Flyback power supply driving multiplexing control method and system based on switch state recognition

Abstract

The present application relates to power supply control technical field, especially to a kind of based on switch state identification's flyback power drive multiplex control method and system.The method includes the following steps: obtaining the auxiliary winding voltage data of flyback topology side;Auxiliary winding voltage data is carried out level transient polarity deconstruction, and switch state identification data is obtained;According to switch state identification data, asymmetric delay timing generation is carried out, and multiplex control timing data is obtained;According to multiplex control timing data, discharge anti backflow switching is carried out, and loop switching control data is obtained;Loop switching control data is carried out asymmetric dead zone optimization, and drive multiplex control strategy is obtained.The present application is by switch state identification to auxiliary winding voltage, and drive multiplex of synchronous rectification path and active absorption path is realized accordingly, improves the timing control precision of flyback power supply, inhibits sharp peak stress, reduces light load reverse backflow and overlap conduction risk, to improve system operating stability and energy utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of power control technology, and in particular to a flyback power supply drive multiplexing control method and system based on switch state recognition. Background Technology

[0002] Flyback switching power supplies are widely used in PD fast charging, power adapters, industrial auxiliary power supplies, and low-power isolated power supply scenarios due to their relatively simple circuit structure, convenient isolation implementation, small size, and low cost. With the increasing demands for high power density and high conversion efficiency, the traditional secondary-side structure using rectifier diodes is no longer sufficient to meet low-loss requirements. Therefore, synchronous rectification technology has gradually become a common solution in flyback power supplies. Existing synchronous rectification drive methods typically include controller-driven and auxiliary winding self-driven methods. The auxiliary winding self-driven method can directly generate drive using transformer-coupled signals, offering advantages such as simple structure and fewer additional components.

[0003] To suppress the peak stress generated by leakage inductance when the main switch is turned off, existing technologies typically employ RC absorption, RCD absorption, or active absorption schemes. While RC or RCD absorption is simple to implement, it suffers from significant energy dissipation and reduced efficiency under light loads. Active absorption can return some of the absorbed energy to the power circuit, offering advantages in efficiency and device stress control, but it usually requires additional control circuitry or a separate drive unit, leading to an increase in the number of external components, a more complex control chain, and higher costs, which is not conducive to high integration and miniaturization designs. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a flyback power supply drive multiplexing control method and system based on switch state recognition, thereby solving at least one of the aforementioned technical problems.

[0005] This application provides a flyback power supply drive multiplexing control method based on switch state recognition, including the following steps: S1. Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data; S2. Generate asymmetric delay timing data based on switch status identification data to obtain multiplexed control timing data; S3. Based on the multiplexing control timing data, perform the backflow prevention switching to obtain the loop switching control data; S4. Perform asymmetric dead-time optimization on the loop switching control data to obtain the drive multiplexing control strategy.

[0006] In this invention, the operating state of the main switch is identified by the voltage of the auxiliary winding on the secondary side of the flyback topology. This eliminates the need for an additional independent detection and control channel, allowing the extraction of conduction, turn-off, and energy release information from the transient changes in the coupling voltage level and polarity, thus reducing the complexity of the drive decision link. Based on the switch state identification results, an asymmetric delay timing sequence is generated, enabling the synchronous rectification path and the active absorption path to obtain differentiated conduction boundaries at different operating stages. This avoids false triggering caused by traditional fixed delay methods under light load, frequency hopping, or insufficient energy release conditions. Furthermore, through backflow prevention switching, the peak stress relief process and the light load reverse energy backflow prevention process are controlled in stages. This allows for timely release of leakage inductance peak stress after the main switch is turned off, and also suppresses reverse current flow caused by the continuous conduction of the absorption branch before the next cycle. By implementing asymmetric dead-time optimization on the loop switching control data, the overlapping conduction of synchronous rectifier and absorber devices at the moment of switching is avoided, thereby improving the timing safety and energy utilization stability during the drive multiplexing process. As a result, it is possible to reduce the number of driving devices while taking into account light-load efficiency, peak suppression capability and system operation reliability.

[0007] Preferably, S1 includes: Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; perform level transient detection on the auxiliary winding voltage data to obtain level transient data; perform polarity boundary identification based on the level transient data to obtain polarity boundary data; perform switching cycle mapping on the polarity boundary data to obtain off-state interval data; and perform switching state reconstruction based on the off-state interval data to obtain switching state identification data.

[0008] This invention acquires the auxiliary winding voltage data of the flyback topology secondary side and sequentially performs level transient detection, polarity boundary identification, switching cycle mapping, and switching state reconstruction. This allows for the direct reconstruction of the actual operating range of the main switch using the transformer coupling signal, eliminating the need for additional independent timing detection branches or complex sampling circuits. By first extracting abrupt changes in the auxiliary winding voltage and then distinguishing different voltage segments based on polarity boundaries, ringing, glitches, and actual switching events can be effectively separated, improving the accuracy of off-state interval identification. Through switching state reconstruction, the system can more stably distinguish between the on, off, and energy release stages, thus avoiding misjudgments in the switching between the synchronous rectification path and the absorption path, which is beneficial for improving the timing accuracy and operational stability of the drive multiplexing control.

[0009] Preferably, S2 includes: State sequence parsing is performed on the switch state identification data to obtain state parsing data; state edge weighting is performed on the state parsing data to obtain asymmetric state data; delay parameter mapping is performed on the asymmetric state data to obtain asymmetric delay data; driving window timing is constructed based on the asymmetric delay data to obtain multiplexed timing data; multiplexed control timing is fused based on the multiplexed timing data to obtain multiplexed control timing data.

[0010] This invention, by parsing state sequences based on switch state identification data and completing state edge weight calibration, delay parameter mapping generation, drive window timing construction, and multiplexed control timing fusion, transforms the discrete switch states identified in the previous stage into executable timing control bases. State sequence parsing clarifies the sequential relationship between the on, off, and energy release stages on the time axis, avoiding discontinuous drive boundaries caused by controlling based on a single state point. State edge weight calibration distinguishes the roles of different edges in control, eliminating the need for symmetrical processing between on-state establishment and off-state cutoff, which is beneficial for adapting to the different timing requirements of peak discharge and light-load backflow prevention in flyback power supplies. By mapping asymmetric states to delay parameters and constructing drive windows, the synchronous rectification path and absorption path can obtain effective conduction intervals that better reflect the actual working process. After multiplexed control timing fusion, the overlap and gaps between different control windows are reduced, lowering the probability of false triggering and switching inaccuracies, thus providing a stable, continuous, and directional timing basis for subsequent discharge and backflow prevention switching.

[0011] Preferably, S3 includes: Multiplexing timing window data is parsed based on multiplexing control timing data to obtain synchronous rectification window data, absorption and discharge window data, and backflow prevention window data. Discharge start boundary locking is performed on the absorption and discharge window data to obtain peak discharge start data. Cutoff boundary prediction processing is performed on the backflow prevention window data to obtain backflow prevention cutoff data. State mutual exclusion determination processing is performed on the peak discharge start data and backflow prevention cutoff data to obtain state mutual exclusion data. Drive conflict shielding processing is performed on the state mutual exclusion data and synchronous rectification window data to obtain drive conflict shielding data. Loop switching trigger generation processing is performed on the drive conflict shielding data to obtain loop switching control data.

[0012] This invention uses window parsing of multiplexed control timing data to divide the synchronous rectification process, absorption and discharge process, and backflow prevention process into corresponding independent control intervals. This avoids control aliasing caused by different functions sharing the same timing boundary in traditional drive multiplexing methods. The absorption and discharge window's discharge start boundary is locked, ensuring that the active absorption branch is only connected within the effective range where peak stress actually occurs, thus preventing premature conduction of absorption devices and unnecessary energy consumption. Simultaneously, the backflow prevention window's cutoff boundary is predicted, which can promptly limit the absorption capacitor's continued energy release to the secondary side or synchronous rectification path after it has absorbed peak energy, reducing the risk of reverse backflow under light load and frequency hopping conditions. By using peak discharge start data and backflow prevention cutoff data for state mutual exclusion determination, and combining this with synchronous rectification window drive conflict shielding, it is possible to effectively prevent erroneous overlap of synchronous rectification devices and absorption devices during switching periods. The loop switching control data generated by the system enables the two control objectives of peak discharge and light-load backflow prevention to switch stably in sequence, which is beneficial to balance peak suppression effect, light-load operation stability and timing safety in the drive reuse process.

[0013] Preferably, S4 includes: Switching edge extraction is performed on the loop switching control data to obtain switching edge data; overlapping interval detection is performed based on the switching edge data to obtain drive conflict interval data; dead zones on the venting side and backflow prevention side are generated based on the drive conflict interval data and switching edge data to obtain dead zone data on the venting side and backflow prevention side; asymmetric difference correction processing is performed on the dead zone data on the venting side and backflow prevention side to obtain asymmetric dead zone correction data; drive boundary reconstruction processing is performed based on the loop switching control data and asymmetric dead zone correction data to obtain the drive reuse control strategy.

[0014] This invention, by extracting switching edges and detecting overlapping intervals in the loop switching control data, can identify potential timing overlaps between the synchronous rectification path and the absorption path during switching. The system generates separate dead zones for the discharge side and the backflow prevention side, ensuring that the peak stress relief stage and the light load backflow prevention stage no longer share the same fixed dead zone. Instead, differentiated isolation times are used based on different switching directions, enabling more targeted suppression of erroneous conduction between Q1 (secondary-side synchronous rectification switch) and Q5 (absorption branch active switch) during edge transition. By applying asymmetric difference correction to the dead zones on both sides, the discharge side prioritizes peak absorption requirements, while the backflow prevention side prioritizes early cutoff requirements, thus balancing stress release and reverse energy suppression control objectives. Through the combined use of loop switching control data and asymmetric dead zone correction data to reconstruct the drive boundary, the drive multiplexing control strategy maintains a more stable mutually exclusive conduction relationship under different load conditions, reducing switching losses and abnormal backflow risks, and improving timing safety and operational reliability during flyback power supply drive multiplexing.

[0015] Preferably, the switching edge extraction includes: The loop switching control data is subjected to leap-type edge extraction and state-jump-type edge extraction to obtain first edge data and second edge data; edge feature extraction is performed based on the first edge data and second edge data to obtain first edge feature data and second edge feature data respectively; edge comparison and fusion are performed on the first edge data and second edge feature data based on the first edge feature data and second edge feature data to obtain switching edge data.

[0016] This invention employs both leapfrog edge extraction and state transition edge extraction on loop switching control data. This allows for the identification of switching events from two perspectives: control level changes and control state transitions. Leapfrog edge extraction is more effective at capturing the actual level leap moments during loop switching, while state transition edge extraction reflects the transition relationships between different states of the control logic, such as bleeding, backflow prevention, or synchronous rectification recovery. Feature extraction of both types of edges distinguishes edge timing, direction of change, duration, and state transition characteristics, thus avoiding misjudging ringing, glitches, or short-term false transitions as genuine switching edges. Edge comparison and fusion unifies the verification of physical level changes and state semantic changes, ensuring that the obtained switching edge data possesses both accurate timing positioning and consistent control semantics. This provides a more stable boundary basis for subsequent overlapping interval detection and asymmetric dead zone generation, reducing the risk of overlapping conduction and reverse backflow caused by misjudgments in drive switching.

[0017] Preferably, the cross-edge extraction includes: Non-stationary waveform baseline identification is performed on the loop switching control data to obtain baseline trajectory data; dynamic threshold envelope is constructed based on the baseline trajectory data to obtain dynamic envelope data; integral energy window accumulation is performed based on the loop switching control data and dynamic envelope data to obtain integral state window data; asymmetric bidirectional differential slope mapping is performed on the integral state window data to obtain asymmetric differential slope data; and cross-cutting feature point identification is performed based on the asymmetric differential slope data to obtain first edge data.

[0018] This invention employs non-stationary waveform baseline identification on loop switching control data to separate baseline changes caused by ringing, drift, or local disturbances during switching from the original control waveform, thus eliminating reliance on a fixed reference level for subsequent edge identification. A dynamic threshold envelope is constructed based on the baseline trajectory, ensuring that effective edges under different load conditions and switching amplitudes are identified within an adaptive threshold range, avoiding missed or false detections under light load or waveform distortion conditions with fixed threshold methods. By integrating the loop switching control data and the dynamic envelope data using an energy window, the continuous and effective switching trend is highlighted, reducing the interference of transient spikes on the identification results. Furthermore, by combining asymmetric bidirectional differential slope mapping, the rate of change for forward and reverse crossings is differentiated, allowing for separate characterization of edge features in the peak discharge direction and the backflow prevention cutoff direction. The first edge data obtained by the system exhibits higher timing positioning accuracy and anti-interference capability.

[0019] Preferably, state transition edge extraction includes: A three-dimensional time-varying state space is constructed from the loop switching control data to obtain state space data; a transition probability matrix is ​​constructed from the state space data to obtain transition probability matrix data; transition entropy is calculated based on the transition probability matrix data to obtain transition entropy time series data; nonlinear window sliding convolution is performed based on the transition entropy time series data to obtain jump feature enhancement data; step threshold value is determined based on the jump feature enhancement data to obtain second edge data.

[0020] This invention constructs a three-dimensional time-varying state space from loop switching control data and generates transition probability matrices and transition entropy time-series data. This expands a single level change into a representation of state transition patterns, thereby more accurately distinguishing between genuine switching events and pseudo-jumps caused by ringing or glitches. Combining nonlinear window sliding convolution to enhance jump features highlights continuous and effective state transition processes and weakens the impact of short-term disturbances on the judgment results. Obtaining the second edge data through step threshold determination improves the stability and anti-interference capability of edge recognition.

[0021] Preferably, edge comparison and fusion includes: Temporal misalignment mapping is performed based on the first edge data and the second edge data to obtain temporal misalignment data; signal-to-noise matching degree is calculated based on the first edge feature data and the second edge feature data to obtain confidence matrix data; temporal fusion is performed on the first edge data and the second edge data based on the confidence matrix data and the temporal misalignment data to obtain switching edge data.

[0022] In this invention, by performing temporal misalignment mapping based on the first and second edge data, the temporal deviations of the two types of edges in terms of sampling time, response sequence, and local drift can be identified in advance. This prevents the extraction results of leap-type edges and state-jump-type edges from being simply regarded as ideal edges occurring simultaneously, but rather corrected according to the temporal misalignment relationship in the actual control process. By calculating the signal-to-noise matching degree based on the first and second edge feature data, the reliability of the two types of edges in terms of change amplitude, duration characteristics, state semantics, and anti-interference ability can be quantified, thereby reducing the probability of misjudgment caused by ringing, glitches, or short-term false jumps affecting a single edge source. By combining the confidence matrix data with the temporal misalignment data for temporal fusion, the obtained switching edge data can simultaneously consider the accuracy of time positioning and the consistency of control semantics. This helps reduce the risk of overlapping conduction and reverse backfeeding during the Q1 and Q5 switching process, improving the reliability of the drive multiplexing control strategy.

[0023] Preferably, this application also provides a flyback power supply drive multiplexing control system based on switch state recognition, used to execute the flyback power supply drive multiplexing control method based on switch state recognition as described above. The flyback power supply drive multiplexing control system based on switch state recognition includes: The level transient polarity deconstruction module is used to acquire the auxiliary winding voltage data of the secondary side of the flyback topology; and to perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data. The asymmetric delay timing generation module is used to generate asymmetric delay timing data based on switch state identification data, thereby obtaining multiplexed control timing data. The backflow prevention switching module is used to switch the backflow prevention based on the multiplexing control timing data to obtain loop switching control data. The asymmetric dead-time optimization module is used to perform asymmetric dead-time optimization on loop switching control data to obtain a drive multiplexing control strategy.

[0024] This invention acquires the auxiliary winding voltage data of the secondary side of a flyback topology and deconstructs its transient polarity. This allows for the reconstruction of the main switch's on / off and energy release states without adding an independent detection branch, providing a continuous and reliable state basis for subsequent drive control. Based on the switch state identification data, an asymmetric delay timing sequence is generated, so that the synchronous rectification path and the active absorption path no longer use a fixed, consistent drive boundary. Instead, differentiated delay windows are established based on different state edges, thereby improving timing adaptability under light load, frequency hopping, and transient switching conditions. Furthermore, based on the multiplexed control timing data, backflow prevention switching is performed, separating the peak stress relief process from the light load reverse backflow prevention process. This allows the active absorption branch to connect promptly when a peak occurs and exit promptly before the next cycle, avoiding interference from continuous discharge of the absorption capacitor to the secondary circuit. By optimizing the loop switching control data with asymmetric dead time, the drive boundaries of the synchronous rectification device and the absorption device are reconstructed, reducing the probability of overlapping switching and lowering the risk of reverse backflow and switching losses. This invention can improve the overall operational reliability of flyback power supplies under multiple operating conditions by reducing the complexity of driving devices and control links while taking into account peak suppression, light-load efficiency and drive multiplexing stability. Attached Figure Description

[0025] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings: Figure 1 A flowchart illustrating the steps of a flyback power supply drive multiplexing control method based on switch state recognition is shown in one embodiment. Figure 2 A flowchart illustrating the steps of a level transient polarity deconstruction method according to one embodiment is shown. Figure 3 A flowchart illustrating the steps of an embodiment of an asymmetric delay timing generation method is shown. Figure 4 A flowchart illustrating the steps of a backflow prevention switching method according to an embodiment is shown; Figure 5 A flowchart illustrating the steps of an embodiment of an asymmetric dead zone optimization method is shown. Detailed Implementation

[0026] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. Functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0028] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] A flyback power supply with a rated output power of 65W and a switching frequency of 100kHz has a controller that acquires secondary auxiliary winding voltage data at a sampling frequency of 2MHz, with a sampling period of... The transient threshold is 0.5 μs. Set to 0.5V, edge confirmation threshold Set to 1.5V, positive level threshold. Set to 2.0V, negative level threshold Set to 2.0V, zero-level threshold Set to 0.3V, ringing interval The time is set to 3μs. The system identifies the primary side conduction state, secondary side freewheeling state, demagnetization end state, and no-load interval state based on the upper transition, lower transition, and near-zero level interval of the auxiliary winding voltage.

[0030] After obtaining the switch status identification data, the system sets the primary side conduction shielding delay. The secondary side freewheeling confirmation delay is 0.8 μs. The demagnetization release delay is 1.2 μs. The hold delay is 0.6 μs under no-load conditions. The time is 5μs, and multiplexing control timing data is generated accordingly. When the primary side conduction state is detected, the system closes the multiplexing loop and maintains backflow prevention isolation; when the secondary side freewheeling state is detected and after passing through... Afterwards, the system connects to the multiplexing loop; when the demagnetization end state is detected and after a period of time... Afterwards, the system shuts down the multiplexing circuit and starts the discharge branch; when an unloaded interval is detected, the system maintains backflow prevention isolation.

[0031] During loop switching control, the system sets the discharge duration. The dead zone on the venting side is 2 μs. The dead zone on the anti-backflow side is 0.7 μs. The safe interval is 0.4 μs. The interval is 0.2μs. The system generates actual drive signals for the discharge branch, the backflow prevention isolation branch, and the multiplexing circuit based on the loop switching control data. This ensures that the discharge branch is allowed to conduct only after the multiplexing circuit is turned off, and that the multiplexing circuit is allowed to connect only after the backflow prevention isolation is released.

[0032] Under the same input voltage, load current, and switching frequency conditions, compared with the fixed-delay multiplexing control method, the number of false triggers caused by auxiliary winding ringing in this embodiment is reduced from approximately 37 times per 10,000 switching cycles to no more than 3 times; the residual voltage of the multiplexing node is reduced from approximately 1.15V after demagnetization to below 0.18V; the peak value of the reverse current from the secondary side to the auxiliary winding is reduced from approximately 160mA to below 35mA; and the drive multiplexing switching delay is shortened from a fixed 2.5μs to less than 1.2μs. This improves the timing accuracy of the flyback power supply drive multiplexing without adding an independent isolation feedback winding, reduces the risk of false turn-on and reverse backflow, and increases the overall full-load efficiency from 90.8% to 91.5%.

[0033] Please see Figures 1 to 5 This application provides a flyback power supply drive multiplexing control method based on switch state recognition, including the following steps: S1. Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data; In one embodiment, the voltage across the auxiliary winding is connected to the analog-to-digital sampling terminal of the controller via a voltage divider circuit, and the voltage divider resistors are respectively... , The analog-to-digital sampling bit depth is 1 bit, and the reference voltage is 1000 kbps. The modulus sampling value of the kth sampling point is The auxiliary winding voltage sampling value is then expressed as: ,in This represents the auxiliary winding voltage value at the k-th sampling time, where k is the sampling sequence number and the sampling period is... Sampling time is The system collects data before startup or during idle periods when there is no switching action. Sample points and calculate the zero-point offset value. The corrected auxiliary winding voltage is obtained: ,in This is the winding polarity correction factor, with a value of +1 or -1, used to ensure that the auxiliary winding voltage is positive during the secondary-side conduction of the flyback topology. If the actual wiring direction of the auxiliary winding is reversed, it will be set to -1 to avoid reversed switching state judgments due to different winding terminal directions.

[0034] When performing level transient polarity deconstruction on the auxiliary winding voltage data, the system calculates the voltage change at adjacent sampling points: Preset transient detection threshold It is 0.5V, when When, mark the k-th sampling point as a positive transient point; when When, mark the k-th sampling point as the reverse transient point; when At this point, the k-th sampling point is marked as the steady-state point. The corresponding transient polarity P(k) is expressed as: Where P(k) = +1 indicates an upward transient in the auxiliary winding voltage, P(k) = -1 indicates a downward transient in the auxiliary winding voltage, and P(k) = 0 indicates no effective transient in the auxiliary winding voltage. To avoid a single noise point being misjudged as a switching edge, the system requires that the transient polarity of L consecutive sampling points be the same, and the cumulative voltage change satisfy: Where L is the number of consecutive confirmation points. The threshold for edge confirmation is set, such as 1.5V. Only sampling intervals that simultaneously satisfy the conditions of consistent transient polarity and cumulative change reaching the threshold are confirmed as valid transient ranges.

[0035] The system is based on the corrected auxiliary winding voltage Perform level partitioning. Set the positive level threshold to... For example, 2.0V, the negative level threshold is For example, 2.0V, the zero-level threshold is For example, 0.3V, and , .like If so, it is determined that the current auxiliary winding is in the positive level region; if If so, it is determined that the current auxiliary winding is in the negative level region; if If so, the auxiliary winding is currently in the near-zero level region; It cycles periodically between positive, negative, or near-zero levels, with the interval between adjacent polarity changes being less than the preset ringing interval. If the value is 3μs, then the auxiliary winding is determined to be in the ringing zone.

[0036] Based on the aforementioned transient polarity and level partitioning (including positive level region, negative level region, near-zero level region, and ringing region), the system performs switch state identification. If the current sampling interval transitions from the near-zero level region or the positive level region to the negative level region, and the transient polarity P(k) = -1, then the primary-side switch of the flyback topology is determined to be in the on state, and a primary-side on-state flag is generated. If the current sampling interval transitions from a negative level region to a positive level region, and the transient polarity P(k) = +1, then it is determined that the primary-side switch is turned off and the secondary-side rectifier circuit enters the freewheeling conduction state, generating a secondary-side freewheeling status flag. If the current sampling interval drops from the positive level region to the near-zero level region, and a decaying ringing region subsequently appears, then the energy release on the secondary side is determined to be complete, and a demagnetization end status marker is generated. If the auxiliary winding voltage remains near zero and no effective positive or reverse transient occurs, it is determined that the current state is intermittent standby or no-load interval, and an no-load interval status mark is generated. The system obtains switch status identification data, including the sampling time. Correcting auxiliary winding voltage Transient polarity P(k), level partitioning results The edge type E(k) and the corresponding switching state S(k). S(k) includes at least the primary edge on state. Secondary side continuous current state Demagnetization completed and no-load interval status The system arranges the above status markers according to the sampling time sequence to form switch status identification data.

[0037] S2. Generate asymmetric delay timing data based on switch status identification data to obtain multiplexed control timing data; In one embodiment, the system takes switch state identification data as input and first extracts state transition points according to the sampling time sequence. If and Then Let this be the starting time of the nth primary side conduction. , This represents the switch state at the (k-1)th sampling time, i.e., the state at the previous sampling point. This represents the switch state at the k-th sampling time, i.e., the state of the current sampling point; if and Then Let this be the starting time of the nth secondary edge continuation. ;like and Then Let this be the end time of the nth demagnetization. .

[0038] The system sets asymmetric delay parameters corresponding to different state edges, where the primary edge conduction and shielding delay is... Secondary side follow-through confirmation delay Demagnetization ends and release delay No-load hold delay The aforementioned delay is not a uniform fixed delay, but is configured separately according to different switching states to avoid false triggering of the auxiliary winding voltage during the primary side conduction, secondary side freewheeling, and demagnetization ringing stages.

[0039] The system generates multiplexed control timing data: when detected At that time, in the interval Internally generated multiplexed prohibition signal Used to shield against the sudden negative level change at the moment the primary side is turned on; when detected At that time, Generating secondary side follow current confirmation signals at all times This is used to allow multiplexed sampling or auxiliary control logic to enter the secondary side freewheeling window; when detected... At that time, Generate demagnetization release signal at all times And disable the secondary side freewheeling confirmation signal. When continuously in The duration of the state satisfies At that time, an idle interval hold signal is generated. .

[0040] S3. Based on the multiplexing control timing data, perform the backflow prevention switching to obtain the loop switching control data; In one embodiment, the system uses multiplexed control timing data as input to control the bleed branch switch. Anti-backflow disconnect switch and multiplex circuit switches Among them, the discharge branch switch Anti-backflow disconnect switch and multiplex circuit switches This refers to the controlled switching device in the secondary auxiliary winding multiplexing control circuit of a flyback topology, or the logic control signal output by the controller to the corresponding controlled switching device. Discharge branch switch. A discharge branch installed between the multiplexing node and the reference ground is used to conduct the discharge resistor after demagnetization, allowing residual charge on the multiplexing node, sample-and-hold capacitor, or parasitic capacitance to be released quickly; anti-backflow isolating switch. It is installed on the isolation path between the auxiliary winding and the multiplexing node to block the reverse current from the multiplexing node to the auxiliary winding during the primary-side conduction or no-load interval; multiplexing circuit switch It is set between the auxiliary winding sampling node and the multiplexing control node to access the auxiliary winding feedback signal in the effective range of the secondary side freewheeling. =1 indicates that the discharge branch is open. =1 indicates that the backflow prevention disconnect switch is open. =1 indicates multiplexing loop connection. The system sets the discharge duration. Backflow prevention maintenance time If the sampling time satisfy: If the current position is determined to be within the original side conduction shielding zone, the system generates the first switching rule: This means closing the discharge branch, disconnecting the multiplexing circuit, and enabling backflow prevention isolation to prevent the energy storage capacitor or sample-and-hold capacitor from releasing charge back into the auxiliary winding during periods of sudden negative level changes in the auxiliary winding. If the sampling time... satisfy: If the current condition is within the valid range of the secondary side's continuous flow, the system generates a second switching rule: This means closing the discharge branch, releasing the backflow prevention isolation, and connecting to the multiplexing circuit so that the auxiliary winding voltage can serve as a usable feedback signal in the secondary freewheeling state. If the sampling time... satisfy: If the current state is determined to be within the residual charge discharge range after demagnetization has ended, the system generates a third switching rule: This means that the discharge branch is opened while the multiplexing circuit is opened and backflow prevention isolation is maintained, so that residual charge on the multiplexing node, sampling holding capacitor, or parasitic capacitor is released through the discharge branch without backflow to the auxiliary winding or secondary rectifier circuit. If the sampling time satisfy: And not entering the next cycle Then the system generates the maintain rules: If there is a continuous no-load interval signal =1, then the hold rule continues until the next primary-side conduction start time is detected. The system generates loop switching control data. ,in It includes primary-side shielding mode (i.e., the first switching rule), secondary-side multiplexing mode (i.e., the second switching rule), demagnetization and discharge mode (i.e., the third switching rule), and no-load backflow prevention mode (i.e., the fourth switching rule).

[0041] S4. Perform asymmetric dead-time optimization on the loop switching control data to obtain the drive multiplexing control strategy.

[0042] In one embodiment, the system performs asymmetric dead-time optimization on the loop switching control data. The system first identifies mode transitions between adjacent sampling times, if... Then Recorded as the mode switching time The system sets different dead times for different mode switching directions, including the dead time for switching from primary-side shielding mode to secondary-side multiplexing mode. Dead time when switching from secondary-side multiplexing mode to demagnetization discharge mode Dead time for switching from demagnetization discharge mode to no-load anti-backflow mode Dead time when switching from no-load anti-backflow mode to original side shielding mode .

[0043] when It is a primary-edge shielding mode and When in secondary-side multiplexing mode, the system in Always keep and will Switch from 1 to 0; when the condition is met Then, Switching to 1 generates secondary-side multiplexing drive instructions. , which is: .when It is a secondary edge reuse mode and When in demagnetizing / discharging mode, the system is First of all Switch to 0 and keep =0; when satisfied Then, Switch to 1 and Switching to 1 completely disconnects the multiplexing circuit before entering the discharge state, which is as follows: .when It is a demagnetization discharge mode and When in no-load anti-backflow mode, the system is First of all Switch to 0; when satisfied At that time, keep =1、 =0, output no-load anti-backflow drive command. .like It is an unloaded anti-backflow mode and If the original edge shielding mode is enabled, the system will continue to maintain... =1、 =0, and in Then, the primary-side shielding drive is allowed to proceed to the next cycle. System-generated driver reuse control strategy ,in , , These are the drive control signals for the discharge branch, the backflow prevention isolation branch, and the multiplexing circuit, respectively.

[0044] Preferably, S1 includes: S11. Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; In one embodiment, the controller acquires the voltage signal across the secondary auxiliary winding of the flyback topology via a voltage divider sampling circuit. Let the voltage divider resistor be... and ADC sampling value The ADC has 1 bit and the reference voltage is 1. The sampling period is Then the auxiliary winding voltage corresponding to the kth sampling point is: The system collects data before startup. 1 idle sampling point is used to obtain the zero-point offset value. And set the polarity correction coefficient according to the direction of the winding terminals. The corrected auxiliary winding voltage is obtained: ,in A value of +1 or -1 is used to ensure that the auxiliary winding voltage remains positive during the secondary freewheeling period. The system obtains auxiliary winding voltage data, including at least the sampling time. ADC sample value Auxiliary winding voltage and correction of auxiliary winding voltage .

[0045] S12. Perform level transient detection on the auxiliary winding voltage data to obtain level transient data; In one embodiment, the system adjusts the auxiliary winding voltage accordingly. Calculate the voltage change between adjacent sampling points: Set a transient detection threshold. Edge confirmation threshold The number of consecutive confirmed points is L=3. When And satisfy: At that time, the system determines that the k-th sampling point is within the valid level transient interval; if only a single-point change is satisfied but not a continuous cumulative change, it is determined to be a sampling glitch or ringing disturbance and is not considered a valid transient output. The system obtains the level transient data, including the transient sampling time. Voltage change Transient valid marker And the transient direction marker P(k).

[0046] S13. Based on the transient level data, polarity boundary identification is performed to obtain polarity boundary data; In one embodiment, the system identifies the polarity boundary based on the transient direction marker P(k) in the level transient data. When and At that time, the sampling point is marked as the positive polarity boundary. ;when and At that time, the sampling point is marked as the reverse polarity boundary. The system records the occurrence times of adjacent polarity boundaries. and If the following conditions are met: ,in If the boundary value is not specified, it is determined to be a ringing boundary after demagnetization and is not considered a valid boundary for the switching cycle; if the interval between adjacent polarity boundaries is greater than or equal to 3 μs, it is retained as a valid polarity boundary. The system obtains polarity boundary data, including boundary times. Boundary type or Valid boundary marking .

[0047] S14. Perform switching cycle mapping on the polarity boundary data to obtain the off-interval data; In one embodiment, the system determines the switching cycle boundary of the flyback topology based on effective polarity boundary data. If a reverse polarity boundary is detected... Furthermore, a positive polarity boundary was subsequently detected within the same period. Then The corresponding time is recorded as the primary side conduction start time. ,Will The corresponding time is recorded as the start time of the secondary side continuation flow after the primary side is turned off. The system continues to detect the boundary between the positive level region and the near-zero level region. From greater than Descending to Then this moment is recorded as the demagnetization end time. This forms the off-interval data for the nth switching cycle: ,in This indicates the effective turn-off interval from when the primary side is turned off until the secondary side continues to flow after demagnetization ends.

[0048] S15. Reconstruct the switch state based on the off interval data to obtain switch state identification data.

[0049] In one embodiment, the system reconstructs the switching state of each sampling point based on the off-interval data, the correction auxiliary winding voltage, and a level threshold. A positive level threshold is set. =2.0V, negative level threshold =2.0V, zero-level threshold =0.3V. If Then it is marked as the primary side is on. ;like lie in Inner and Then it is marked as a secondary side continuous flow state. ;like near ,and Then it is marked as the end of demagnetization. If all M consecutive sampling points satisfy If no valid polarity boundary is detected, it is marked as an unloaded interval state. The system will , P(k), boundary type / edge type, off interval, and switch state S(k) are arranged in the sampling time sequence to form switch state identification data. S(k) includes at least the primary side conduction state. Secondary side continuous current state Demagnetization completed and no-load interval status .

[0050] Preferably, S2 includes: S21. Perform state sequence parsing based on the switch state identification data to obtain state parsing data; In one embodiment, the system takes switch state identification data as input and extracts consecutive identical switch state intervals according to the sampling time sequence. If Then Recorded as the state transition time And represent the previous state interval as ,in The state begins at time . The end time of the state. This represents the duration of the state. From this, we obtain the state parsing data. .

[0051] S22. Perform state edge weight calibration on the state analysis data to obtain asymmetric state data; In one embodiment, the system determines the edge priority marker based on the state type of adjacent state intervals. This marker is only used to distinguish different switching directions. If Then it is marked as a secondary edge access edge. ;like Then it is marked as the demagnetization release edge. ;like Then it is marked as an unloaded retaining edge. ;like Then it is marked as the original edge shielding edge. The system obtains asymmetric state data. , For the first The switching state before the state switching edge, i.e., the "previous state". For the first The switching state after a state switching edge, i.e., the "next state".

[0052] S23. Generate asymmetric delay data by mapping delay parameters based on asymmetric state data; In one embodiment, the system according to Mapping to the corresponding delay parameter. If =1, then the delay for generating the secondary edge access is... ;like =2, then a demagnetization release delay is generated. ;like =3, then generate no-load hold delay ;like =4, then generate primary edge shielding delay Delay data is represented as ,in This is the asymmetric delay corresponding to the edge of the state.

[0053] S24. Construct the driving window timing based on the asymmetric delay data to obtain the reused timing data; In one embodiment, the system uses the state transition time as the basis for the decision. Calculate the relative delay time starting from the window start point. .like =4 and Then the original edge shielding window is generated. =1; if =1 and Then a secondary edge reuse window is generated. =1; if =2 and Then a demagnetization release window is generated. =1; if =3 and Then an empty hold window is generated. =1.

[0054] S25. Perform multiplexing control timing fusion based on the multiplexing timing data to obtain multiplexing control timing data.

[0055] In one embodiment, the system fuses multiple window signals from the same sampling time. If =1, then the multiplexing disable signal is output first; if =0 and =1, then output the secondary side multiplexing enable signal; if =1, then the secondary side multiplexing enable signal is turned off and the demagnetization release signal is output; if If =1, then the no-load holding signal is output.

[0056] Preferably, S3 includes: S31. Based on the multiplexing control timing data, the multiplexing timing window is parsed to obtain the synchronous rectification window data, the absorption and discharge window data, and the backflow prevention window data. In one embodiment, the system takes multiplexed control timing data as input. If =1 and =0, then the corresponding continuous time period will be resolved as a synchronous rectification window. ;like =1, then it is interpreted as an absorption / discharge window. ;like =1 or =1, then it is interpreted as an anti-backflow window. .get , and Three types of window data.

[0057] S32. Lock the discharge start boundary of the absorption and discharge window data to obtain the peak discharge start data; In one embodiment, the system detects the absorption and discharge window. The rising boundary. If at the sampling time... satisfy =0 and =1, then Locked to the starting boundary of the discharge Set the discharge duration. ,when At that time, a spike discharge start flag is generated. ;when hour, The system received spike discharge start data. .

[0058] S33. Based on the backflow prevention window data, perform cutoff boundary prediction processing to obtain backflow prevention cutoff data; In one embodiment, the system protects the backflow prevention window. The start and end boundaries are identified. If =1 or =1, then the starting point of the current continuous interval is recorded as the starting boundary for preventing backflow. If a synchronous rectification window is subsequently detected... The starting point Then This serves as a candidate boundary for lifting backflow prevention. The system generates a backflow prevention cutoff flag. ,when hour, After entering the synchronous rectification window, This yields the backflow prevention cutoff data. .

[0059] S34. Perform state mutual exclusion judgment processing based on peak discharge start data and backflow prevention cutoff data to obtain state mutual exclusion data; In one embodiment, the system for , and synchronous rectification window marking Perform a mutual exclusion check. If =1, then it is determined to be a discharge priority state, and multiplexing circuit access is prohibited; if =0 and =1, then it is determined to be a backflow prevention priority state, and the connection of the reuse circuit is prohibited; if =0、 =0 and If this condition is met, the system is determined to be in a synchronous rectification enabled state. The system then receives mutually exclusive state data. Its status includes discharge priority, backflow prevention priority, and synchronous rectification permission.

[0060] S35. Perform drive conflict masking processing based on state mutual exclusion data and synchronous rectification window data to obtain drive conflict masking data; In one embodiment, the system uses mutually exclusive state data. Generate blocking rules. If... To prioritize venting, the synchronous rectification drive and the multiplexing circuit drive are disabled. ;like To prevent backflow, the synchronous rectification drive, multiplexing circuit drive, and discharge false triggering are shielded. ;like Synchronous rectification is permitted, and lie in Within this area, multiplexed circuits are permitted for connection. The driver conflict masking data is represented as ,in , , These represent the permissible markings for venting, backflow prevention, and reuse circuits, respectively.

[0061] S36. Perform loop switching trigger generation processing on the drive conflict masking data to obtain loop switching control data.

[0062] In one embodiment, the system according to Generate a circuit switching trigger signal. If =1, then let =1、 =1、 =0, enter demagnetization and discharge mode; if =1 and =0, then let =0、 =1、 =0, enter anti-backflow mode; if =1, then let =0、 =0、 =1, enter secondary edge reuse mode; if all three are 0, then remain unchanged. =0、 =1、 =0. To release the desired state of the branch circuit breaker, To prevent backflow into the desired state of the disconnect switch, This represents the desired state of the multiplexed circuit switch. A value of 1 indicates that the corresponding switch is on or the corresponding function is enabled, while a value of 0 indicates that the corresponding switch is off or the corresponding function is disabled.

[0063] Preferably, S4 includes: S41. Extract the switching edge from the loop switching control data to obtain the switching edge data; In one embodiment, the system takes loop switching control data as input. If Then Let it be the r-th switching edge. Where x∈{d,b,m}, for , as well as One of the situations. When The change from 0 to 1 is marked as a rising edge. ,when The change from 1 to 0 is marked as a falling edge. .

[0064] In one embodiment, the system takes loop switching control data as input and extracts edges from two dimensions: changes in switch combinations and mode state transitions. The system first constructs a switch combination vector: ,like Then we get a straddle edge; if This yields state-transition edges. The system performs time-domain comparison and fusion of bridging edges and state-transition edges to obtain switching edge data, including... This represents the switch combination state before the r-th switching edge occurs. This represents the switch combination state after the r-th switching edge occurs. , , These represent the state changes of the discharge branch, the backflow prevention isolation branch, and the reuse circuit at this edge, respectively. This is the edge-front loop switching mode. This is the edge-back loop switching mode. As the source of the switching signal edge, This indicates the direction of mode transition.

[0065] S42. Detect overlapping intervals based on switching edge data to obtain drive conflict interval data; In one embodiment, the system detects whether there is a conflict between the desired states of adjacent switching edges. If within the interval Internal satisfaction =1 and =1, then the discharge branch and the reuse circuit overlap; if =1 and =0, then the backflow prevention isolation during the release period is determined to be lifted; if =1 and If the value is 1, it is determined that the multiplexed loop is connected but the isolation has not been released. When any of the above conditions are met, the interval is marked as a drive conflict interval.

[0066] In one embodiment, the system uses switching edge data In , and adjacent edge times , The desired switching state after reconstructing the edge. For the interval. The system selects the desired switch combination within this interval as If satisfied and If the conditions are met, it is determined that the discharge branch and the multiplexing circuit overlap; if the conditions are met... and If the conditions are met, the backflow prevention isolation during the release period is deemed lifted; =1 and If the value is 1, it indicates that the multiplexed circuit has been connected but the isolation has not been lifted. When any of the above conditions are met, [the following will occur]. This is denoted as the driving conflict interval, and the driving conflict interval data is obtained.

[0067] S43. Generate the venting side dead zone and the anti-backflow side dead zone based on the drive conflict interval data and the switching edge data to obtain the venting side dead zone data and the anti-backflow side dead zone data. In one embodiment, if the switching edge corresponds to entering the demagnetization discharge mode, a discharge-side dead zone is generated before the discharge branch is turned on. ,in Maintain within the dead zone =0、 =1、 =0, discharge is allowed only after the multiplexed circuit is disconnected and isolation is completed. If the switching edge corresponds to entering the secondary side multiplexing mode, a backflow prevention dead zone is generated. ,in Maintain within the dead zone =0, access can only be reused after the backflow prevention isolation is lifted.

[0068] S44. Perform asymmetric difference correction on the dead zone data of the venting side and the dead zone data of the backflow prevention side to obtain asymmetric dead zone correction data. In one embodiment, the system compares the endpoints of the bleed-side dead zone. and the end point of the backflow prevention dead zone and set a safety interval. If the demagnetizing discharge mode is entered, the following conditions are met. Then the discharge allowance boundary will be modified to If the conditions are met when entering the secondary edge reuse mode. Then the reuse allowable boundary will be modified to . For multiplex circuit switch The shutdown boundary moment is the point in time when the multiplexed loop switches from the access state to the disconnect state. , .

[0069] S45. Based on the loop switching control data and the asymmetric dead zone correction data, the drive boundary is reconstructed to obtain the drive reuse control strategy.

[0070] In one embodiment, the system corrects data based on asymmetric dead zones. Reconstructing the actual driving signal , , .like If it is located within any dead zone, then output the same value. =0、 =1、 =0; if and In demagnetizing / discharging mode, the output is... =1、 =1、 =0; if and If it is in secondary-side multiplexing mode, then the output is... =0、 =0、 =1; if If it is in primary-side shielding mode or no-load backflow prevention mode, then the output... =0、 =1、 =0.

[0071] Preferably, the switching edge extraction includes: The loop switching control data is subjected to cross-edge extraction and state transition edge extraction to obtain the first edge data and the second edge data. In one embodiment, the system uses loop switching control data. First, construct a switch combination vector as input. .like Then This is denoted as the crossing edge moment. Record the switch combinations before and after the change. , Obtain the first edge data , This represents the switch combination state before the i-th crossover edge occurs. This represents the switch combination state after the i-th crossover edge occurs. If Then Let the edge time of the state transition be denoted as . And record pattern changes Obtain the second edge data , This refers to the edge-before-the-loop switching mode before the j-th state transition edge occurs. This refers to the edge-after-loop switching mode after the j-th state transition edge occurs.

[0072] Edge features are extracted based on the first edge data and the second edge data to obtain the first edge feature data and the second edge feature data, respectively; In one embodiment, the system extracts the number of switch changes from the first edge data. And based on the changing branch mark edge source ,For example The change is marked on the vent side edge. The change is marked on the anti-backflow side edge. The change is marked as a reused side edge. For the second edge data, the system extracts the mode transition direction. The changes between the primary-side shielding mode, secondary-side multiplexing mode, demagnetization discharge mode, and no-load backflow prevention mode are recorded as mode edge features. This yields the first edge feature data. and second edge feature data .

[0073] The first edge feature data and the second edge feature data are compared and fused to obtain the switching edge data.

[0074] In one embodiment, the system sets an edge alignment tolerance. =1 sampling period If it exists If the leaping edge and the state-transition edge correspond to the same switch, the earlier time is taken as the fusion edge time. and will , , , , , Merge into switching edge data. If only the first edge data exists and there is no corresponding second edge data, retain it as a switch edge; if only the second edge data exists and there is no corresponding first edge data, retain it as a mode edge.

[0075] Preferably, the cross-edge extraction includes: Non-stationary waveform baseline identification is performed on the loop switching control data to obtain baseline trajectory data; In one embodiment, the system takes loop switching control data as input and... , , These are respectively used as three state waveforms. Where x∈{d,b,m}. If a certain path In continuous If the value remains unchanged within a certain number of sampling points, then this stable value is taken as the current baseline. ,Right now ,in =4. This yields the baseline trajectory data. .

[0076] Dynamic threshold envelopes are constructed based on baseline trajectory data to obtain dynamic envelope data; In one embodiment, a dynamic threshold envelope is constructed based on baseline trajectory data to obtain dynamic envelope line data. The system revolves around each baseline. Construct upper and lower envelopes, and set envelope tolerances. =0.25, then: ,when or When this occurs, the waveform of that path is determined to deviate from the stable baseline. Dynamic envelope data is then obtained from this. .

[0077] Based on the loop switching control data and dynamic envelope data, the integral energy window is accumulated to obtain the integral state window data; In one embodiment, integral energy window accumulation is performed based on loop switching control data and dynamic envelope data to obtain integral state window data. The system sets the integral window length N_I=3, and calculates the following for each channel's state waveform: , For the x-th integral state window data at the k-th sampling time, The sampling sequence number within the integration window. This is the current sampling sequence number. The length of the integration window. For the xth loop in Switch status data at any given time. The baseline trajectory data of the x-th loop at the i-th sampling point. The sampling period. If... If it remains stable, then Approaching 0; if If a leapfrog change occurs, then The value increases within a short window. This yields the integral state window data. .

[0078] Asymmetric bidirectional differential slope mapping is performed on the integral state window data to obtain asymmetric differential slope data; In one embodiment, asymmetric bidirectional differential slope mapping is performed on the integral state window data to obtain asymmetric differential slope data. The system sets the forward differential length. =2, Backward Difference Length =4, and calculate: , ,in and Different values ​​are used to reflect the asymmetry of state changes before and after the switch. This yields asymmetric difference slope data. .

[0079] The first edge data is obtained by identifying cross-feature points based on asymmetric difference slope data.

[0080] In one embodiment, cross-feature point identification is performed based on asymmetric differential slope data to obtain first edge data. If a certain path satisfies... ,and , Then Identified as the moment of the road crossing edge .when When, it is marked as crossing the upper edge. ;when When, mark as the lower crossing edge .

[0081] Preferably, state transition edge extraction includes: A three-dimensional time-varying state space is constructed from the loop switching control data to obtain state space data; In one embodiment, the system uses loop switching control data as input to construct a three-dimensional state vector from the states of the three loop switches: ,because , , All values ​​are either 0 or 1, therefore X(k) can correspond to 8 state combinations. The system forms state space data according to the sampling time sequence. .

[0082] Transition probability matrix data is obtained by constructing a transition probability matrix from the state space data; In one embodiment, the system statistically analyzes the sliding interval. =State within 8 sampling points Jump to status Number of times , This is the initial state combination before the state transition. Let be the target state combination after the state transition, and 'a' and 'b' be the state combination numbers in the three-dimensional state space, where 'a' represents the state number before the transition and 'b' represents the state number after the transition. Both can take values ​​between 1 and 8. The following calculations are performed: , To start from the same initial state within the sliding window corresponding to the k-th sampling time. Starting from the beginning, the total number of transitions to all possible target states, when the denominator is 0, let... =0, thus obtaining the transition probability matrix data. .

[0083] The transition entropy is calculated based on the transition probability matrix data to obtain the transition entropy time series data; In one embodiment, the system calculates for each sampling time: ,in, = This is used to avoid taking the logarithm of 0. A larger H(k) indicates a more concentrated occurrence of state transitions, resulting in better transition entropy time-series data. The complexity of the current circuit switching state is measured by statistically analyzing the transition probabilities of the three-way switch state combinations within a short time window; the transition entropy is low in a stable state and increases near state transitions or switching boundaries.

[0084] Nonlinear window sliding convolution is performed on the time series data of transfer entropy to obtain jump feature enhancement data; In one embodiment, the system sets the window radius r=2 and calculates: ,in Enhance the jump feature data at the k-th sampling time. This is the current sampling sequence number. The offset index within the sliding convolution window. The radius of the sliding convolution window. It is a hyperbolic tangent nonlinear compression function. The transfer entropy data corresponding to the (k+i)th sampling point. This is the transfer entropy data corresponding to the (k+i-1)th sampling point. The nonlinear normalized scaling parameter, with a value of 0.1, is used to limit the influence of anomalous spikes and obtain data with enhanced jump features. . It is used to represent the change in transfer entropy between adjacent sampling points. If the loop state is stable, the change in transfer entropy is small; if a mode jump occurs, the transfer entropy will show a sudden increase or decrease. This is used for nonlinear compression of the magnitude of changes. When the entropy change is small, the output is close to 0; when the entropy change is significant, the output gradually approaches 1. This highlights the true jumps while limiting the infinite amplification of abnormal spikes. The jump responses of multiple sampling points are accumulated within a local time window by summing, avoiding the direct triggering of edge detection by a single noise point. Y(k) will only increase significantly when there are continuous or locally concentrated significant entropy changes.

[0085] The step threshold value is determined based on the jump feature enhancement data to obtain the second edge data.

[0086] In one embodiment, the system sets a step threshold value. =2.0, if ,and Then Determined as a state transition edge moment And record the jump direction. This yields the second edge data.

[0087] Preferably, edge comparison and fusion includes: Time-domain misalignment mapping is performed based on the first edge data and the second edge data to obtain time-domain misalignment data; In one embodiment, a time-domain misalignment mapping is performed based on the first edge data and the second edge data to obtain time-domain misaligned data. The system uses the first edge data and the second edge data as input to calculate the time-domain misalignment between them: , For the i-th straddle edge moment, At the edge of the j-th state transition, if the following conditions are met... Then the edges of that group are denoted as alignment candidates, where The system obtains time-domain misaligned data. ,in =1 indicates that alignment is possible. =0 indicates that alignment is not possible.

[0088] The signal-to-noise matching degree is calculated based on the first edge feature data and the second edge feature data to obtain the confidence matrix data; In one embodiment, the system uses first edge feature data Second edge feature data Perform a match. This is the enhanced data for the transition feature at the j-th state transition edge corresponding to the k-th sampling time. If... and For the same switching direction, for example To reuse the side edge corresponding For changes in secondary side multiplexing mode / entry into secondary side multiplexing mode, the direction matching flag is... =1, otherwise 0. The system sets the first edge strength threshold. =1, second edge reinforcement threshold =2.0, and construct the confidence matrix: , thus obtaining the confidence matrix data .

[0089] The first edge data and the second edge data are fused in the time domain based on the confidence matrix data and the time domain misalignment data to obtain the switching edge data.

[0090] In one embodiment, the system targets the first edge data. Each first edge record in the dataset is matched. Let the i-th first edge record include the time of the first edge. Type of circuit and switch edge type ,in The system receives data at the second edge. Searching for the content that satisfies and The smallest j-th second edge record, where the second edge record includes the second edge time. Edge-front loop switching mode and edge-back loop switching mode If a matching item exists that meets the conditions, then the i-th first edge record and the j-th second edge record correspond to the same switch, and the earlier edge time of the two is taken as the fusion edge time. And merge to generate the r-th switching edge data: If the i-th first edge record does not have a second edge record that satisfies the condition, then the first edge record is retained as a switch edge; if the second edge data... If a second edge record exists that does not match any of the first edge records, then that second edge record is retained as a mode edge. The system sorts the merged edges, switch edges, and mode edges in ascending order of edge timing to obtain the switching edge data.

[0091] Preferably, this application also provides a flyback power supply drive multiplexing control system based on switch state recognition, used to execute the flyback power supply drive multiplexing control method based on switch state recognition as described above. The flyback power supply drive multiplexing control system based on switch state recognition includes: The level transient polarity deconstruction module is used to acquire the auxiliary winding voltage data of the secondary side of the flyback topology; and to perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data. The asymmetric delay timing generation module is used to generate asymmetric delay timing data based on switch state identification data, thereby obtaining multiplexed control timing data. The backflow prevention switching module is used to switch the backflow prevention based on the multiplexing control timing data to obtain loop switching control data. The asymmetric dead-time optimization module is used to perform asymmetric dead-time optimization on loop switching control data to obtain a drive multiplexing control strategy.

[0092] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0093] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A flyback power supply drive multiplexing control method based on switch state recognition, characterized in that, Includes the following steps: S1. Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data; S2. Generate asymmetric delay timing data based on switch status identification data to obtain multiplexed control timing data; S3. Based on the multiplexing control timing data, perform the backflow prevention switching to obtain the loop switching control data; S4. Perform asymmetric dead-time optimization on the loop switching control data to obtain the drive multiplexing control strategy.

2. The method according to claim 1, characterized in that, S1 includes: Obtain the auxiliary winding voltage data of the secondary side of the flyback topology; Level transient detection is performed on the auxiliary winding voltage data to obtain level transient data; Polarity boundary identification is performed based on the transient level data to obtain polarity boundary data; switching cycle mapping is performed on the polarity boundary data to obtain the off-state interval data; The switch state is reconstructed based on the off interval data to obtain switch state identification data.

3. The method according to claim 1, characterized in that, S2 include: State sequence parsing is performed based on switch state identification data to obtain state parsing data; State edge weights are calibrated on the state analysis data to obtain asymmetric state data; Delay parameter mapping is performed based on asymmetric state data to generate asymmetric delay data; driving window timing is constructed based on asymmetric delay data to obtain multiplexed timing data; multiplexed control timing is fused based on multiplexed timing data to obtain multiplexed control timing data.

4. The method according to claim 1, characterized in that, S3 include: Based on the multiplexing control timing data, the multiplexing timing window is parsed to obtain synchronous rectification window data, absorption and discharge window data, and backflow prevention window data; The data from the absorption and discharge window is used to lock the discharge start boundary to obtain the peak discharge start data; the data from the backflow prevention window is used to predict the cutoff boundary to obtain the backflow prevention cutoff data; the data from the peak discharge start data and the backflow prevention cutoff data are used to determine the mutual exclusion of states to obtain the mutual exclusion of states. Drive conflict masking is performed based on state mutual exclusion data and synchronous rectification window data to obtain drive conflict masking data; loop switching trigger generation is performed on the drive conflict masking data to obtain loop switching control data.

5. The method according to claim 1, characterized in that, S4 include: Switching edge extraction is performed on the loop switching control data to obtain switching edge data; overlapping interval detection is performed based on the switching edge data to obtain drive conflict interval data; dead zones on the venting side and backflow prevention side are generated based on the drive conflict interval data and switching edge data to obtain dead zone data on the venting side and backflow prevention side; asymmetric difference correction processing is performed on the dead zone data on the venting side and backflow prevention side to obtain asymmetric dead zone correction data; drive boundary reconstruction processing is performed based on the loop switching control data and asymmetric dead zone correction data to obtain the drive reuse control strategy.

6. The method according to claim 5, characterized in that, Switching edge extraction includes: The loop switching control data is subjected to leap-type edge extraction and state-jump-type edge extraction to obtain first edge data and second edge data; edge feature extraction is performed based on the first edge data and second edge data to obtain first edge feature data and second edge feature data respectively; edge comparison and fusion are performed on the first edge data and second edge feature data based on the first edge feature data and second edge feature data to obtain switching edge data.

7. The method according to claim 6, characterized in that, Cross-edge extraction includes: Non-stationary waveform baseline identification is performed on the loop switching control data to obtain baseline trajectory data; dynamic threshold envelope is constructed based on the baseline trajectory data to obtain dynamic envelope data; integral energy window accumulation is performed based on the loop switching control data and dynamic envelope data to obtain integral state window data; asymmetric bidirectional differential slope mapping is performed on the integral state window data to obtain asymmetric differential slope data; and cross-cutting feature point identification is performed based on the asymmetric differential slope data to obtain first edge data.

8. The method according to claim 6, characterized in that, State transition edge extraction includes: A three-dimensional time-varying state space is constructed from the loop switching control data to obtain state space data; a transition probability matrix is ​​constructed from the state space data to obtain transition probability matrix data; transition entropy is calculated based on the transition probability matrix data to obtain transition entropy time series data; nonlinear window sliding convolution is performed based on the transition entropy time series data to obtain jump feature enhancement data; step threshold value is determined based on the jump feature enhancement data to obtain second edge data.

9. The method according to claim 6, characterized in that, Edge alignment and fusion include: Temporal misalignment mapping is performed based on the first edge data and the second edge data to obtain temporal misalignment data; signal-to-noise matching degree is calculated based on the first edge feature data and the second edge feature data to obtain confidence matrix data; temporal fusion is performed on the first edge data and the second edge data based on the confidence matrix data and the temporal misalignment data to obtain switching edge data.

10. A flyback power supply drive multiplexing control system based on switch state recognition, characterized in that, For executing the flyback power supply drive multiplexing control method based on switch state recognition as described in claim 1, the flyback power supply drive multiplexing control system based on switch state recognition includes: The level transient polarity deconstruction module is used to acquire the auxiliary winding voltage data of the secondary side of the flyback topology; and to perform level transient polarity deconstruction on the auxiliary winding voltage data to obtain switch state identification data. The asymmetric delay timing generation module is used to generate asymmetric delay timing data based on switch state identification data, thereby obtaining multiplexed control timing data. The backflow prevention switching module is used to switch the backflow prevention based on the multiplexing control timing data to obtain loop switching control data. The asymmetric dead-time optimization module is used to perform asymmetric dead-time optimization on loop switching control data to obtain a drive multiplexing control strategy.

Images