GaN-based power supply full-bridge resonant energy efficiency optimization method and system based on edge computing

Abstract

This invention provides a method and system for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing. It uses a fixed control cycle to periodically sample and calculate instantaneous energy efficiency information, continuously updating the energy efficiency queue. The average energy efficiency of the queue is compared with the previously optimized benchmark energy efficiency for initial deviation. Only when the deviation exceeds a trigger threshold is continuous multi-cycle verification initiated to determine the energy efficiency trend direction and amplitude. The adjustment direction is confirmed only when the trend amplitude exceeds a trend threshold, and then compared with the locked benchmark value. A single-step adjustment amount is output only when the deviation exceeds a locked threshold. After adjustment, resampling and verification are performed. If the energy efficiency deviation decreases, the benchmark and locked values ​​are updated; otherwise, the process regresses and outputs the adjustment amount in reverse until convergence. This invention effectively filters out false triggers caused by noise and transient disturbances, while reducing communication latency through edge-local closed-loop execution, thus improving the energy efficiency optimization stability of the full-bridge resonant converter under all operating conditions.

Description

Technical Field

[0001] This invention relates to the fields of power conversion technology and edge computing control, and more specifically, to a method and system for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing. Background Technology

[0002] With the increasing demands for power density and conversion efficiency in applications such as communication base stations, data centers, and on-board charging of electric vehicles, full-bridge resonant converters based on gallium nitride power devices have become the mainstream topology in the field of high-efficiency power conversion due to their high-frequency switching capability and soft-switching characteristics.

[0003] However, in actual engineering operation, the energy efficiency of the full-bridge resonant converter is affected by the dynamic coupling of multiple factors: random sudden changes in load current, wide range fluctuations in input voltage, real-time drift of junction temperature of power devices, and parameter degradation of resonant capacitors and inductors as they age. All of these factors can cause the actual operating point of the resonant cavity to deviate from the optimal frequency point determined in the design stage, resulting in increased cycle energy, intensified hard switching of switching transistors, and increased losses of magnetic components, leading to a significant deterioration in overall energy efficiency.

[0004] To address the aforementioned energy efficiency degradation issues, traditional optimization schemes often employ a centralized architecture, uploading sampled data to a central processing unit or cloud for parameter scanning and optimal frequency calculation. This approach relies on reliable communication links, inherently suffers from transmission delays, and cannot meet the microsecond-level response speed requirements of high-frequency converters, failing completely in the event of network anomalies. Secondly, while some localized solutions attempt to adjust via preset energy efficiency tables, such open-loop control lacks the ability to perceive changes in actual operating conditions. When resonant cavity parameters shift due to temperature or aging, the fixed tables become inapplicable. Thirdly, existing schemes generally use a single absolute value threshold comparison logic to determine whether to initiate optimization. This method is sensitive to sampling noise and transient load disturbances, easily triggering frequent erroneous adjustments during steady-state operation, introducing additional switching losses and reducing system reliability. Finally, existing decision-making processes lack a closed-loop verification mechanism for the adjustment direction. If the initial adjustment direction is incorrect, multiple trials are required for correction, resulting in slow convergence and oscillations near the optimal operating point, failing to quickly and stably converge to the true energy-optimal operating point. Summary of the Invention

[0005] In view of the above problems, the purpose of this invention is to provide a GaN-based power supply full-bridge resonant energy efficiency optimization method and system based on edge computing. By deploying the edge controller on the power module side to independently execute the complete process of sampling, queue updating and threshold comparison, the communication dependence on the cloud or host computer is eliminated, and local closed-loop autonomous energy efficiency optimization is achieved. Specifically: First, by constructing a constant-length energy efficiency queue and updating it in a rolling manner, the energy efficiency information window on which the decision is based is ensured to have a stable time span. Then, after the initial trigger, by continuing to perform verification for a preset number of cycles and calculating the trend amplitude and trend direction, occasional sampling noise and load transient disturbances are distinguished from the actual continuous energy efficiency drift. Next, by setting a direction threshold, the trend amplitude is screened a second time, and the adjustment direction is determined only when the trend is sufficiently significant, further ensuring that the determination of the adjustment direction is based on substantial energy efficiency changes. Then, by starting a waiting timer after the adjustment command is output to pause all sampling comparison operations, the subsequent energy efficiency response results are ensured to be obtained after the converter enters a new steady state. Finally, by performing a closed-loop comparison between the adjusted energy efficiency deviation and the pre-adjustment deviation and setting a reverse backoff mechanism when the direction is incorrect, the optimization process is ensured to always converge towards the direction of energy efficiency improvement. At the same time, after confirming that the direction is correct, a dual update strategy of benchmark energy efficiency and locking the benchmark value is adopted to ensure the independence and accuracy of each level of criteria.

[0006] The first aspect of this invention provides a method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing, the method comprising: Based on a preset control cycle, instantaneous energy efficiency information is sampled and calculated to update the energy efficiency queue; Based on the updated energy efficiency queue, the average energy efficiency is calculated, and the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action is obtained. If the first energy efficiency deviation exceeds the preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude. If the trend amplitude exceeds the preset trend threshold, the adjustment direction is determined according to the trend direction; Calculate the average energy efficiency of the latest energy efficiency queue and compare it with the locked baseline value after the last optimization action to obtain the second energy efficiency deviation; If the second energy efficiency deviation exceeds the preset locking threshold, then a single-step adjustment amount is output in the adjustment direction and the adjustment is executed; Determine whether the first energy efficiency deviation after adjustment has been reduced; If so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; If not, then roll back and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

[0007] In this scheme, the step of sampling and calculating instantaneous energy efficiency information based on a preset control cycle to update the energy efficiency queue specifically includes: At the arrival of each control cycle, the input voltage, input current, output voltage, and output current of the converter are sampled synchronously. The input power is obtained by multiplying the input voltage and the input current, and the output power is obtained by multiplying the output voltage and the output current. The ratio of the output power to the input power is used as instantaneous energy efficiency information; The instantaneous energy efficiency information is appended to the tail of the energy efficiency queue, and the historical energy efficiency information at the head of the energy efficiency queue is removed to maintain a constant queue length. The energy efficiency queue is used to store instantaneous energy efficiency information for multiple consecutive control cycles.

[0008] In this scheme, if the first energy efficiency deviation exceeds a preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude, specifically including: When the first energy efficiency deviation exceeds the preset trigger threshold, continue to perform sampling and queue update for a preset number of consecutive verification cycles; Calculate the current energy efficiency queue average value after each cycle is completed; After completing all verification cycles, calculate the difference between the latest energy efficiency queue average and the initial energy efficiency queue average. Use the absolute value of the difference as the trend amplitude, and determine the trend direction based on the positive or negative sign of the difference. A positive value indicates that energy efficiency is on an upward trend, and a negative value indicates that energy efficiency is on a downward trend.

[0009] In this scheme, if the trend amplitude exceeds a preset trend threshold, the adjustment direction is determined based on the trend direction, specifically including: The trend amplitude is compared with a preset direction threshold. If the trend amplitude is less than the direction threshold, then abandon this round of optimization and clear the verification count; If the trend amplitude reaches or exceeds the direction threshold, then the positive or negative sign of the trend direction is read. If the trend direction is negative, then the adjustment direction is determined to be the adjustment direction to improve energy efficiency; If the trend direction is positive, then the adjustment direction is determined to be a fine-tuning direction that maintains the current energy efficiency trend; The adjustment direction is used to indicate the direction of frequency increase or decrease of subsequent adjustment output or dead time adjustment.

[0010] In this scheme, the step of outputting a single-step adjustment amount and performing adjustment according to the adjustment direction specifically includes: Based on the adjustment direction, an adjustment command equal to the single adjustment step size is generated from the read single adjustment step size base; The adjustment command is output to the drive circuit of the full-bridge converter to control the switching frequency to change by one step unit in the adjustment direction. After the adjustment command is output, a waiting timer is started, and no new energy efficiency sampling and comparison operations are performed during the timer's countdown. After the timer expires, the sampling and energy efficiency deviation calculation are re-executed to obtain the adjusted energy efficiency response result.

[0011] In this solution, determining whether the adjusted first energy efficiency deviation has decreased specifically includes: Calculate the first energy efficiency deviation obtained after adjustment and compare it with the first energy efficiency deviation recorded before adjustment; If the first energy efficiency deviation after adjustment is less than the first energy efficiency deviation before adjustment, then the adjustment direction is confirmed to be correct. The benchmark energy efficiency is updated according to the average value of the energy efficiency queue obtained by the current sampling cycle after adjustment, and the locked benchmark value is updated according to the average value of the energy efficiency queue after adjustment and after verification by multiple consecutive control cycles and tending to be stable. If the first energy efficiency deviation after adjustment is not less than the first energy efficiency deviation before adjustment, it is determined that the adjustment direction is wrong, a reverse adjustment command is generated, and an adjustment amount of twice the step size base is output in the direction opposite to the adjustment direction, and the waiting and re-acquisition verification are executed again.

[0012] A second aspect of the present invention provides a GaN-based power supply full-bridge resonant energy efficiency optimization system based on edge computing, comprising a GaN-based power supply full-bridge resonant energy efficiency optimization method program based on edge computing, wherein the GaN-based power supply full-bridge resonant energy efficiency optimization method program based on edge computing is executed by the processor to perform the following steps: Based on a preset control cycle, instantaneous energy efficiency information is sampled and calculated to update the energy efficiency queue; Based on the updated energy efficiency queue, the average energy efficiency is calculated, and the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action is obtained. If the first energy efficiency deviation exceeds the preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude. If the trend amplitude exceeds the preset trend threshold, the adjustment direction is determined according to the trend direction; Calculate the average energy efficiency of the latest energy efficiency queue and compare it with the locked baseline value after the last optimization action to obtain the second energy efficiency deviation; If the second energy efficiency deviation exceeds the preset locking threshold, then a single-step adjustment amount is output in the adjustment direction and the adjustment is executed; Determine whether the first energy efficiency deviation after adjustment has been reduced; If so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; If not, then roll back and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

[0013] A third aspect of the present invention provides a computer-readable storage medium comprising a method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant circuit based on edge computing. When the method program is executed by a processor, it implements the steps of the method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant circuit based on edge computing as described in any of the preceding claims.

[0014] This invention provides a method and system for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing. It uses a fixed control cycle to periodically sample and calculate instantaneous energy efficiency information, continuously updating the energy efficiency queue. The average energy efficiency of the queue is compared with the previously optimized benchmark energy efficiency for initial deviation. Only when the deviation exceeds a trigger threshold is continuous multi-cycle verification initiated to determine the energy efficiency trend direction and amplitude. The adjustment direction is confirmed only when the trend amplitude exceeds a trend threshold, and then compared with the locked benchmark value. A single-step adjustment amount is output only when the deviation exceeds a locked threshold. After adjustment, resampling and verification are performed. If the energy efficiency deviation decreases, the benchmark and locked values ​​are updated; otherwise, the process regresses and outputs the adjustment amount in reverse until convergence. This invention effectively filters out false triggers caused by noise and transient disturbances, while reducing communication latency through edge-local closed-loop execution, thus improving the energy efficiency optimization stability of the full-bridge resonant converter under all operating conditions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope.

[0016] Figure 1 The flowchart of a GaN-based full-bridge resonant power supply energy efficiency optimization method based on edge computing is shown in this invention. Figure 2 A schematic diagram of the structure of a GaN-based full-bridge resonant power supply efficiency optimization circuit based on edge computing according to the present invention; Figure 3 A block diagram of a GaN-based full-bridge resonant power supply efficiency optimization system based on edge computing is shown. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Unless otherwise defined, all terms (including technical and scientific terms) used in embodiments of this invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in a common dictionary shall be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as being interpreted in an idealized or highly formalized sense, unless expressly defined in this embodiment of the invention.

[0019] The terms "first," "second," and similar words used in the embodiments of this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "an," "a," or "the" do not indicate a quantity limitation, but rather indicate the presence of at least one. Similarly, terms such as "including" or "comprising" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The steps preceding or following the steps in the method of the embodiments of this invention are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0020] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0021] Figure 1 The flowchart of a GaN-based power supply full-bridge resonant energy efficiency optimization method based on edge computing is shown.

[0022] like Figure 1 As shown, the first aspect of this invention discloses a method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing, the method comprising: S102, Based on the preset control cycle, sample and calculate instantaneous energy efficiency information to update the energy efficiency queue; S104, Calculate the average energy efficiency based on the updated energy efficiency queue, and obtain the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action; S106, if the first energy efficiency deviation exceeds the preset trigger threshold, then continuously collect instantaneous energy efficiency information of multiple cycles and update the energy efficiency queue to determine the trend direction and trend amplitude. S108, if the trend amplitude exceeds the preset trend threshold, then the adjustment direction is determined according to the trend direction; S110, calculate the average energy efficiency of the latest energy efficiency queue, and obtain the second energy efficiency deviation by comparing it with the locked baseline value after the last optimization action; S112, if the second energy efficiency deviation exceeds the preset locking threshold, then output a single-step adjustment amount in the adjustment direction and perform the adjustment; S114, determine whether the adjusted first energy efficiency deviation has been reduced; S116, if so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; S118, if not, then backtrack and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

[0023] It should be noted that, in this embodiment, during the power-on initialization phase, the edge controller reads the preset trigger threshold, direction threshold, lock threshold, queue length parameter, single adjustment step size base and maximum number of trial steps from the non-volatile memory area. At the same time, it clears the historical energy efficiency queue and fills the queue with the instantaneous energy efficiency information obtained from the first sampling calculation until it is full.

[0024] After entering the main loop, the edge controller periodically triggers the sampling process according to a preset fixed control cycle. At the arrival of each cycle, it synchronously collects the input and output voltage and current information of the converter and calculates the current instantaneous energy efficiency information accordingly. This instantaneous value is then appended to the tail of the energy efficiency queue, and the oldest value at the head of the queue is removed to maintain a constant queue length. Subsequently, the edge controller calculates the arithmetic mean of all energy efficiency values ​​in the queue as the current average energy efficiency, and performs a difference calculation with the baseline energy efficiency recorded after the last successful optimization to obtain the first energy efficiency deviation. If the absolute value of this deviation does not exceed the trigger threshold, it is determined that the current energy efficiency is still within the acceptable range, and it directly waits for the next control cycle; if it exceeds the threshold, the second-level verification is initiated, i.e., maintaining the current operating state and continuing to execute sampling and queue updates for a preset number of consecutive verification cycles. The current queue average value is calculated for each cycle. After the verification cycle is completed, the difference between the last average value and the average value at the beginning of the first verification is calculated, and its absolute value is used as the trend amplitude. The trend direction is determined based on the sign of the difference. If the trend amplitude is less than the direction threshold, it is considered an interference pulse and the current optimization is abandoned; if it reaches or exceeds the threshold, the adjustment direction is determined based on the trend direction. Subsequently, the edge controller calculates the difference between the verified final trend value and the previously maintained locked benchmark value to obtain the second energy efficiency deviation. If the absolute value of the second energy efficiency deviation does not exceed the locking threshold, it is determined that the energy efficiency is within the optimal neighborhood and no adjustment command is output, while the benchmark energy efficiency is updated to the current average value; if it exceeds the threshold, the actual deviation is confirmed and a single-step adjustment amount is output to the drive circuit along the determined direction.

[0025] After adjustment, the edge controller starts a waiting timer to allow the converter to enter a new transient equilibrium. Once the timer expires, sampling and chain comparison are re-executed, comparing the adjusted energy efficiency deviation with the deviation before adjustment to determine if the adjustment is effective. If effective, the baseline energy efficiency is updated and the baseline value is locked; if ineffective, it reverts and outputs a double-step adjustment amount for re-verification, and dynamically adjusts the thresholds at each level based on recent energy efficiency fluctuations after each successful adjustment. This embodiment independently completes the entire closed-loop optimization process at the edge, from energy efficiency deviation identification, trend confirmation, endpoint determination to adjustment execution and effect verification, effectively overcoming the problem of frequent malfunctions due to noise interference in single threshold decisions.

[0026] According to an embodiment of the present invention, the step of sampling and calculating instantaneous energy efficiency information based on a preset control cycle to update the energy efficiency queue specifically includes: At the arrival of each control cycle, the input voltage, input current, output voltage, and output current of the converter are sampled synchronously. The input power is obtained by multiplying the input voltage and the input current, and the output power is obtained by multiplying the output voltage and the output current. The ratio of the output power to the input power is used as instantaneous energy efficiency information; The instantaneous energy efficiency information is appended to the tail of the energy efficiency queue, and the historical energy efficiency information at the head of the energy efficiency queue is removed to maintain a constant queue length. The energy efficiency queue is used to store instantaneous energy efficiency information for multiple consecutive control cycles.

[0027] It should be noted that, in this embodiment, the edge controller is internally configured with voltage and current sensing sensor circuits electrically connected to the input and output terminals of the converter, used to collect input DC voltage, input DC current, output DC voltage, and output DC current, respectively. At the arrival of each control cycle, the edge controller initiates multi-channel synchronous sampling, simultaneously acquiring the instantaneous values ​​of the above four electrical quantities. Subsequently, the energy efficiency calculation logic multiplies the input voltage value by the input current value to obtain the instantaneous input power value of the current cycle; it also multiplies the output voltage value by the output current value to obtain the instantaneous output power value of the current cycle. Further, the instantaneous output power value is divided by the instantaneous input power value, and the resulting ratio is used as the instantaneous energy efficiency characterization value for that cycle. This characterization value is dimensionless and reflects the proportion of input electrical energy converted into output electrical energy by the converter at that instant. After completing the calculation of the instantaneous energy efficiency characterization value, the edge controller performs a queue update operation: appending the instantaneous value to the tail of the energy efficiency queue, while removing the oldest stored historical energy efficiency characterization value from the head of the queue, thus ensuring that the queue always retains the energy efficiency information of the most recent consecutive control cycles. By using the aforementioned constant-length rolling update method, the energy efficiency queue provides a data base that is continuous in time and consistent in length for all subsequent threshold comparisons and trend determinations, ensuring that the historical information window on which each round of optimization decisions is based always maintains the same time span.

[0028] According to an embodiment of the present invention, if the first energy efficiency deviation exceeds a preset trigger threshold, then continuously collecting instantaneous energy efficiency information for multiple cycles and updating the energy efficiency queue to determine the trend direction and trend amplitude specifically includes: When the first energy efficiency deviation exceeds the preset trigger threshold, continue to perform sampling and queue update for a preset number of consecutive verification cycles; Calculate the current energy efficiency queue average value after each cycle is completed; After completing all verification cycles, calculate the difference between the latest energy efficiency queue average and the initial energy efficiency queue average. Use the absolute value of the difference as the trend amplitude, and determine the trend direction based on the positive or negative sign of the difference. A positive value indicates that energy efficiency is on an upward trend, and a negative value indicates that energy efficiency is on a downward trend.

[0029] It should be noted that in this embodiment, when the edge controller determines that the first energy efficiency deviation has exceeded the primary trigger threshold, it enters the second-level verification stage. In this stage, the edge controller keeps all the converter's operating parameters unchanged, does not output any adjustment commands, and only continues to perform sampling and queue update operations, continuously performing a preset number of verification cycles. At the arrival of each verification cycle, the edge controller obtains the current instantaneous energy efficiency value and updates the energy efficiency queue using the same sampling and calculation method as the main loop. It then immediately calculates the arithmetic mean of all energy efficiency values ​​in the updated queue and temporarily stores this average in an internal buffer. After all the preset number of verification cycles are completed, the edge controller reads the initial queue average at the end of the first verification cycle and the final queue average at the end of the last verification cycle from the buffer, and performs a difference calculation between the two. If the difference is positive, it indicates that the overall energy efficiency is increasing within the verification window; if the difference is negative, it indicates that the overall energy efficiency is decreasing. The edge controller records the sign of the difference as the trend direction. Simultaneously, the edge controller calculates the absolute value of this difference and uses it as the trend amplitude. The magnitude of this amplitude reflects the overall drastic degree of energy efficiency change within the verification window. Through the above-described continuous multi-cycle verification and trend extraction process, this embodiment can further confirm whether the energy efficiency change has continuity and directionality, effectively filtering out single, occasional exceedances caused by sampling noise or transient load disturbances.

[0030] According to an embodiment of the present invention, if the trend amplitude exceeds a preset trend threshold, determining the adjustment direction based on the trend direction specifically includes: The trend amplitude is compared with a preset direction threshold. If the trend amplitude is less than the direction threshold, then abandon this round of optimization and clear the verification count; If the trend amplitude reaches or exceeds the direction threshold, then the positive or negative sign of the trend direction is read. If the trend direction is negative, then the adjustment direction is determined to be the adjustment direction to improve energy efficiency; If the trend direction is positive, then the adjustment direction is determined to be a fine-tuning direction that maintains the current energy efficiency trend; The adjustment direction is used to indicate the direction of frequency increase or decrease of subsequent adjustment output or dead time adjustment.

[0031] It should be noted that in this embodiment, after the edge controller determines the trend direction and trend amplitude, it compares the trend amplitude with a preset intermediate direction threshold to determine whether the direction confirmation condition is met. If the trend amplitude is less than the intermediate direction threshold, it indicates that although a deviation was detected in the initial triggering stage, the actual drift of energy efficiency within the subsequent continuous verification window is limited and insufficient to constitute a stable optimization requirement. At this time, the edge controller determines that the current trigger is an occasional disturbance event, abandons the current optimization process, and resets the internal counter used to record the verification process. Then it returns to the main loop to wait for the next control cycle, without outputting any adjustment commands throughout the process. If the trend amplitude reaches or exceeds the intermediate direction threshold, the edge controller confirms that the energy efficiency has undergone a substantial and continuous deviation. At this time, it further reads the previously determined trend direction sign. If the trend direction sign is negative, that is, the energy efficiency is declining, the edge controller determines that the adjustment direction is towards improving energy efficiency, such as increasing or decreasing the switching frequency, depending on the characteristics of the resonant cavity. If the trend direction sign is positive, that is, the energy efficiency is rising, the adjustment direction is determined to be a fine-tuning direction to maintain the current energy efficiency trend, so as to avoid over-adjustment. By using the above-mentioned intermediate direction threshold determination, this embodiment ensures that the adjustment direction is only output when the trend amplitude is sufficiently significant, and the determination of the adjustment direction is strictly based on the measured trend sign, thereby ensuring the correctness of the adjustment action direction and the robustness of the decision.

[0032] According to an embodiment of the present invention, the step of outputting a single-step adjustment amount and performing adjustment in the adjustment direction specifically includes: Based on the adjustment direction, an adjustment command equal to the single adjustment step size is generated from the read single adjustment step size base; The adjustment command is output to the drive circuit of the full-bridge converter to control the switching frequency to change by one step unit in the adjustment direction. After the adjustment command is output, a waiting timer is started, and no new energy efficiency sampling and comparison operations are performed during the timer's countdown. After the timer expires, the sampling and energy efficiency deviation calculation are re-executed to obtain the adjusted energy efficiency response result.

[0033] It should be noted that, in this embodiment, after confirming the need for adjustment through a series of threshold comparisons, the edge controller first reads a pre-set single adjustment step size base from its internal storage. This step size base represents the minimum frequency increment or dead time increment changed by a single adjustment action. Based on the adjustment direction and this step size base, the edge controller generates a corresponding adjustment command. This adjustment command is output to the drive circuit of the full-bridge converter, which controls the operating frequency of the switching transistors to change by one step unit along the determined adjustment direction, for example, increasing or decreasing the frequency by one step. The moment the adjustment command is output, the edge controller immediately starts its internal wait timer. During the entire timer's duration, the edge controller pauses all new sampling and comparison operations to prevent non-steady-state electrical quantities during the converter's transient process from interfering with subsequent judgments. Only after the wait timer expires does the edge controller resume sampling and re-execute energy efficiency calculations and deviation acquisition to capture the true energy efficiency response after the converter enters a new steady state following adjustment. Through the above-described process of adjusting output and waiting for transient equilibrium, this implementation method ensures that each adjustment effect evaluation is based on stable operating conditions, avoiding misjudgments caused by transient oscillations.

[0034] According to an embodiment of the present invention, determining whether the adjusted first energy efficiency deviation has decreased specifically includes: Calculate the first energy efficiency deviation obtained after adjustment and compare it with the first energy efficiency deviation recorded before adjustment; If the first energy efficiency deviation after adjustment is less than the first energy efficiency deviation before adjustment, then the adjustment direction is confirmed to be correct. The benchmark energy efficiency is updated according to the average value of the energy efficiency queue obtained by the current sampling cycle after adjustment, and the locked benchmark value is updated according to the average value of the energy efficiency queue after adjustment and after verification by multiple consecutive control cycles and tending to be stable. If the first energy efficiency deviation after adjustment is not less than the first energy efficiency deviation before adjustment, it is determined that the adjustment direction is wrong, a reverse adjustment command is generated, and an adjustment amount of twice the step size base is output in the direction opposite to the adjustment direction, and the waiting and re-acquisition verification are executed again.

[0035] It should be noted that in this embodiment, after the edge controller completes the adjustment output and waits for transient equilibrium, it re-executes sampling and energy efficiency deviation calculation to obtain the adjusted first energy efficiency deviation. The edge controller compares this adjusted first energy efficiency deviation with the first energy efficiency deviation recorded before adjustment to determine whether energy efficiency has been effectively improved. If the absolute value of the adjusted first energy efficiency deviation is less than the absolute value of the first energy efficiency deviation before adjustment, the current adjustment direction is confirmed to be correct and the adjustment action is effective. At this time, the edge controller adopts a dual update strategy: updating the average energy efficiency queue value obtained by the current sampling cycle after adjustment to a new benchmark energy efficiency for subsequent primary trigger comparison; and continuing to execute verification for multiple consecutive control cycles until the average energy efficiency queue value tends to stabilize, then updating the stabilized queue average value to a new locking benchmark value for subsequent final locking comparison. If the absolute value of the adjusted first energy efficiency deviation is not less than the absolute value of the first energy efficiency deviation before adjustment, the adjustment direction is determined to be incorrect. The edge controller immediately generates a reverse adjustment command, outputting an adjustment amount of twice the step size base in the direction opposite to the current adjustment direction to accelerate the correction speed of the incorrect direction, and then executes waiting and re-sampling verification again. Through the above deviation comparison and direction correction process, this implementation method achieves adaptive verification and error rollback of the adjustment direction, ensuring that the optimization process always converges in the direction of energy efficiency improvement.

[0036] It is worth mentioning that it also includes: During the sampling process of multiple consecutive control cycles, the rate of change of output voltage and the rate of change of output current are collected synchronously. The rate of change of the output voltage is compared with a preset load rise rate threshold, and the rate of change of the output current is compared with a preset load fall rate threshold. If the rate of change of the output voltage exceeds the load rise rate threshold or the rate of change of the output current exceeds the load fall rate threshold, it is determined that the load will undergo a step change, and an advance adjustment amount is pre-output according to the energy efficiency queue. After the load change occurs, energy efficiency sampling and comparison are performed again, and the advance adjustment amount is corrected based on the comparison results.

[0037] It should be noted that in this embodiment, the edge controller executes the conventional energy efficiency sampling and optimization process while simultaneously running a load transient prediction subprocess in parallel. During the sampling process of each control cycle, the edge controller not only collects the instantaneous amplitudes of voltage and current but also calculates the rate of change of output voltage and output current over time. The edge controller compares the rate of change of output voltage with a preset load rise rate threshold and the rate of change of output current with a preset load fall rate threshold. When the rate of change of output voltage exceeds the load rise rate threshold, it indicates that the load is rapidly increasing; when the rate of change of output current exceeds the load fall rate threshold, it indicates that the load is rapidly decreasing. When either of these conditions is met, the edge controller predicts that the load will undergo a step change. Before the actual load step change occurs, the edge controller outputs a lead adjustment amount based on the energy efficiency optimal trend direction reflected by the current energy efficiency queue, adjusting the switching frequency or dead time in advance. After the actual load step change occurs, the edge controller re-executes the complete energy efficiency sampling and chain threshold comparison process, correcting the lead adjustment amount based on the measured results. Through the aforementioned pre-adjustment mechanism, this implementation method effectively compensates for the response lag caused by sudden load changes and improves the timeliness of energy efficiency optimization under dynamic operating conditions.

[0038] It is worth mentioning that it also includes: During the energy efficiency optimization process, the switching frequency value and the corresponding average value of the energy efficiency queue are recorded synchronously after each effective adjustment to form a frequency-energy efficiency mapping sequence. When the offset of the frequency point corresponding to the maximum energy efficiency value in the frequency-energy efficiency mapping sequence relative to the initial nominal frequency point exceeds a preset drift identification threshold, it is determined that the resonant cavity parameters have drifted significantly, and the graded adjustment mode is activated. The first stage is the coarse adjustment stage, which outputs an adjustment amount with a coarse adjustment step size larger than the normal step size until the energy efficiency deviation is reduced to within the preset coarse adjustment termination threshold. The second stage is the fine-tuning stage, which restores the normal step size and re-executes the acquisition and comparison until it converges to within the locked threshold. After fine-tuning is completed, the new frequency point is recorded as the updated nominal frequency point.

[0039] It should be noted that in this embodiment, the edge controller continuously executes the resonant parameter drift identification sub-process during long-term operation. After each effective adjustment, the edge controller associates and records the adjusted switching frequency value with the corresponding energy efficiency queue average value, gradually forming a frequency-energy efficiency mapping sequence. The edge controller periodically scans this mapping sequence, locates the frequency point corresponding to the maximum energy efficiency value in the sequence, and calculates the offset of this frequency point relative to the initial nominal frequency point. When this offset exceeds the preset drift identification threshold, the edge controller determines that the resonant cavity parameters have changed significantly due to factors such as temperature changes, device aging, or tolerance drift, and at this time, a graded adjustment mode is initiated. The first level is the coarse adjustment stage, where the edge controller outputs the adjustment amount with a coarse adjustment step size larger than the normal step size, quickly scanning the frequency range to approach the new optimal energy efficiency range, until the current energy efficiency deviation is reduced to within the preset coarse adjustment termination threshold. Subsequently, the second level, the fine adjustment stage, is entered, where the edge controller restores the normal step size and re-executes the standard chain threshold comparison process until the energy efficiency deviation converges to within the locking threshold, completing the fine positioning. After fine-tuning, the edge controller records the frequency point corresponding to the newly found maximum energy efficiency value as the updated nominal frequency point, which serves as the new benchmark for subsequent drift determination. Through the above-described graded adjustment process, this implementation method ensures both rapid acquisition capability under large drift and accurate final positioning.

[0040] This invention provides a GaN-based power supply full-bridge resonant energy efficiency optimization method based on edge computing, applied to a GaN-based power supply full-bridge resonant energy efficiency optimization circuit based on edge computing. For example... Figure 2 As shown, the energy efficiency optimization circuit includes: The sampling module 21 is used to synchronously acquire the input voltage, input current, output voltage and output current of the converter according to a preset control cycle, and output the corresponding electrical quantity signals. Energy efficiency calculation module 22 is used to calculate instantaneous energy efficiency information based on the electrical quantity signal; The queue storage module 23 is used to store the instantaneous energy efficiency information of multiple consecutive control cycles, forming an energy efficiency queue; Threshold comparison module 24 is used to read the energy efficiency queue and perform comparisons sequentially, and output the adjustment enable signal after three-level progressive verification; The direction determination module 25 is used to determine the direction of energy efficiency trends and generate corresponding adjustment direction indicators; The adjustment amount generation module 26 is used to generate a single-step adjustment amount signal and output it to the drive circuit according to the adjustment enable signal and the adjustment direction indicator; The threshold adaptive correction module 27 is used to generate a threshold correction signal based on the energy efficiency fluctuation amplitude after successful adjustment, so as to dynamically adjust each threshold parameter.

[0041] It should be noted that, in this embodiment, the GaN-based full-bridge resonant energy efficiency optimization circuit based on edge computing is integrated into the edge controller on the power module side. This edge controller uses a digital signal processor or a field-programmable gate array as its core computing carrier. The sampling module is equipped with multiple voltage and current sensors at its front end, connected to the input and output terminals of the converter, respectively. Each sensor converts the collected analog electrical quantities into digital electrical quantities and outputs the corresponding digital electrical quantity signals to the energy efficiency calculation module. After receiving the electrical quantity signals, the energy efficiency calculation module performs multiplication and division operations to obtain the instantaneous energy efficiency characterization value of the current control cycle and outputs this characterization value to the queue storage module. The queue storage module is equipped with a circular buffer, storing the instantaneous energy efficiency information of multiple consecutive control cycles in a first-in-first-out rolling update manner, forming a time-continuous energy efficiency queue for subsequent modules to read. The threshold comparison module reads the energy efficiency queue data from the queue storage module and sequentially performs primary trigger threshold comparison, intermediate direction threshold comparison, and final lock threshold comparison. Each comparison only proceeds to the next level when the conditions of the previous level are met. After all three levels are passed, an adjustment enable signal is output to the adjustment amount generation module. The direction determination module and the chain threshold comparison module work together. When comparing intermediate direction thresholds, the module generates an adjustment direction identifier based on the trend sign, which is also transmitted to the adjustment amount generation module. After receiving the adjustment enable signal and the adjustment direction identifier, the adjustment amount generation module generates a corresponding single-step adjustment amount signal, which is output to the drive circuit of the full-bridge converter via the drive interface to change the switching frequency or dead time. After each successful adjustment, the threshold adaptive correction module collects the current energy efficiency fluctuation amplitude and generates a threshold correction signal based on it, which is fed back to the chain threshold comparison module to dynamically adjust each threshold parameter to match the current operating conditions. Through the coordinated work of the above circuit modules, this implementation fully realizes the entire closed-loop energy efficiency optimization at the hardware level, from signal sampling, energy efficiency calculation, queue storage, three-level progressive threshold determination, direction confirmation, adjustment amount output to threshold adaptive correction. All decisions are completed locally at the edge side without relying on external communication.

[0042] Figure 3 A block diagram of a GaN-based full-bridge resonant power supply efficiency optimization system based on edge computing is shown.

[0043] like Figure 3 As shown, the second aspect of the present invention discloses a GaN-based power supply full-bridge resonant energy efficiency optimization system 3 based on edge computing, including a memory 31 and a processor 32. The memory includes a GaN-based power supply full-bridge resonant energy efficiency optimization method program based on edge computing. When the processor executes the GaN-based power supply full-bridge resonant energy efficiency optimization method program based on edge computing, it performs the following steps: Based on a preset control cycle, instantaneous energy efficiency information is sampled and calculated to update the energy efficiency queue; Based on the updated energy efficiency queue, the average energy efficiency is calculated, and the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action is obtained. If the first energy efficiency deviation exceeds the preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude. If the trend amplitude exceeds the preset trend threshold, the adjustment direction is determined according to the trend direction; Calculate the average energy efficiency of the latest energy efficiency queue and compare it with the locked baseline value after the last optimization action to obtain the second energy efficiency deviation; If the second energy efficiency deviation exceeds the preset locking threshold, then a single-step adjustment amount is output in the adjustment direction and the adjustment is executed; Determine whether the first energy efficiency deviation after adjustment has been reduced; If so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; If not, then roll back and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

[0044] It should be noted that, in this embodiment, during the power-on initialization phase, the edge controller reads the preset trigger threshold, direction threshold, lock threshold, queue length parameter, single adjustment step size base and maximum number of trial steps from the non-volatile memory area. At the same time, it clears the historical energy efficiency queue and fills the queue with the instantaneous energy efficiency information obtained from the first sampling calculation until it is full.

[0045] After entering the main loop, the edge controller periodically triggers the sampling process according to a preset fixed control cycle. At the arrival of each cycle, it synchronously collects the input and output voltage and current information of the converter and calculates the current instantaneous energy efficiency information accordingly. This instantaneous value is then appended to the tail of the energy efficiency queue, and the oldest value at the head of the queue is removed to maintain a constant queue length. Subsequently, the edge controller calculates the arithmetic mean of all energy efficiency values ​​in the queue as the current average energy efficiency, and performs a difference calculation with the baseline energy efficiency recorded after the last successful optimization to obtain the first energy efficiency deviation. If the absolute value of this deviation does not exceed the trigger threshold, it is determined that the current energy efficiency is still within the acceptable range, and it directly waits for the next control cycle; if it exceeds the threshold, the second-level verification is initiated, i.e., maintaining the current operating state and continuing to execute sampling and queue updates for a preset number of consecutive verification cycles. The current queue average value is calculated for each cycle. After the verification cycle is completed, the difference between the last average value and the average value at the beginning of the first verification is calculated, and its absolute value is used as the trend amplitude. The trend direction is determined based on the sign of the difference. If the trend amplitude is less than the direction threshold, it is considered an interference pulse and the current optimization is abandoned; if it reaches or exceeds the threshold, the adjustment direction is determined based on the trend direction. Subsequently, the edge controller calculates the difference between the verified final trend value and the previously maintained locked benchmark value to obtain the second energy efficiency deviation. If the absolute value of the second energy efficiency deviation does not exceed the locking threshold, it is determined that the energy efficiency is within the optimal neighborhood and no adjustment command is output, while the benchmark energy efficiency is updated to the current average value; if it exceeds the threshold, the actual deviation is confirmed and a single-step adjustment amount is output to the drive circuit along the determined direction.

[0046] After adjustment, the edge controller starts a waiting timer to allow the converter to enter a new transient equilibrium. Once the timer expires, sampling and chain comparison are re-executed, comparing the adjusted energy efficiency deviation with the deviation before adjustment to determine if the adjustment is effective. If effective, the baseline energy efficiency is updated and the baseline value is locked; if ineffective, it reverts and outputs a double-step adjustment amount for re-verification, and dynamically adjusts the thresholds at each level based on recent energy efficiency fluctuations after each successful adjustment. This embodiment independently completes the entire closed-loop optimization process at the edge, from energy efficiency deviation identification, trend confirmation, endpoint determination to adjustment execution and effect verification, effectively overcoming the problem of frequent malfunctions due to noise interference in single threshold decisions.

[0047] According to an embodiment of the present invention, the step of sampling and calculating instantaneous energy efficiency information based on a preset control cycle to update the energy efficiency queue specifically includes: At the arrival of each control cycle, the input voltage, input current, output voltage, and output current of the converter are sampled synchronously. The input power is obtained by multiplying the input voltage and the input current, and the output power is obtained by multiplying the output voltage and the output current. The ratio of the output power to the input power is used as instantaneous energy efficiency information; The instantaneous energy efficiency information is appended to the tail of the energy efficiency queue, and the historical energy efficiency information at the head of the energy efficiency queue is removed to maintain a constant queue length. The energy efficiency queue is used to store instantaneous energy efficiency information for multiple consecutive control cycles.

[0048] It should be noted that, in this embodiment, the edge controller is internally configured with voltage and current sensing sensor circuits electrically connected to the input and output terminals of the converter, used to collect input DC voltage, input DC current, output DC voltage, and output DC current, respectively. At the arrival of each control cycle, the edge controller initiates multi-channel synchronous sampling, simultaneously acquiring the instantaneous values ​​of the above four electrical quantities. Subsequently, the energy efficiency calculation logic multiplies the input voltage value by the input current value to obtain the instantaneous input power value of the current cycle; it also multiplies the output voltage value by the output current value to obtain the instantaneous output power value of the current cycle. Further, the instantaneous output power value is divided by the instantaneous input power value, and the resulting ratio is used as the instantaneous energy efficiency characterization value for that cycle. This characterization value is dimensionless and reflects the proportion of input electrical energy converted into output electrical energy by the converter at that instant. After completing the calculation of the instantaneous energy efficiency characterization value, the edge controller performs a queue update operation: appending the instantaneous value to the tail of the energy efficiency queue, while removing the oldest stored historical energy efficiency characterization value from the head of the queue, thus ensuring that the queue always retains the energy efficiency information of the most recent consecutive control cycles. By using the aforementioned constant-length rolling update method, the energy efficiency queue provides a data base that is continuous in time and consistent in length for all subsequent threshold comparisons and trend determinations, ensuring that the historical information window on which each round of optimization decisions is based always maintains the same time span.

[0049] According to an embodiment of the present invention, if the first energy efficiency deviation exceeds a preset trigger threshold, then continuously collecting instantaneous energy efficiency information for multiple cycles and updating the energy efficiency queue to determine the trend direction and trend amplitude specifically includes: When the first energy efficiency deviation exceeds the preset trigger threshold, continue to perform sampling and queue update for a preset number of consecutive verification cycles; Calculate the current energy efficiency queue average value after each cycle is completed; After completing all verification cycles, calculate the difference between the latest energy efficiency queue average and the initial energy efficiency queue average. Use the absolute value of the difference as the trend amplitude, and determine the trend direction based on the positive or negative sign of the difference. A positive value indicates that energy efficiency is on an upward trend, and a negative value indicates that energy efficiency is on a downward trend.

[0050] It should be noted that in this embodiment, when the edge controller determines that the first energy efficiency deviation has exceeded the primary trigger threshold, it enters the second-level verification stage. In this stage, the edge controller keeps all the converter's operating parameters unchanged, does not output any adjustment commands, and only continues to perform sampling and queue update operations, continuously performing a preset number of verification cycles. At the arrival of each verification cycle, the edge controller obtains the current instantaneous energy efficiency value and updates the energy efficiency queue using the same sampling and calculation method as the main loop. It then immediately calculates the arithmetic mean of all energy efficiency values ​​in the updated queue and temporarily stores this average in an internal buffer. After all the preset number of verification cycles are completed, the edge controller reads the initial queue average at the end of the first verification cycle and the final queue average at the end of the last verification cycle from the buffer, and performs a difference calculation between the two. If the difference is positive, it indicates that the overall energy efficiency is increasing within the verification window; if the difference is negative, it indicates that the overall energy efficiency is decreasing. The edge controller records the sign of the difference as the trend direction. Simultaneously, the edge controller calculates the absolute value of this difference and uses it as the trend amplitude. The magnitude of this amplitude reflects the overall drastic degree of energy efficiency change within the verification window. Through the above-described continuous multi-cycle verification and trend extraction process, this embodiment can further confirm whether the energy efficiency change has continuity and directionality, effectively filtering out single, occasional exceedances caused by sampling noise or transient load disturbances.

[0051] According to an embodiment of the present invention, if the trend amplitude exceeds a preset trend threshold, determining the adjustment direction based on the trend direction specifically includes: The trend amplitude is compared with a preset direction threshold. If the trend amplitude is less than the direction threshold, then abandon this round of optimization and clear the verification count; If the trend amplitude reaches or exceeds the direction threshold, then the positive or negative sign of the trend direction is read. If the trend direction is negative, then the adjustment direction is determined to be the adjustment direction to improve energy efficiency; If the trend direction is positive, then the adjustment direction is determined to be a fine-tuning direction that maintains the current energy efficiency trend; The adjustment direction is used to indicate the direction of frequency increase or decrease of subsequent adjustment output or dead time adjustment.

[0052] It should be noted that in this embodiment, after the edge controller determines the trend direction and trend amplitude, it compares the trend amplitude with a preset intermediate direction threshold to determine whether the direction confirmation condition is met. If the trend amplitude is less than the intermediate direction threshold, it indicates that although a deviation was detected in the initial triggering stage, the actual drift of energy efficiency within the subsequent continuous verification window is limited and insufficient to constitute a stable optimization requirement. At this time, the edge controller determines that the current trigger is an occasional disturbance event, abandons the current optimization process, and resets the internal counter used to record the verification process. Then it returns to the main loop to wait for the next control cycle, without outputting any adjustment commands throughout the process. If the trend amplitude reaches or exceeds the intermediate direction threshold, the edge controller confirms that the energy efficiency has undergone a substantial and continuous deviation. At this time, it further reads the previously determined trend direction sign. If the trend direction sign is negative, that is, the energy efficiency is declining, the edge controller determines that the adjustment direction is towards improving energy efficiency, such as increasing or decreasing the switching frequency, depending on the characteristics of the resonant cavity. If the trend direction sign is positive, that is, the energy efficiency is rising, the adjustment direction is determined to be a fine-tuning direction to maintain the current energy efficiency trend, so as to avoid over-adjustment. By using the above-mentioned intermediate direction threshold determination, this embodiment ensures that the adjustment direction is only output when the trend amplitude is sufficiently significant, and the determination of the adjustment direction is strictly based on the measured trend sign, thereby ensuring the correctness of the adjustment action direction and the robustness of the decision.

[0053] According to an embodiment of the present invention, the step of outputting a single-step adjustment amount and performing adjustment in the adjustment direction specifically includes: Based on the adjustment direction, an adjustment command equal to the single adjustment step size is generated from the read single adjustment step size base; The adjustment command is output to the drive circuit of the full-bridge converter to control the switching frequency to change by one step unit in the adjustment direction. After the adjustment command is output, a waiting timer is started, and no new energy efficiency sampling and comparison operations are performed during the timer's countdown. After the timer expires, the sampling and energy efficiency deviation calculation are re-executed to obtain the adjusted energy efficiency response result.

[0054] It should be noted that, in this embodiment, after confirming the need for adjustment through a series of threshold comparisons, the edge controller first reads a pre-set single adjustment step size base from its internal storage. This step size base represents the minimum frequency increment or dead time increment changed by a single adjustment action. Based on the adjustment direction and this step size base, the edge controller generates a corresponding adjustment command. This adjustment command is output to the drive circuit of the full-bridge converter, which controls the operating frequency of the switching transistors to change by one step unit along the determined adjustment direction, for example, increasing or decreasing the frequency by one step. The moment the adjustment command is output, the edge controller immediately starts its internal wait timer. During the entire timer's duration, the edge controller pauses all new sampling and comparison operations to prevent non-steady-state electrical quantities during the converter's transient process from interfering with subsequent judgments. Only after the wait timer expires does the edge controller resume sampling and re-execute energy efficiency calculations and deviation acquisition to capture the true energy efficiency response after the converter enters a new steady state following adjustment. Through the above-described process of adjusting output and waiting for transient equilibrium, this implementation method ensures that each adjustment effect evaluation is based on stable operating conditions, avoiding misjudgments caused by transient oscillations.

[0055] According to an embodiment of the present invention, determining whether the adjusted first energy efficiency deviation has decreased specifically includes: Calculate the first energy efficiency deviation obtained after adjustment and compare it with the first energy efficiency deviation recorded before adjustment; If the first energy efficiency deviation after adjustment is less than the first energy efficiency deviation before adjustment, then the adjustment direction is confirmed to be correct. The benchmark energy efficiency is updated according to the average value of the energy efficiency queue obtained by the current sampling cycle after adjustment, and the locked benchmark value is updated according to the average value of the energy efficiency queue after adjustment and after verification by multiple consecutive control cycles and tending to be stable. If the first energy efficiency deviation after adjustment is not less than the first energy efficiency deviation before adjustment, it is determined that the adjustment direction is wrong, a reverse adjustment command is generated, and an adjustment amount of twice the step size base is output in the direction opposite to the adjustment direction, and the waiting and re-acquisition verification are executed again.

[0056] It should be noted that in this embodiment, after the edge controller completes the adjustment output and waits for transient equilibrium, it re-executes sampling and energy efficiency deviation calculation to obtain the adjusted first energy efficiency deviation. The edge controller compares this adjusted first energy efficiency deviation with the first energy efficiency deviation recorded before adjustment to determine whether energy efficiency has been effectively improved. If the absolute value of the adjusted first energy efficiency deviation is less than the absolute value of the first energy efficiency deviation before adjustment, the current adjustment direction is confirmed to be correct and the adjustment action is effective. At this time, the edge controller adopts a dual update strategy: updating the average energy efficiency queue value obtained by the current sampling cycle after adjustment to a new benchmark energy efficiency for subsequent primary trigger comparison; and continuing to execute verification for multiple consecutive control cycles until the average energy efficiency queue value tends to stabilize, then updating the stabilized queue average value to a new locking benchmark value for subsequent final locking comparison. If the absolute value of the adjusted first energy efficiency deviation is not less than the absolute value of the first energy efficiency deviation before adjustment, the adjustment direction is determined to be incorrect. The edge controller immediately generates a reverse adjustment command, outputting an adjustment amount of twice the step size base in the direction opposite to the current adjustment direction to accelerate the correction speed of the incorrect direction, and then executes waiting and re-sampling verification again. Through the above deviation comparison and direction correction process, this implementation method achieves adaptive verification and error rollback of the adjustment direction, ensuring that the optimization process always converges in the direction of energy efficiency improvement.

[0057] It is worth mentioning that it also includes: During the sampling process of multiple consecutive control cycles, the rate of change of output voltage and the rate of change of output current are collected synchronously. The rate of change of the output voltage is compared with a preset load rise rate threshold, and the rate of change of the output current is compared with a preset load fall rate threshold. If the rate of change of the output voltage exceeds the load rise rate threshold or the rate of change of the output current exceeds the load fall rate threshold, it is determined that the load will undergo a step change, and an advance adjustment amount is pre-output according to the energy efficiency queue. After the load change occurs, energy efficiency sampling and comparison are performed again, and the advance adjustment amount is corrected based on the comparison results.

[0058] It should be noted that in this embodiment, the edge controller executes the conventional energy efficiency sampling and optimization process while simultaneously running a load transient prediction subprocess in parallel. During the sampling process of each control cycle, the edge controller not only collects the instantaneous amplitudes of voltage and current but also calculates the rate of change of output voltage and output current over time. The edge controller compares the rate of change of output voltage with a preset load rise rate threshold and the rate of change of output current with a preset load fall rate threshold. When the rate of change of output voltage exceeds the load rise rate threshold, it indicates that the load is rapidly increasing; when the rate of change of output current exceeds the load fall rate threshold, it indicates that the load is rapidly decreasing. When either of these conditions is met, the edge controller predicts that the load will undergo a step change. Before the actual load step change occurs, the edge controller outputs a lead adjustment amount based on the energy efficiency optimal trend direction reflected by the current energy efficiency queue, adjusting the switching frequency or dead time in advance. After the actual load step change occurs, the edge controller re-executes the complete energy efficiency sampling and chain threshold comparison process, correcting the lead adjustment amount based on the measured results. Through the aforementioned pre-adjustment mechanism, this implementation method effectively compensates for the response lag caused by sudden load changes and improves the timeliness of energy efficiency optimization under dynamic operating conditions.

[0059] It is worth mentioning that it also includes: During the energy efficiency optimization process, the switching frequency value and the corresponding average value of the energy efficiency queue are recorded synchronously after each effective adjustment to form a frequency-energy efficiency mapping sequence. When the offset of the frequency point corresponding to the maximum energy efficiency value in the frequency-energy efficiency mapping sequence relative to the initial nominal frequency point exceeds a preset drift identification threshold, it is determined that the resonant cavity parameters have drifted significantly, and the graded adjustment mode is activated. The first stage is the coarse adjustment stage, which outputs an adjustment amount with a coarse adjustment step size larger than the normal step size until the energy efficiency deviation is reduced to within the preset coarse adjustment termination threshold. The second stage is the fine-tuning stage, which restores the normal step size and re-executes the acquisition and comparison until it converges to within the locked threshold. After fine-tuning is completed, the new frequency point is recorded as the updated nominal frequency point.

[0060] It should be noted that in this embodiment, the edge controller continuously executes the resonant parameter drift identification sub-process during long-term operation. After each effective adjustment, the edge controller associates and records the adjusted switching frequency value with the corresponding energy efficiency queue average value, gradually forming a frequency-energy efficiency mapping sequence. The edge controller periodically scans this mapping sequence, locates the frequency point corresponding to the maximum energy efficiency value in the sequence, and calculates the offset of this frequency point relative to the initial nominal frequency point. When this offset exceeds the preset drift identification threshold, the edge controller determines that the resonant cavity parameters have changed significantly due to factors such as temperature changes, device aging, or tolerance drift, and at this time, a graded adjustment mode is initiated. The first level is the coarse adjustment stage, where the edge controller outputs the adjustment amount with a coarse adjustment step size larger than the normal step size, quickly scanning the frequency range to approach the new optimal energy efficiency range, until the current energy efficiency deviation is reduced to within the preset coarse adjustment termination threshold. Subsequently, the second level, the fine adjustment stage, is entered, where the edge controller restores the normal step size and re-executes the standard chain threshold comparison process until the energy efficiency deviation converges to within the locking threshold, completing the fine positioning. After fine-tuning, the edge controller records the frequency point corresponding to the newly found maximum energy efficiency value as the updated nominal frequency point, which serves as the new benchmark for subsequent drift determination. Through the above-described graded adjustment process, this implementation method ensures both rapid acquisition capability under large drift and accurate final positioning.

[0061] A third aspect of the present invention provides a computer-readable storage medium comprising a method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant circuit based on edge computing. When the method program is executed by a processor, it implements the steps of the method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant circuit based on edge computing as described in any of the preceding claims.

[0062] In summary, this invention provides a method and system for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing. It uses a fixed control cycle to periodically sample and calculate instantaneous energy efficiency information, continuously updating the energy efficiency queue. The average energy efficiency of the queue is compared with the previously optimized benchmark energy efficiency for initial deviation. Only when the deviation exceeds a trigger threshold is continuous multi-cycle verification initiated to determine the energy efficiency trend direction and amplitude. The adjustment direction is confirmed only when the trend amplitude exceeds the trend threshold, and then compared with the locked benchmark value. A single-step adjustment amount is output only when the deviation exceeds the locking threshold. After adjustment, resampling and verification are performed. If the energy efficiency deviation decreases, the benchmark and locked values ​​are updated; otherwise, the process regresses and outputs the adjustment amount in reverse until convergence. This invention effectively filters out false triggers caused by noise and transient disturbances, while reducing communication latency through edge-local closed-loop execution, thus improving the energy efficiency optimization stability of the full-bridge resonant converter under all operating conditions.

[0063] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0064] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing, characterized in that, The method includes: Based on a preset control cycle, instantaneous energy efficiency information is sampled and calculated to update the energy efficiency queue; Based on the updated energy efficiency queue, the average energy efficiency is calculated, and the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action is obtained. If the first energy efficiency deviation exceeds the preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude. If the trend amplitude exceeds the preset trend threshold, the adjustment direction is determined according to the trend direction; Calculate the average energy efficiency of the latest energy efficiency queue and compare it with the locked baseline value after the last optimization action to obtain the second energy efficiency deviation; If the second energy efficiency deviation exceeds the preset locking threshold, then a single-step adjustment amount is output in the adjustment direction and the adjustment is executed; Determine whether the first energy efficiency deviation after adjustment has been reduced; If so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; If not, then roll back and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

2. The method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing according to claim 1, characterized in that, The step of sampling and calculating instantaneous energy efficiency information based on a preset control cycle to update the energy efficiency queue specifically includes: At the arrival of each control cycle, the input voltage, input current, output voltage, and output current of the converter are sampled synchronously. The input power is obtained by multiplying the input voltage and the input current, and the output power is obtained by multiplying the output voltage and the output current. The ratio of the output power to the input power is used as instantaneous energy efficiency information; The instantaneous energy efficiency information is appended to the tail of the energy efficiency queue, and the historical energy efficiency information at the head of the energy efficiency queue is removed to maintain a constant queue length. The energy efficiency queue is used to store instantaneous energy efficiency information for multiple consecutive control cycles.

3. The method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing according to claim 1, characterized in that, If the first energy efficiency deviation exceeds a preset trigger threshold, then instantaneous energy efficiency information for multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude, specifically including: When the first energy efficiency deviation exceeds the preset trigger threshold, continue to perform sampling and queue update for a preset number of consecutive verification cycles; Calculate the current energy efficiency queue average value after each cycle is completed; After completing all verification cycles, calculate the difference between the latest energy efficiency queue average and the initial energy efficiency queue average. Use the absolute value of the difference as the trend amplitude, and determine the trend direction based on the positive or negative sign of the difference. A positive value indicates that energy efficiency is on an upward trend, and a negative value indicates that energy efficiency is on a downward trend.

4. The method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing according to claim 1, wherein if the trend amplitude exceeds a preset trend threshold, the adjustment direction is determined according to the trend direction, specifically including: The trend amplitude is compared with a preset direction threshold. If the trend amplitude is less than the direction threshold, then abandon this round of optimization and clear the verification count; If the trend amplitude reaches or exceeds the direction threshold, then the positive or negative sign of the trend direction is read. If the trend direction is negative, then the adjustment direction is determined to be the adjustment direction to improve energy efficiency; If the trend direction is positive, then the adjustment direction is determined to be a fine-tuning direction that maintains the current energy efficiency trend; The adjustment direction is used to indicate the direction of frequency increase or decrease of subsequent adjustment output or dead time adjustment.

5. The method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing according to claim 1, characterized in that, The step of outputting a single-step adjustment amount and performing adjustment according to the adjustment direction specifically includes: Based on the adjustment direction, an adjustment command equal to the single adjustment step size is generated from the read single adjustment step size base; The adjustment command is output to the drive circuit of the full-bridge converter to control the switching frequency to change by one step unit in the adjustment direction. After the adjustment command is output, a waiting timer is started, and no new energy efficiency sampling and comparison operations are performed during the timer's countdown. After the timer expires, the sampling and energy efficiency deviation calculation are re-executed to obtain the adjusted energy efficiency response result.

6. The method for optimizing the energy efficiency of a GaN-based full-bridge resonant power supply based on edge computing according to claim 1, characterized in that, The determination of whether the adjusted first energy efficiency deviation has decreased specifically includes: Calculate the first energy efficiency deviation obtained after adjustment and compare it with the first energy efficiency deviation recorded before adjustment; If the first energy efficiency deviation after adjustment is less than the first energy efficiency deviation before adjustment, then the adjustment direction is confirmed to be correct. The benchmark energy efficiency is updated according to the average value of the energy efficiency queue obtained by the current sampling cycle after adjustment, and the locked benchmark value is updated according to the average value of the energy efficiency queue after adjustment and after verification by multiple consecutive control cycles and tending to be stable. If the first energy efficiency deviation after adjustment is not less than the first energy efficiency deviation before adjustment, it is determined that the adjustment direction is wrong, a reverse adjustment command is generated, and an adjustment amount of twice the step size base is output in the direction opposite to the adjustment direction, and the waiting and re-acquisition verification are executed again.

7. A GaN-based full-bridge resonant power supply efficiency optimization system based on edge computing, characterized in that, The system includes a memory and a processor. The memory includes a program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant circuit based on edge computing. When the processor executes the program, the GaN-based power supply full-bridge resonant circuit optimization method program based on edge computing performs the following steps: Based on a preset control cycle, instantaneous energy efficiency information is sampled and calculated to update the energy efficiency queue; Based on the updated energy efficiency queue, the average energy efficiency is calculated, and the first energy efficiency deviation between the average energy efficiency and the baseline energy efficiency after the previous optimization action is obtained. If the first energy efficiency deviation exceeds the preset trigger threshold, then the instantaneous energy efficiency information of multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude. If the trend amplitude exceeds the preset trend threshold, the adjustment direction is determined according to the trend direction; Calculate the average energy efficiency of the latest energy efficiency queue and compare it with the locked baseline value after the last optimization action to obtain the second energy efficiency deviation; If the second energy efficiency deviation exceeds the preset locking threshold, then a single-step adjustment amount is output in the adjustment direction and the adjustment is executed; Determine whether the first energy efficiency deviation after adjustment has been reduced; If so, update the benchmark energy efficiency and lock the benchmark value according to the latest adjusted energy efficiency queue; If not, then roll back and output the adjustment amount in reverse, and re-execute the acquisition and adjustment.

8. The GaN-based full-bridge resonant power supply efficiency optimization system based on edge computing according to claim 7, characterized in that, The step of sampling and calculating instantaneous energy efficiency information based on a preset control cycle to update the energy efficiency queue specifically includes: At the arrival of each control cycle, the input voltage, input current, output voltage, and output current of the converter are sampled synchronously. The input power is obtained by multiplying the input voltage and the input current, and the output power is obtained by multiplying the output voltage and the output current. The ratio of the output power to the input power is used as instantaneous energy efficiency information; The instantaneous energy efficiency information is appended to the tail of the energy efficiency queue, and the historical energy efficiency information at the head of the energy efficiency queue is removed to maintain a constant queue length. The energy efficiency queue is used to store instantaneous energy efficiency information for multiple consecutive control cycles.

9. The GaN-based full-bridge resonant power supply efficiency optimization system based on edge computing according to claim 7, characterized in that, If the first energy efficiency deviation exceeds a preset trigger threshold, then instantaneous energy efficiency information for multiple cycles is continuously collected and the energy efficiency queue is updated to determine the trend direction and trend amplitude, specifically including: When the first energy efficiency deviation exceeds the preset trigger threshold, continue to perform sampling and queue update for a preset number of consecutive verification cycles; Calculate the current energy efficiency queue average value after each cycle is completed; After completing all verification cycles, calculate the difference between the latest energy efficiency queue average and the initial energy efficiency queue average. Use the absolute value of the difference as the trend amplitude, and determine the trend direction based on the positive or negative sign of the difference. A positive value indicates that energy efficiency is on an upward trend, and a negative value indicates that energy efficiency is on a downward trend.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer-readable storage medium includes a method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant system based on edge computing. When the method program is executed by a processor, it implements the steps of the method program for optimizing the energy efficiency of a GaN-based power supply full-bridge resonant system based on edge computing as described in any one of claims 1 to 6.

Images