Consistency adaptive control device and method for modular power converter

Abstract

The application discloses a consistency adaptive control device and method of a modular power converter, and relates to the technical field of power electronics control. The device comprises an aging evaluation unit, a consistency distribution unit, a calibration unit and a correction record unit; the aging evaluation unit estimates the health state quantity of each power module, and the consistency distribution unit distributes the power task according to the health state quantity; the calibration unit collects the measured electrical parameters and converts the measured health state quantity to compare them according to the calibration period, and corrects the health state quantity and the distribution weight when the deviation exceeds the threshold value; the correction record unit records the correction record, and rolls back according to the correction record when the operation index after correction deteriorates to exceed the set degree, and no longer corrects in the current period, and retries in the same direction with a halved step length in the next calibration period, and marks an abnormal module and reduces the upper limit of the distribution weight when the rollback is triggered for two times in succession. The application solves the problems of open-loop operation of life-weighted distribution, estimation error accumulation and error correction that cannot be rolled back, and is suitable for multi-module power conversion occasions such as solid-state transformers.

Description

Technical Field

[0001] This invention relates to the field of power electronic control technology, and specifically to a consistency adaptive control device and method for a modular power converter. Background Technology

[0002] Modular power conversion equipment, represented by solid-state transformers (SST), consists of multiple power modules cascaded or connected in parallel. Due to manufacturing variations, differences in heat dissipation conditions, and varying operating times, the power devices in each power module have inconsistent aging rates. If the power load is shared equally among all modules, the modules that age faster will deteriorate more rapidly, creating a vicious cycle that ultimately leads to overstress failure of individual modules.

[0003] Several published documents describe the approach of weighted power allocation based on the aging state of each module (i.e., lifetime-weighted allocation). International patent application WO2009007459A2 (Cherokee, filed in 2007) discloses a scheme for allocating load according to the aging state of power modules; US patent US10128684B2 (VPS, filed in 2013) discloses a power system that adjusts the output of each module according to its health state; Japanese patent JP7258132B2 (Mitsubishi Electric, filed in 2019) discloses a scheme for allocating drive to each module in a laser driver power supply according to its degradation level; and Chinese patent application CN122246856A discloses a method for operating and controlling modular power electronic equipment under all operating conditions, using junction temperature closed-loop prediction and... Lifetime-weighted allocation is used for module-level scheduling; Chinese patent application CN118899901A discloses a scheme for determining module priority and calling based on module lifetime assessment results; Chinese patent application CN121530128A discloses a power allocation method for a multi-module converter, which sets an average efficiency observer based on a consensus algorithm and adjusts the phase shift angle according to the efficiency observation value of each sub-module to achieve power allocation. The "consistency" referred to in that case refers to a collaborative observation mechanism in which the efficiency of each module converges to the average efficiency. The allocation goal is to achieve optimal efficiency, and it does not involve aging assessment, calibration closed loop, and correction record rollback, which is different from the consistency control of power allocation tasks based on aging state in this application.

[0004] The aforementioned literature all discloses the general concept of "weighted allocation based on aging / lifetime status"; other published literature involves online identification of aging parameters of power devices and online correction of thermal model parameters. However, none of the existing technologies address: in the scenario of weighted allocation of power based on aging, periodically calibrating the aging assessment value using the measured electrical parameters of the module and directly linking the calibration results to correct the allocation weights; furthermore, they do not address recording and monitoring the calibration correction action itself and rolling it back when the operating indicators deteriorate after correction. The existing solutions lack a closed-loop linkage between the calibration process and the allocation weights; the correction action is unrecorded, unsupervised, and unrolled. Once a correction is performed in the wrong direction (e.g., due to momentary sensor failure or model mismatch leading to misjudgment), the incorrect allocation weights will be solidified or even amplified, which is the root cause of inaccurate allocation.

[0005] In summary, the common shortcomings of existing technologies are: the lack of closed-loop linkage between calibration and weight allocation in lifetime-weighted allocation scenarios, and the lack of recording, monitoring, and rollback mechanisms for the correction actions themselves.

[0006] On the same filing date as this application, the applicant separately filed two invention patent applications: "A Real-Time Cooperative Control Device and Method for a Multi-Physics Coupled Power Converter" and "A Feedforward Pre-adjustment Device and Method for a Power Converter Based on Load Prediction." This application relates to consistent adaptive control of modular power converters; the two applications filed on the same date respectively relate to real-time closed-loop cooperative control of multi-physics coupled power converters across time scales and feedforward pre-adjustment based on load prediction. Summary of the Invention

[0007] To address the issues of lack of calibration-allocation closed-loop linkage and inability to roll back error correction in existing lifetime weighted allocation technologies, this invention provides a consistency adaptive control device and method for modular power converters.

[0008] The present invention adopts the following technical solution: A consistency adaptive control device for a modular power converter, the modular power converter comprising N power modules, where N is an integer not less than 2, the device comprising: an aging assessment unit for estimating the health status of each power module; a consistency allocation unit for weighted allocation of power tasks to each power module according to the health status; a calibration unit for collecting measured electrical parameters of each power module at calibration cycles, comparing the measured health status determined based on the measured electrical parameters with the health status estimated by the aging assessment unit, and correcting the allocation weights of the health status and the consistency allocation unit when the deviation exceeds a set threshold; and a correction recording unit for recording correction records for each correction, the correction record including at least the health status before and after correction, the allocation weights before and after correction, the triggering reason, and a timestamp; the correction recording unit also, within a set observation window after correction, when the operating indicators deteriorate beyond a set level, rollback the health status and the allocation weights to the state before correction according to the correction record, after rollback, no further correction is performed in the current calibration cycle, and retrying in the next calibration cycle with a halved correction step size in the same direction.

[0009] Furthermore, the calibration unit statistically analyzes the deviation sequence of multiple consecutive calibration cycles: when the deviation is consistently less than the lower threshold, the calibration cycle is extended; when the frequency of deviation exceeding the upper threshold exceeds a set number, the calibration cycle is shortened.

[0010] Furthermore, the measured electrical parameters include at least two of the following: on-state voltage drop or on-state resistance, switching delay, and junction temperature of the power devices in each power module.

[0011] Furthermore, the health status quantity is an estimated remaining useful life or a normalized aging coefficient; the consistency allocation unit determines the allocation weight of each power module according to W_i = H_i / ΣH_j.

[0012] Furthermore, the operating indicators include at least one of the following: bus voltage ripple, tracking deviation of the actual output power of each power module from the allocated power task, and peak junction temperature of the power device; the determination of deterioration exceeding the set level is as follows: the increase of the tracking deviation exceeds 2 percentage points, or the peak junction temperature of the power device rises by more than 5°C, or the increase of the bus voltage ripple exceeds 20%. If any of these conditions are met, the deterioration is determined to exceed the set level.

[0013] Furthermore, the calibration period is determined based on the aging drift rate of the power device and the noise level of the measured electrical parameters, and the calibration period is longer than the control period of the modular power converter and shorter than the time when the aging state of the power device changes significantly; the initial value of the calibration period is 1 hour to 168 hours.

[0014] Furthermore, the calibration unit is also used to verify the tracking deviation between the actual output power of each power module and the power task assigned by the consistency allocation unit, and to trigger a temporary shortening of the calibration cycle when the tracking deviation exceeds a set threshold.

[0015] A consistency adaptive control method for a modular power converter includes the following steps: S201: estimating the health status of each power module; S202: weighting the power tasks of each power module according to the health status; S203: collecting measured electrical parameters according to the calibration cycle, and comparing the measured health status determined based on the measured electrical parameters with the estimated health status; S204: when the deviation exceeds a threshold, correcting the health status and the assigned weights and writing it into the correction record; S205: monitoring the corrected operating indicators, and when the deterioration exceeds a set level, rolling back to the state before correction according to the correction record. After rolling back, no further correction is performed in the current calibration cycle, and the correction is retried in the next calibration cycle with a halved correction step size along the same direction.

[0016] The beneficial effects of this invention are: (1) The calibration unit adds a closed loop to the aging assessment: the calibration results directly link to the correction of the allocation weights, and the original open-loop weighted allocation of lifespan is upgraded to closed-loop adaptive allocation, thus eliminating the long-term accumulation of assessment errors; (2) The correction record unit makes each correction traceable and rollback: after rollback, the current cycle is frozen, the next cycle is halved and retried, and if rollback is triggered twice in a row, the abnormal module is marked and its allocation weight limit is reduced, forming a tiered fault tolerance protocol. Error correction is no longer solidified or amplified, and the robustness of the allocation mechanism is improved accordingly. (3) The calibration closed loop suppresses the long-term accumulation of evaluation error. Under the specified test conditions, the tracking deviation of the actual output power of each power module to the power allocation task is expected to be controlled within 5%. The remaining life of each module with different aging degrees tends to be consistent, and the life of the whole machine is extended. (4) All of the above effects can be measured and reproduced from the outside in product-level bench testing without obtaining the source code of the control program. Attached Figure Description

[0017] Figure 1 This is a block diagram of the consistency adaptive control device according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the calibration closed-loop process according to an embodiment of the present invention. Figure 3 This is a flowchart of the consistency adaptive control method according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the tiered fault-tolerant timing of an embodiment of the present invention; Figure 5This is a flowchart illustrating the adaptive adjustment of the calibration cycle according to an embodiment of the present invention. Figure 6 This is a system structure diagram of the application of the present invention in a solid-state transformer according to Embodiment 1; Figure 7 This is a schematic diagram illustrating the principle of tracking deviation trend of actual power distribution commands compared to the present invention and the uncalibrated scheme.

[0018] Explanation of markings in the diagram: 100-Consistency Adaptive Control Device; 110 - Aging Assessment Unit; 120 - Consistent Allocation Unit; 130 - Calibration Unit; 140 - Correction Recording Unit; 200-Power Module Group; 210-1 to 210-N power modules; 300-load. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that the specific embodiments described are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0020] [Device Architecture] as follows Figure 1 As shown, the consistency adaptive control device 100 includes an aging assessment unit 110, a consistency allocation unit 120, a calibration unit 130, and a correction recording unit 140. The aging assessment unit 110 estimates the health status of each power module 210-1 to 210-N in the power module group 200; the consistency allocation unit 120 allocates the power tasks of each module according to the health status; the calibration unit 130 collects the measured electrical parameters of each module according to the calibration cycle, converts the measured health status into a measured health status and compares it with the estimated health status; when the deviation exceeds the threshold, the health status and allocation weights are corrected; the correction recording unit 140 records the correction record for each correction and rolls back according to the correction record when the operating indicators deteriorate after the correction.

[0021] [Health Status Quantity and Weighted Allocation] The health status quantity H_i (i = 1, ..., N) is the estimated remaining useful life or normalized aging coefficient, ranging from 0 to 1, where 1 represents a brand new state. The aging assessment unit 110 adopts an aging model based on Miner's linear cumulative damage law: the junction temperature cycle amplitude ΔT_i and the cycle number n_i of each level of junction temperature cycle are statistically analyzed using the rainflow counting method. The cumulative damage D = Σ(n_i / N_i), where N_i is the number of failure cycles given by the device datasheet lifetime curve under the corresponding junction temperature cycle amplitude ΔT_i, obtained by fitting the lifetime curve; the health status quantity H = 1−D, and the model parameters are updated online with the operating data. The consistency allocation unit 120 determines the allocation weight of each module according to W_i = H_i / ΣH_j, so that the module with the lower health status quantity shares a smaller power task, and the sum of the power tasks of each module equals the total load demand. The health status values ​​of each module output by the aging assessment unit 110 are also output to the multi-physics field collaborative control system via the signal interface. This serves as the input for the collaborative control system to adjust the weights of the junction temperature target and efficiency target, the switching frequency, and the coolant flow rate, thereby realizing the linkage between the aging assessment results and the multi-physics field collaborative control.

[0022] [Calibration Closed Loop] such as Figure 2 As shown, calibration unit 130 triggers calibration according to calibration cycle T_c: it collects at least two of the measured electrical parameters of each module—the on-state voltage drop or on-resistance of the power device, switching delay, and junction temperature; it converts the measured electrical parameters into measured health status quantities through aging characteristic mapping, and compares them with the estimated health status quantities output by aging assessment unit 110 to obtain the deviation Δ; when |Δ| exceeds the set threshold, it corrects the evaluation value of aging assessment unit 110 with the measured health status quantity, and simultaneously updates the allocation weight of consistency allocation unit 120. On-state voltage drop and on-resistance monotonically increase with device aging, and switching delay drifts with aging. These are all recognized aging-sensitive electrical parameters in the field and can be measured online without disassembling the module. The quantitative relationship of aging characteristic mapping is as follows: the rate of increase of on-state voltage drop relative to the new product reference value is the main mapping quantity. A rate of increase within 5% is mapped to the healthy period, 5% to 15% is mapped to the mid-aging period, and above 15% is mapped to the late-aging period. The mapping relationship is derived from the device datasheet and accelerated aging test calibration data, and is implemented using piecewise linear lookup table. When multiple measured electrical parameters are used, the measured health status quantity is obtained by weighting and fusing them with a weight of 0.6 for on-state voltage drop, 0.3 for switching delay, and 0.1 for junction temperature. The selection of the weighting coefficients is based on the correlation coefficients between the rate of change of on-state voltage drop, switching delay, junction temperature and cumulative damage obtained through accelerated aging tests, and the correlation coefficients are normalized before determination.

[0023] [Calibration Cycle Setting Basis] The calibration cycle T_c is determined based on the following: T_c must be much longer than the converter's control cycle (microseconds to milliseconds) to avoid calibration actions interfering with normal control; T_c must be much shorter than the time it takes for the power devices to undergo significant changes in their aging state (on the order of months) to avoid excessive accumulation of deviations within the calibration interval; the noise level of the measured electrical parameters is also considered, as higher noise levels result in lower reliability of a single calibration, thus a longer calibration cycle is preferable for filtering and averaging. Taking all these factors into account, the initial value of the calibration cycle ranges from 1 hour to 168 hours, with a typical value of 24 hours.

[0024] [Adaptive Adjustment of Calibration Cycle and Verification of Tracking Deviation] (e.g.) Figure 5 As shown, calibration unit 130 statistically analyzes the deviation sequence over K consecutive calibration cycles: when the deviation is consistently below the lower threshold, it indicates slow aging drift or good model fit, and the calibration cycle is extended (by a factor not exceeding 2) to reduce measurement and computational overhead; when the frequency of deviations exceeding the upper threshold exceeds a set number, it indicates accelerated aging or model mismatch, and the calibration cycle is shortened (by a factor not less than 0.5) to improve tracking capability. The adjustment range of the calibration cycle is 0.1 to 10 times the initial value. Calibration unit 130 also continuously verifies the tracking deviation between the actual output power of each power module and the power task allocated by consistency allocation unit 120. When the tracking deviation exceeds a set threshold (typically 5%), it indicates an anomaly in the allocation execution process, triggering a temporary shortening of the calibration cycle, and prioritizing the investigation of allocation execution anomalies.

[0025] [Revision Records and Tiered Fault Tolerance Protocol] For example Figure 4As shown, the correction recording unit 140 writes a correction record for each correction, the content of which includes at least: the health status before and after the correction, the allocation weight before and after the correction, the triggering reason (the specific parameters and values ​​of the deviation exceeding the threshold), and a timestamp; the correction records are stored in a ring, with a capacity of not less than 1000 records, and are stored on a non-volatile storage medium, so that the records are not lost after the device is powered off. After the correction is executed, the correction recording unit 140 continuously monitors at least one of the following operating indicators within the set observation window: bus voltage ripple, tracking deviation of the actual output power of each module from the allocated power task, and peak junction temperature of the power device; the quantitative judgment standard for deterioration exceeding the set degree is: the increase of the tracking deviation within the observation window exceeds 2 percentage points, or the peak junction temperature of the power device rises by more than 5°C, or the increase of the bus voltage ripple exceeds 20%, and any condition is met to determine that the deterioration exceeds the set degree (or logic, to avoid ambiguity in judgment when the direction of change of multiple indicators is inconsistent); the typical value of the observation window is 1 hour. When the operating indicators deteriorate beyond the set level, it is determined that the correction direction or step size is inappropriate. The health status quantity and the allocated weight are rolled back to the state before the correction according to the correction record. After the rollback, the correction will not be performed again in the current calibration cycle. In the next calibration cycle, the correction step size will be halved and the same direction will be retried. If the rollback is triggered twice in a row, the corresponding power module will be marked as an abnormal module, its allocation weight limit will be reduced, and an operation and maintenance prompt signal will be output.

[0026] [Method and Flow] as follows Figure 3 As shown, the consistency adaptive control method includes: S201, estimating the health status of each power module; S202, weighting the power tasks of each module according to the health status; S203, collecting measured electrical parameters according to the calibration cycle, converting the measured health status and comparing them; S204, when the deviation exceeds the threshold, correcting the health status and the assigned weight and writing it into the correction record; S205, monitoring the corrected operating indicators, and when the deterioration exceeds the set level, rolling back according to the correction record, no further correction in this cycle, and retrying in the same direction with half the step size in the next cycle.

[0027] [Example 1: Solid-State Transformer in AI Computing Center] For example... Figure 6As shown, the solid-state transformer input is a 10kV medium-voltage power grid. Each phase consists of three cascaded power modules 210-1 to 210-3, and the output 750V DC bus supplies power to a computing load of 300. The rated power of the entire unit is 1MW. The aging assessment unit 110 estimates the health status of each module using a cumulative damage model, with sampled inputs including module current, junction temperature cycle, and operating time. The consistency allocation unit 120 adjusts the modulation ratio of each module according to weights to achieve power allocation. The calibration unit 130 collects the on-state voltage drop and junction temperature of the power devices in each module every 24 hours, converts the measured health status quantities, and compares and corrects them. The correction recording unit 140 uses ring storage with a capacity of 1000 records, stored on a non-volatile storage medium, and the observation window is 1 hour after correction. Parameter selection criteria: the initial 24-hour calibration cycle is between the control cycle (microsecond level) and the time of significant aging changes (monthly level); the 1-hour observation window covers the complete statistical process of bus voltage ripple and junction temperature peak.

[0028] The [Reproducible Measurement Protocol] specifies the following test conditions: At least one module with an aging degree difference of no less than 20% is artificially set in each power module (equivalent to modules of the same model that have been running for different durations); the load is 50% to 100% of the rated power; the ambient temperature is 25℃±2℃; the power measurement point is the output terminal of each power module; the power measurement accuracy is no less than ±0.5%; and the statistical duration is no less than 1 hour. According to the above protocol, those skilled in the art can externally measure and reproduce the tracking accuracy of the calibration closed loop on a product-level test bench; based on the mechanism of closed-loop calibration and tiered fault tolerance, the tracking deviation of the actual output power of each power module from the allocated power task is expected to be controlled within 5%; however, if the calibration unit and correction recording unit are disconnected, and only the open-loop lifespan weighted allocation is retained on the same test bench, the tracking deviation will gradually increase with the accumulation of evaluation errors (e.g., ...). Figure 7 (The qualitative trend shown).

[0029] [Example 2: Electric Vehicle Supercharging Equipment] The device of the present invention is applied to a 350kW to 1000kW electric vehicle supercharging solid-state transformer, wherein the high-voltage side consists of multiple cascaded power modules and the low-voltage side consists of multiple power modules connected in parallel. The differences from Example 1 are: the measured electrical parameters are the on-resistance and switching delay; the initial value of the calibration cycle is 12 hours; and the observation window is 30 minutes after correction. Its aging assessment, calibration closed loop, correction record, and tiered fault tolerance protocol are the same as in Example 1, and will not be repeated.

[0030] [Example 3: Aviation Power System] The device of the present invention is applied to the parallel power generation channel of an aviation 270V DC power system: the aging assessment unit estimates the health status of each power generation channel, the consistency allocation unit allocates the output of each channel according to the health status, the calibration unit collects the measured electrical parameters of each channel according to the flight take-off and landing cycle to perform calibration, and the correction recording unit records the correction and executes the tiered fault tolerance protocol. The principles for determining the number of modules, calibration cycle, and observation window are the same as in Example 1, and the specific values ​​are determined by those skilled in the art according to the platform's operating conditions.

[0031] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A consistency adaptive control device for a modular power converter, characterized in that, The modular power converter includes N power modules, where N is an integer not less than 2, and the device includes: An aging assessment unit is used to estimate the health status of each power module. A consistency allocation unit, electrically connected to the aging assessment unit, is used to weighted allocate the power tasks of each power module according to the health status quantity; The calibration unit, electrically connected to the aging assessment unit, is used to collect the measured electrical parameters of each power module according to the calibration cycle, compare the measured health status quantity determined based on the measured electrical parameters with the health status quantity estimated by the aging assessment unit, and correct the allocation weight of the health status quantity and the consistency allocation unit when the deviation exceeds a set threshold. The correction recording unit, electrically connected to the calibration unit, is used to record the correction record for each correction. The correction record includes at least the health status quantity before and after the correction, the allocation weight before and after the correction, the triggering reason, and the timestamp. Within the set observation window after the correction, when the operating index deteriorates beyond a set level, the correction recording unit rolls back the health status quantity and the allocation weight to the state before the correction according to the correction record. After the rollback, the correction will not be performed again in the current calibration cycle, and it will be retried in the next calibration cycle with a correction step size of half in the same direction.

2. The consistency adaptive control device for the modular power converter according to claim 1, characterized in that, The calibration unit statistically analyzes the deviation sequence of multiple consecutive calibration cycles: when the deviation is consistently less than the lower threshold, the calibration cycle is extended; when the frequency of deviations exceeding the upper threshold exceeds a set number, the calibration cycle is shortened.

3. The consistency adaptive control device for the modular power converter according to claim 1, characterized in that, The measured electrical parameters include at least two of the following: on-state voltage drop or on-state resistance of the power devices in each power module, switching delay, and junction temperature.

4. The consistency adaptive control device for the modular power converter according to claim 1, characterized in that, The health status quantity is an estimated remaining useful life or a normalized aging coefficient; the consistency allocation unit determines the allocation weight of each power module according to W_i = H_i / ΣH_j, where H_i is the health status quantity of the i-th power module, W_i is the allocation weight of the i-th power module, and ΣH_j is the sum of the health status quantities of all N power modules.

5. The consistency adaptive control device for a modular power converter according to claim 1, characterized in that, The operating indicators include at least one of the following: bus voltage ripple, tracking deviation of the actual output power of each power module from the allocated power task, and peak junction temperature of the power device; the determination that the operating indicators deteriorate beyond the set level within the set observation window is as follows: the increase in tracking deviation exceeds 2 percentage points, or the peak junction temperature of the power device rises by more than 5°C, or the increase in bus voltage ripple exceeds 20%. If any of these conditions are met, the deterioration is determined to exceed the set level.

6. The consistency adaptive control device for the modular power converter according to claim 1, characterized in that, The calibration period is determined based on the aging drift rate of the power device and the noise level of the measured electrical parameters. The calibration period is longer than the control period of the modular power converter and shorter than the time when the aging state of the power device changes significantly. The initial value of the calibration period is 1 hour to 168 hours.

7. The consistency adaptive control device for a modular power converter according to claim 1, characterized in that, The calibration unit is also used to verify the tracking deviation between the actual output power of each power module and the power task assigned by the consistency allocation unit, and to trigger a temporary shortening of the calibration cycle when the tracking deviation exceeds a set threshold.

8. A consistency adaptive control method for a modular power converter, characterized in that, Includes the following steps: S201: Estimate the health status parameters of each power module in the modular power converter; S202: Allocate power tasks to each power module according to the weighted health status quantity; S203: Collect the measured electrical parameters of each power module according to the calibration cycle, and compare the measured health status quantity determined based on the measured electrical parameters with the estimated health status quantity; S204: When the deviation exceeds the set threshold, correct the health status quantity and the assigned weight, and write the correction record. The correction record includes at least the health status quantity before and after correction, the assigned weight before and after correction, the triggering reason, and the timestamp. S205: Monitor the corrected operating indicators. When the operating indicators deteriorate beyond a set level, roll back the health status quantity and the allocated weight to the state before the correction according to the correction record. After the rollback, no further correction will be performed in the current calibration cycle, and the correction will be retried in the same direction with a halved correction step size in the next calibration cycle.

9. The consistency adaptive control method for a modular power converter according to claim 8, characterized in that, Step S203 includes: statistically analyzing the deviation sequence of multiple consecutive calibration cycles; extending the calibration cycle when the deviation is consistently less than the lower threshold; and shortening the calibration cycle when the frequency of deviations exceeding the upper threshold exceeds a set number of times.

10. The consistency adaptive control method for a modular power converter according to claim 8, characterized in that, Step S205 includes: when rollback is triggered in two consecutive retries, the corresponding power module is marked as an abnormal module, its allocation weight limit is reduced, and an operation and maintenance prompt signal is output.

Images