripple The calculation method is as follows: (1) Based on the power balance principle and the charge integral characteristics of equivalent capacitance, a ripple estimation model ΔV is constructed: (2) (3) (4) in, For grid power, This is an estimated load power value. This is the effective value of the grid voltage. This represents the current amplitude output by the voltage loop. This is the load current.

[0035] Second harmonic ripple peak-to-peak value It can be represented as: (5) Where f is the grid frequency, C is the equivalent capacitance of the DC bus, and V dc_ave This represents the average value of the DC bus voltage. Ripple error It can be represented as: (6) The discrimination threshold ε can be expressed as: ε=k (7) Where k is the proportionality coefficient, and the value of k ranges from 0.1 to 0.3.

[0036] Under steady-state conditions, the equivalent capacitance C is calibrated iteratively on a minute-by-minute basis. The specific process includes: S11. Define instantaneous ripple error : (8) S12. Using a fixed-benchmark split model, the equivalent capacitance actually involved in the model calculation is obtained. : (9) in, The equivalent capacitance is the factory-standard nominal value and remains fixed during the control cycle; This is the equivalent capacitance deviation correction amount; for Set a limit protection to prevent excessive correction of deviation due to abnormal operating conditions.

[0037] The instantaneous ripple error e is smoothed by mean filtering. slow Mean filtering is implemented using a sliding window averaging method, with a sliding window duration of 30–60 seconds; channel error e slow It can be represented as: (10) Based on channel error Correction amount for equivalent capacitance deviation Make corrections: (11) in, The equivalent capacitance correction ratio is adjusted in real time based on the nominal value of the equivalent capacitance, with a single correction step not exceeding 1% of the nominal value of the equivalent capacitance; the correction cycle is on the order of minutes.

[0038] when When >0, This reduces the equivalent capacitance. ;when When <0, This increases the equivalent capacitance. By measuring the equivalent capacitance To enable real-time updates It converges to zero.

[0039] Under dynamic operating conditions, the load power is estimated using the control cycle as the step size. Make corrections: (12) (13) (14) in, (k+1) represents the estimated load power for the next control cycle. (k) represents the estimated load power for the current control cycle. For load power increment, This is the load power correction factor. The range of values ​​is This is the DC feedback voltage.

[0040] when - If the value is greater than 0, the compensation amount is too large, so the estimated load power value should be increased. ;when - If the value is less than 0, the compensation is insufficient, so the estimated load power should be reduced. By estimating the load power To enable real-time updates It converges to zero.

[0041] Compensation amount High-precision matching with real-time AC ripple, compensation amount It can be represented as: (15) Among them, the estimated load power The initial value is determined by the grid power during the system startup phase. Determined, that is .

[0042] Example 1 Determine the actual steady-state ripple And construct a ripple estimation model ΔV The DC bus voltage of the totem-pole PFC is sampled in real time through a voltage divider and conditioning circuit. This sampled voltage is then periodically acquired by an ADC (analog-to-digital converter) to acquire continuously changing analog signals from nature. The continuous analog values ​​are converted into discrete digital values ​​for the control algorithm to process, resulting in the sampled DC bus voltage value V. dc .

[0043] Extracting actual steady-state ripple : The actual steady-state ripple is calculated based on the voltage reference value and the DC bus voltage sampling value. The calculation method is as follows: (1) Among them, V dc_ref This is the voltage reference value; Constructing a ripple estimation model for bus voltage : Based on the power balance principle and the integral characteristic of capacitor charge, a bus voltage ripple estimation model is constructed. : (2) (3) (4) in, For grid power, This is an estimated load power value. The initial value is determined by the grid power during the system startup phase. Determined, that is , This is the sampled value of the DC bus voltage; This is the effective value of the grid voltage. This represents the current amplitude output by the voltage loop. This is the load current.

[0044] Example 2 System steady-state and dynamic discrimination Based on actual steady-state ripple The system operating condition is determined by comparing the error with the peak-to-peak value of the ripple. The peak-to-peak value of the second harmonic ripple can be expressed as: (5) Where, ΔV ripple The peak-to-peak value of the second harmonic ripple is given, where f is the grid frequency, C is the equivalent capacitance of the DC bus, and V is the peak value of the second harmonic ripple. dc_ave This represents the average value of the DC bus voltage. Define ripple error : (6) Set the discrimination threshold ε: ε=k (7) Where k is the proportionality coefficient. Considering the combined effects of power grid frequency fluctuations, sampling noise, and capacitor parameter drift, the value of k ranges from 0.1 to 0.3, with a typical value of 0.2.

[0045] The judgment logic is as follows: When ΔV err When the value is greater than ε, it indicates that the actual ripple deviates significantly from the current estimation model, and the system is determined to be in a dynamic operating condition. When ΔV err When ≤ε, it indicates that the model is basically consistent with reality, and the system is judged to be in steady state.

[0046] This discrimination provides a basis for subsequent dual-timescale parameter correction. The discrimination flowchart is as follows: Figure 3 As shown.

[0047] Example 3 Ripple correction, after which a compensation amount f(C, P) is obtained that is highly accurate in matching the real-time AC ripple. load ) Ripple correction under steady-state conditions: After determining that the system is in steady-state operation, the bus voltage ripple residual is mainly caused by the slow drift of the bus equivalent capacitance parameter C, such as temperature effects and device aging. At this point, a slow iterative calibration is performed on the equivalent capacitance to maintain the long-term accuracy of the ripple estimation model. The implementation process is as follows: Figure 4 As shown.

[0048] First, define the instantaneous ripple error e: (8) To accurately reflect the physical characteristics of capacitors, a split-type model with a fixed reference and slow adjustment for deviations is adopted. The equivalent capacitance actually involved in the model calculation is represented as follows: (9) Wherein, C0 is the nominal value of the capacitor at the factory, which remains fixed within the control cycle; ΔC is the capacitor deviation correction amount, updated slowly at the minute level. This strategy only fine-tunes the deviation amount ΔC without changing the reference C0, which conforms to the physical characteristic that the capacitance value will only drift slightly around the reference value due to temperature and aging, ensuring the long-term accuracy and control stability of the model. At the same time, a limiting protection is set for ΔC to prevent excessive correction of the deviation due to abnormal operating conditions.

[0049] Long-term mean filtering, slow iterative correction, and directional correction logic: Considering the slow-varying characteristics of capacitor parameter drift, the instantaneous ripple error e is not directly used in the control. Instead, it is subjected to long-term mean filtering to obtain a smoothed slow-channel error e. slow Long-term mean filtering is achieved using a sliding window averaging method, with the preferred sliding window duration being 30 to 60 seconds, corresponding to 30 to 60 one-second update cycles.

[0050] (10) The capacitance deviation ΔC is corrected using only the slow channel error signal: (11) According to the ripple estimation model ΔV formula, the ripple estimation model ΔV is inversely proportional to the capacitance value C. Therefore, the capacitance value is directionally corrected based on the polarity of the ripple error: when e slow When the value is greater than 0, it indicates that the actual steady-state ripple is... If the ripple estimation model ΔV is greater than the value of C, then C needs to be reduced; when e slow When <0, it indicates actual steady-state ripple. If the value is less than ΔV in the ripple estimation model, C needs to be increased.

[0051] Where, k c The capacitance correction ratio is adjusted in real time according to the actual situation and the nominal value of the capacitor used. The value is very small to ensure that the single correction step does not exceed 1% of the nominal value of the capacitor. The update cycle is on the order of minutes, such as once every 30 seconds or 1 minute. Only the deviation ΔC is finely adjusted so that the equivalent capacitance C always drifts slightly near the reference value C0. This can effectively compensate for the parameter deviation caused by temperature and aging and ensure the steady-state accuracy of the ripple model.

[0052] Slow correction of bus capacitance, nominal capacitance C0=220uF, If C < 0.1C0, take 1uF and correct once every 30 seconds. For the new capacitor, the actual capacitance C = 218uF, with an initial deviation of 2uF, therefore, correct twice. After correction... C is 2uF. For long-term aging capacitors, C=178uF, initial deviation 42uF, correction times 42, after correction... C is 42uF.

[0053] Example 4 Ripple correction under dynamic operating conditions: When the system is determined to be in a dynamic operating condition, such as when a sudden load change causes fluctuations in the bus voltage, the ripple error increases significantly. In this situation, during a brief dynamic process, the load power estimate P is used to... load Rapid correction, actively adjusting the amplitude of the compensation ripple in the feedback path, and reconstructing the DC feedback voltage V of the voltage loop. dcf This indirectly shapes the input error of the voltage loop, prompting the controller to adjust rapidly and achieve rapid bus voltage tracking. Its implementation logic is as follows: Figure 5 As shown.

[0054] Load power estimate P load The initial value is determined by the grid power P during the system startup phase. ref Determined, i.e., P load (0) = P ref .

[0055] Core mechanism: By modifying P load Indirectly shape the voltage loop feedback signal.

[0056] Reconstructing DC feedback voltage: (14) When dynamic operating conditions occur, P load The current estimate deviates from the actual load power, causing ΔV to fail to accurately compensate for V. dc The actual ripple in V makes dcf Residual ripple is mixed in. This invention corrects P quickly. load By changing the magnitude of ΔV, V can be actively adjusted. dcf This allows the voltage loop to sense the amplified DC error, thereby accelerating the adjustment process.

[0057] Error definition and directional correction logic: During the dynamic process, C is considered constant, and ΔV and P load They are inversely proportional. Based on this, the direction of correction is determined: when V... dc_ref -V dcf >0: Indicates that the current P load The estimated value is too small, resulting in an excessively large ΔV. In the feedback path, due to the excessive compensation, the feedback V of the voltage loop is... dcf The value is too small, resulting in a voltage loop input error V. dc_ref -V dcf Too large, it's prone to overtuning. P needs to be increased. load This reduces ΔV, thereby raising V. dcf To appropriately reduce errors and maintain the stability of adjustment.

[0058] When V dc_ref -V dcf <0: indicates that P is currently in the P state. load The estimated value is too high, resulting in a lower ΔV. In the feedback path, due to insufficient compensation, the feedback V of the voltage loop is... dcf The error is too large, causing the voltage loop input error V to be too high. dc_ref -V dcf The value is too small, resulting in insufficient adjustment. To accelerate the response, P needs to be reduced. load This increases ΔV, thereby decreasing V. dcf This amplifies the voltage loop error, forcing the PI controller to output a larger current reference, thus rapidly increasing the bus voltage.

[0059] Example 5 A fast iterative approach with the control period as the step size is used to process P. load Real-time updates: (12) = (13) Among them, P load (k+1) represents the estimated load power for the next control cycle, P load (k) represents the estimated load power for the current control cycle, ΔP. load k represents the load power increment. p This is the load power correction factor. The load power is iteratively updated in real time using this formula, resulting in a residual V. dc_ref -V dcf The voltage converges to zero, thereby achieving the goal of correcting DC bus voltage ripple and realizing rapid voltage tracking under dynamic operating conditions.

[0060] According to the power relationship P=V dc I, The maximum allowable current change within a single control cycle is limited by hardware current limiting. limit With control period T s The maximum allowable power fluctuation per cycle is determined jointly by: P max =V dc_ref I limit T s Combined with the maximum residual voltage e of the bus during load abrupt change v,max , to obtain k p Theoretically, the overall hardware current limiting can reach tens of amperes, but T s At the microsecond level, via I limit T s After scaling, the allowable power correction amplitude per cycle is very small, therefore kp Only for ( The current fluctuations generated in a single iteration are on the order of magnitude () and are far below the current limiting threshold of the device.

[0061] Dynamic P load Quick fix, P load 4kW, k p =5× 10μs iteration, 50% to 100% sudden increase: P load Iterated from 10 million to 20 million, 120 iterations, converged in 1.2ms, and the maximum correction per step was 50 million.

[0062] Example 6 The compensation amount f(C, P) load ) and the original sampling voltage V dc By superimposing the values, the DC feedback voltage V that filters out the ripple is obtained. dcf The V dcf Replace the original V dc The signal is fed into the outer loop PI regulator to achieve ripple feedforward compensation. The control block diagram is as follows: Figure 6 As shown.

[0063] (15) This invention achieves root cause suppression and stable filtering of bus second harmonic ripple by time-division discrimination of steady state and dynamic state, thereby improving the accuracy of steady state ripple suppression while ensuring dynamic response speed.

Claims

1. A method for suppressing DC bus ripple in a totem-pole power factor correction circuit, characterized in that, Specifically, the following steps are included: S1. Given voltage reference value V dc_ref Collect DC bus voltage sampling value V dc Obtain the actual steady-state ripple And construct a ripple estimation model ΔV; S2. Based on the DC bus voltage sampling value V dc Actual steady-state ripple Determine the peak-to-peak value of the second harmonic ripple. and ripple error According to the second harmonic ripple peak-to-peak value Set a discrimination threshold ε, and determine the system operating condition based on the discrimination threshold ε: when When ≤ε, the system is considered to be in steady state. When the value is greater than ε, the system is determined to be in a dynamic operating condition. S3. Estimated equivalent capacitance C under steady-state conditions or load power under dynamic conditions based on the ripple estimation model ΔV. Correction, compensation amount obtained after correction ; S4. The compensation amount The DC bus voltage sampling value V dc The DC feedback voltage V is obtained by superposition. dcf Using the DC feedback voltage V dcf Replace the DC bus voltage sampling value V dc To achieve ripple feedback.

2. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 1, characterized in that, The DC bus voltage is sampled in real time by a voltage dividing and conditioning circuit, and the DC bus voltage sampling value V is obtained through an analog-digital converter dc The actual steady-state ripple is calculated in the following manner: (1) Based on the power balance principle and the charge integral characteristics of equivalent capacitance, a ripple estimation model ΔV is constructed: (2) (3) (4) wherein, is the grid power, is the load power estimate, is the grid voltage effective value, is the current magnitude of the voltage loop output; is the load current.

3. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 1, characterized in that, The peak-to-peak value of the second harmonic ripple It can be represented as: (5) where f is the grid frequency, C is the equivalent capacitance of the DC bus, V dc_ave is the average value of the DC bus voltage; The ripple error Can be expressed as: (6) The discrimination threshold ε can be expressed as: ε = k (7) Where k is the proportionality coefficient, and the value of k ranges from 0.1 to 0.

3.

4. The method of claim 2, wherein, Under the aforementioned steady-state operating condition, the equivalent capacitance C undergoes minute-level iterative calibration. The specific process includes: S11. Defining instantaneous ripple error : (8) S12. Using the benchmark fixed split-body model, obtain the equivalent capacitance actually participating in the model calculation : (9) in, The equivalent capacitance is the factory-specified nominal value and remains fixed during the control cycle. This is the equivalent capacitance deviation correction amount; for the aforementioned Set a limit protection to prevent excessive correction of deviation due to abnormal operating conditions.

5. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 4, characterized in that, The instantaneous ripple error e is subjected to mean filtering to obtain the smoothed channel error e. slow The mean filtering is implemented using a sliding window averaging method, with the sliding window duration being 30–60 seconds; the channel error e slow It can be represented as: (10) According to the channel error The equivalent capacitance deviation correction amount Make corrections: (11) wherein, is the equivalent capacitance correction ratio, which is adjusted in real time according to the nominal value of the equivalent capacitance, and the single correction step is not more than 1% of the nominal value of the equivalent capacitance; the correction period is minute level.

6. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 5, characterized in that, when When >0, Then reduce the equivalent capacitance. ;when When <0, Then increase the equivalent capacitance. ; by the equivalent capacitance To enable real-time updates It converges to zero.

7. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 4, characterized in that, Under the aforementioned dynamic operating conditions, the estimated load power is calculated using the control cycle as the step size. Make corrections: (12) (13) (14) in, (k+1) represents the estimated load power for the next control cycle. (k) represents the estimated load power for the current control cycle. For load power increment, This is the load power correction factor. The range of values ​​is This is the DC feedback voltage.

8. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 7, characterized in that, when - If the value is greater than 0, the compensation amount is too large, so the estimated load power value should be increased. ;when - If the value is less than 0, the compensation is insufficient, so the estimated load power value should be reduced. ; By estimating the load power To enable real-time updates It converges to zero.

9. The DC bus ripple suppression method for the totem-pole power factor correction circuit according to claim 7, characterized in that, The compensation amount The compensation amount is matched with the real-time AC ripple with high precision. It can be represented as: (15) Among them, the estimated load power The initial value is determined by the grid power during the system startup phase. Determined, that is .