Multi-port integration method of integrated energy router suitable for multi-energy access

Abstract

A kind of integrated multi-port method of comprehensive energy router suitable for multiple energy access, using the steps of topology layer integration, access layer integration, control layer integration, integrates multiple ports on the same group of shared switch bridge arm array and the same shared magnetic element at the physical topology level, standardizes heterogeneous energy interface into a unified physical port at the access level, standardizes heterogeneous energy interface into a unified physical port at the access level, solves the problems of port dispersion and linear expansion of hardware scale, lack of standardization of heterogeneous energy access interface and insufficient suppression of multi-port circulation and power coupling. The hardware scale is sublinearly expanded, the power density is significantly improved, and the modular expansion is feasible; heterogeneous energy truly realizes plug and play and supports runtime re-identification; under the integrated topology, the dynamic magnetic path coupling coefficient tensor is first proposed; three-stage closed-loop coupling produces overall gain beyond simple addition.

Description

Technical Field

[0001] This invention belongs to the field of integrated energy router technology, and relates to a multi-port integration method for integrated energy routers suitable for multi-energy access. Background Technology

[0002] With the large-scale deployment of distributed photovoltaic, energy storage, wind power, and other renewable energy sources, scenarios such as distributed power generation parks, integrated photovoltaic-storage-charging stations, and integrated generation-grid-load-storage demonstration projects have placed new demands on power electronic hubs that handle the access and energy distribution of multiple energy sources. As such power electronic hubs, integrated energy routers need to simultaneously support the access of heterogeneous energy sources such as photovoltaic, energy storage, wind power, and the power grid; they need to handle multiple types of output, including AC loads, DC loads, and charging piles; and they need to achieve energy distribution and control between various ports.

[0003] In existing technologies, the port integration methods of multi-port energy routers can be broadly categorized into three engineering directions.

[0004] The first approach is a magnetically integrated multi-port energy router based on a multi-winding high-frequency transformer. Patent document CN106374451B discloses a DC residential energy router and its control method based on a multi-port converter. This router consists of four H-bridge modules, two bidirectional Buck / Boost converter modules, four external series inductors, and a high-frequency four-winding transformer. While this approach achieves port integration at the magnetic component level, each port still has an independent H-bridge (four in total, each containing four power switching devices), and the total number of switching devices increases linearly with the number of ports. It is geared towards residential load scenarios and lacks a universal access interface for heterogeneous energy sources such as wide-voltage-range photovoltaic arrays, variable-frequency wind power, and industrial frequency grids. The control layer focuses on energy management and power balance between the source, load, and storage, without providing suppression strategies for port coupling or circulating current problems that may be introduced by magnetic integration itself.

[0005] The second approach is a multi-port energy router with a common DC bus and independent converters at each port (i.e., a weak sharing form based on the idea of ​​multi-port converters sharing a common bridge arm). Patent document CN116031920A discloses a hierarchical energy coordination control strategy for a multi-port power electronic device. This device includes six ports—photovoltaic, wind power, energy storage, grid connection, AC load, and DC load—sharing a single DC bus, and proposes a hierarchical energy coordination control strategy. In this approach, each port remains an independent converter, physically connected only through the DC bus. Magnetic components and switching devices are not shared between ports, and the hardware scales linearly with the number of ports. Coordination control is located at the energy management layer (such as maximum power point tracking, constant bus voltage, and mode switching), and does not involve switch-level decoupling.

[0006] The third direction is multi-port energy routers and plug-and-play control methods under a cascaded power electronic transformer architecture, namely, high-frequency isolated bus topology. Patent document CN117458605A discloses a fully controllable flexible power distribution system and method based on a DC multi-port power exchanger. It employs a multi-port DC power router based on a cascaded high-frequency transformer, combined with a droop control method to achieve plug-and-play access for source, load, and storage. In this direction, plug-and-play access still relies on the independent power electronic converters at the front end of each distributed power source; coordinated control uses a droop control method, relying on the DC bus voltage as the information carrier. The port topology is independent, avoiding port coupling issues under shared magnetic circuits or shared switch arms; the number of magnetic components still increases with the number of ports.

[0007] Furthermore, patent document CN102624234B discloses a full-bridge three-port DC-DC converter and its control method, which achieves the dual sharing concept of switching device multiplexing and magnetic component multiplexing on a three-port scale. However, it is only applicable to a three-port scale, the secondary side uses diode rectification and does not support bidirectional operation, it does not support AC ports, and it does not propose quantifiable topology integration design indicators. Patent document CN114244139B discloses a multi-winding high-frequency magnetically coupled converter and its power decoupling control method, but each port is still configured with an independent full-bridge, and switching device multiplexing is not implemented.

[0008] For plug-and-play access of heterogeneous energy sources, patent document CN107171361A discloses an intelligent distribution station and a grid-source-load-storage coordinated control system based on grid-source-load-storage coordinated control. It explicitly proposes a plug-and-play interface for distributed power sources, including a handshake protocol, grid connection mode switching, and energy management strategy distribution. This existing technology treats plug-and-play as an interface layer concept, relying on the independent power electronic converters at the front end of each distributed power source. It only performs interface handshakes and connects with upper-level power commands, without identifying the physical characteristics of the energy type (such as impedance characteristics, volt-ampere characteristics, and state of charge feedback in the 1Hz to 100Hz frequency band), and cannot automatically re-identify the energy type and switch modulation modes based on changes in operating conditions during operation.

[0009] In summary, the existing technology has the following problems: First, the dispersed ports lead to excessive size and increased costs associated with independent converters. The sharing of magnetic components and switching devices is treated separately and not implemented synchronously within the same physical topology. Furthermore, the lack of quantifiable topology integration design metrics results in linear hardware scaling, low power density, and high expansion costs as the number of ports increases. Simultaneously, the lack of engineerable modular expansion rules means that adding new ports requires redesigning the hardware layer rather than incrementally expanding the existing topology. The fundamental technical reason is that the magnetic integration path and the switching multiplexing path belong to different technological evolution paths. Magnetic integration focuses on multi-winding coupling of transformers, while switching multiplexing focuses on time-sharing or multi-level multiplexing of bridge arms; these two have long been considered unrelated research directions. Existing solutions do not use topology integration as a quantifiable design target.

[0010] Second, there is a lack of standardization in the interfaces for heterogeneous energy sources. Heterogeneous distributed energy sources (photovoltaics, energy storage, wind power, and the power grid) differ significantly in voltage levels, frequency characteristics, polarity, and impedance characteristics. The existing plug-and-play concept still relies on the independent power electronic converters at the front end of each energy source in engineering implementation, and a complete algorithm chain for executable port detection, energy type identification, adaptive modulation mapping, and port registration has not been formed. Furthermore, the existing identification process is one-time, and it cannot be re-identified online when encountering changes in energy conditions during operation (such as photovoltaic shading, wind power speed changes, and sudden changes in the state of charge of energy storage), resulting in a mismatch between the modulation strategy and the actual operating conditions.

[0011] Third, there is insufficient suppression of multi-port circulating current and power coupling. In integrated topologies where magnetic components and switching devices are deployed simultaneously, dual internal coupling occurs between multiple ports through magnetic circuit coupling and common bridge arm reuse. When multiple energy sources are connected simultaneously, power coupling and circulating current between ports are easily triggered. Existing decoupling methods either rely on offline small-signal linearization models (such as dq decoupling based on fixed coupling coefficients) or adjust virtual impedance based on reactive power feedback, which are insufficient for adapting to large disturbance conditions (such as photovoltaic shading, sudden wind power changes, and load steps). At the same time, in integrated topologies, magnetic circuit equalization control and inter-port power decoupling control are designed separately, with each having its own closed loop, resulting in conflicting control objectives. Summary of the Invention

[0012] The technical problem to be solved by this invention is to provide a multi-port integration method for a comprehensive energy router suitable for multi-energy access, which solves the problems of port dispersion and linear expansion of hardware scale, lack of standardization of heterogeneous energy access interfaces, and insufficient suppression of multi-port circulating current and power coupling phenomena.

[0013] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a multi-port integration method for a comprehensive energy router suitable for multi-energy access, comprising the following steps: S1, topology layer integration, integrates multiple ports on the same set of shared switch bridge arm arrays and the same shared magnetic element at the physical topology level; S1-1, Design a multi-winding coupled transformer with shared magnetic elements; S1-2, configured with a shared switch bridge arm array, time-division multiplexing + carrier interleaving; S1-3, Design the topology reuse factor; S1-4, Calibrate K mag,0 Inverse calculation of open / closed circuit impedance in matrix; S2, access layer integration, standardizes heterogeneous energy interfaces into physical ports that can be uniformly connected at the access layer; S3, control layer integration, integrates distributed port controls into unified coordination and decoupling control at the control layer.

[0014] In S1-1, the shared magnetic element is a multi-winding coupled transformer, which contains at least four windings. Each winding is connected to one of the energy input port and the load or output port, and all windings are wound on the same magnetic circuit. The multi-winding coupled transformer has a magnetic circuit coupling coefficient K. mag .

[0015] In S1-2, the shared switch arm array includes multiple common arms, which are shared by at least two ports among the energy input port and the load and output ports in a time-division multiplexing manner; the time-division multiplexing rule is the switching cycle T of the shared switch arm array. sw Divided into N slot Each time slot is assigned to a group of ports of equal length; the common bridge arm array is interleaved modulated on the carrier phase, making N arm The carrier phase difference between each arm is 360° / N. arm The degree is used to average the magnetic flux pulsation of the magnetic core.

[0016] In S1-3, the topology reuse factor RF of the multi-port integrated topology core is not less than 1.5; the topology reuse factor RF, where N port N represents the total number of ports on the integrated energy router. arm RF represents the total number of bridge arms in the shared switch bridge arm array; when RF is 1.0, it represents the traditional design where each port occupies one bridge arm exclusively; when RF is 1.5, it represents that every 2 bridge arms are shared by 3 ports in a time-sharing manner, which is the lower boundary design point for shared switching devices.

[0017] In S1-4, the magnetic circuit coupling coefficient K mag The nominal coupling coefficient K between the i-th winding and the j-th winding mag,ij,0 The calibration method is obtained through joint back-calculation of short-circuit impedance and open-circuit impedance tests. It constitutes the step-by-step output of the S1 topology layer integration in stage S1, specifically: Z short,ij Z is the input impedance measured from the i-th winding under the condition that the j-th winding is short-circuited and the other windings are open-circuited; open,ij The input impedance is measured from the i-th winding with all windings open; sqrt is the square root operation; the measurement is completed during the factory commissioning phase, resulting in an N... w Line N w Column symmetric matrix K mag,0 , where N w The total number of windings is represented by the off-diagonal elements of this matrix, which characterize the initial magnetic circuit coupling between ports, while the diagonal elements are always 1.

[0018] Specifically, S2 includes: S2-1 samples the port voltage, port frequency, port polarity, and port impedance of the access energy input port through a plug-and-play access interface, and adds a small injected signal within the sampling window to measure the frequency response of the port impedance characteristics. S2-2, perform coarse identification of energy type; S2-3, based on the determined energy type, map the modulation method of the access port to the corresponding channel in the shared switch bridge arm array; S2-4: Write the energy type, port voltage, port frequency, port polarity, port impedance, and port impedance frequency response characteristics into the port attribute table as a record to complete the port registration; at the same time, encode the above characteristics into the access feature vector of the port and store it in the port access feature library. S2-5, Re-identification in runtime.

[0019] In S2-2, when the port frequency is no greater than 5 Hz and the port polarity is stable, the energy type is determined to be DC; when the port frequency is in the range of 45 Hz to 65 Hz, the energy type is determined to be AC ​​power frequency; when the port frequency exceeds the range of 45 Hz to 65 Hz and is time-varying, the energy type is determined to be AC ​​variable frequency.

[0020] In S3, specifically including, S3-1, Desired power command generation; S3-2, Online generation of dynamic decoupling matrix; based on the port attribute table and the magnetic circuit coupling coefficient K. mag Generate the decoupling matrix D(t) online; S3-3, the desired power command P is obtained through the decoupling matrix D(t). ref,i The mapping is represented by the duty cycle and phase shift angle of each common arm in the shared switch arm array, and the magnetic circuit equalization compensation command ΔΦ is superimposed. bal To avoid magnetization of the magnetic core; S3-4, Magnetic circuit equalization and collaborative compensation; Magnetic circuit equalization compensation command ΔΦ balThe generation depends not only on the flux integral residual, but also on the decoupling matrix D(t) and the desired power command P. ref .

[0021] S3 also includes, S3-5, periodically sample the actual power P at each port. meas,i (t) and port circulation i circ ; S3-6, Virtual Impedance Adaptive Feedback and Reverse Trigger Re-identification.

[0022] In S3-6, determine the port circulating current i. circ With rated current I n Is the ratio no greater than 3%, and determine the actual power P of each port? meas,i (t) and desired power command P ref,i Check if the error between them is within 2% of the expected power command; if the result is yes, maintain the current decoupling matrix D(t) and virtual impedance Z. v Enter steady-state operation; when the determination result is negative, dynamically adjust the virtual impedance Z according to the adaptive mapping law. v And return to S3-2.

[0023] The main beneficial effects of this invention are as follows: Simultaneously achieve magnetic component sharing and switching device sharing within the same physical topology, and propose quantifiable topology integration design indicators and engineerable modular expansion rules to solve the problems of port dispersion and linear expansion of hardware scale.

[0024] A complete algorithm chain is established for port attribute detection, energy type self-identification, adaptive modulation mapping, and port registration, and online re-identification based on historical access characteristics is supported, thereby solving the problem of lack of standardization of heterogeneous energy access interfaces.

[0025] Based on the online generation method of decoupling matrix using dynamic magnetic circuit coupling coefficient tensor and the virtual impedance adaptive feedback mechanism with magnetic circuit equalization collaborative compensation, a deeply coupled closed-loop structure of decoupling control and magnetic circuit equalization is formed, thereby solving the problem of insufficient suppression of multi-port circulating current and power coupling phenomena.

[0026] The hardware scale is sublinearly scalable, the power density is significantly improved, and modular expansion is feasible; heterogeneous energy sources are truly plug-and-play and support operational re-identification; a dynamic magnetic circuit coupling coefficient tensor is proposed for the first time under integrated topology; three-stage closed-loop coupling generates an overall gain that exceeds simple summation. Attached Figure Description

[0027] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0028] Figure 1This is an overall block diagram of the integrated energy router of the present invention.

[0029] Figure 2 This is a flowchart of the present invention.

[0030] Figure 3 This is a topology diagram of the shared switch bridge arm array and multi-winding coupled transformer of the present invention. Detailed Implementation

[0031] like Figures 1-3 This paper presents a multi-port integration method for integrated energy routers suitable for multi-energy access, consisting of three mutually cooperating stages: Stage S1 (topology layer integration), Stage S2 (access layer integration), and Stage S3 (control layer integration). Stage S1 integrates multiple ports onto the same set of shared switch bridge arm arrays and the same shared magnetic element at the physical topology level; Stage S2 standardizes heterogeneous energy interfaces into uniformly connectable physical ports at the access level; and Stage S3 integrates the dispersed port controls into unified coordination and decoupling control at the control level.

[0032] The product implementing this method is an integrated energy router, the structure of which is described in stage S1 below.

[0033] Example 1, An integrated photovoltaic, energy storage, and charging station with 8 source terminals and 4 negative terminals.

[0034] This embodiment takes an integrated photovoltaic, energy storage and charging station as a typical scenario, with a port scale of 12 ports: 8 source ports and 4 negative ports.

[0035] 1.1 Device Composition and Port Configuration like Figure 1 As shown, the integrated energy router in this embodiment includes an energy input port 1 (including 3 photovoltaic ports 11, 2 energy storage ports 12, 1 wind power port 13, and 1 grid port 14), a plug-and-play access interface 2, a multi-port integrated topology core 3 (including a shared switch bridge arm array 31 and a shared magnetic component 32), load and output ports 4 (including 1 AC load port, 1 DC load port, and 4 charging pile ports), and an inter-port power coordination and decoupling controller 5.

[0036] The rated parameters of each port are as follows: each photovoltaic port has a rated power of 250 kW and a wide voltage adaptation range of 400V to 1200V DC; each energy storage port has a rated power of 500 kW, a voltage of 600V DC, and is bidirectional; the wind power port has a rated power of 300 kW and a frequency converter AC input (frequency range of 20Hz to 80Hz); the grid port corresponds to the 380V public power grid; the DC load port is rated at 750V; the AC load port is rated at 380V; and each charging pile port is rated at 60 kW.

[0037] 1.2 Complete Implementation Process The multi-port integration method consists of three complementary phases: Phase S1 (topology layer integration), Phase S2 (access layer integration), and Phase S3 (control layer integration). Phase S1 integrates multiple ports at the physical topology level onto the same set of shared switch arm arrays and the same shared magnetic element. Phase S2 standardizes heterogeneous energy interfaces into uniformly connectable physical ports at the access layer. Phase S3 integrates distributed port controls into unified, coordinated, and decoupled control at the control layer. The product implementing this method is an integrated energy router, the structure of which is described in Phase S1 below.

[0038] This embodiment strictly follows the three-stage integration method described above: stage S1 topology layer integration, stage S2 access layer integration, and stage S3 control layer integration. Specific implementation data for each step in each stage are provided below.

[0039] Phase S1: Topology Layer Integration The integrated energy router includes an energy input port, a plug-and-play access interface, a multi-port integrated topology core, load and output ports, and an inter-port power coordination and decoupling controller. The multi-port integrated topology core consists of a shared switch bridge arm array and shared magnetic components.

[0040] In this embodiment, the energy input port 1 actually includes 3 photovoltaic ports 11, 2 energy storage ports 12, 1 wind power port 13 and 1 grid port 14 (a total of 8 source ports); the load and output port 4 actually includes 1 AC load port, 1 DC load port and 4 charging pile ports (a total of 4 negative ports); the inter-port power coordination and decoupling controller 5 is implemented based on a hardware unit that combines a field-programmable gate array and a digital signal processor.

[0041] The shared magnetic element is a multi-winding coupled transformer, which contains at least four windings. Each winding is connected to one of the energy input port and the load or output port, and all windings are wound on the same magnetic circuit. The multi-winding coupled transformer has a magnetic circuit coupling coefficient K. mag .

[0042] In this embodiment, the multi-winding coupled transformer 321 is configured as an eight-winding coupled transformer, with all windings wound on the same magnetic core 322 to form a common magnetic circuit; the magnetic core 322 is made of PM-type ferrite (material is manganese-zinc ferrite such as 3C95 or PC95, saturation magnetic flux density 0.5 Tesla, and peak operating magnetic flux density 0.25 Tesla); the number of turns of the winding corresponding to each port is N. pv Take 24 turns (photovoltaic port), N es Take 15 turns (energy storage port), N wt Take 18 turns (wind power port), N gridTake 12 turns (grid port), N dc Take 19 turns (DC load port), N cp Take 12 turns (charging pile port); the equivalent leakage inductance of each winding is typically 5 microhenries to 30 microhenries; the number of magnetic elements N in this embodiment mag There is 1.

[0043] A shared switch arm array comprises multiple common arms, which are shared by at least two ports among the energy input port and the load and output ports in a time-division multiplexing manner. The time-division multiplexing is specifically implemented according to the following rules: the switching cycle T of the shared switch arm array is... sw Divided into N slot Each of the three equal-length time slots is assigned to a group of ports. The same common bridge arm undertakes power conversion tasks for different ports in different time slots. The time slot allocation rule is based on the product of the port's rated power and instantaneous power command, allocated sequentially from largest to smallest, ensuring that high-power ports preferentially occupy bridge arm time slots, while low-power ports share the remaining time slots in a time-sharing manner. Simultaneously, the common bridge arm array undergoes interleaved modulation on the carrier phase, ensuring that N... arm The carrier phase difference between each arm is 360° / N. arm The degree is used to average the magnetic flux pulsation of the magnetic core.

[0044] In this embodiment, the shared switch arm array 31 includes four common arms 311a to 311d and two auxiliary arms, for a total of N arms. arm There are 6; each common bridge arm consists of two silicon carbide MOSFETs (labeled S). 1a With S 1b To S 4a With S 4b The midpoints of the bridge arms (nodes A to D) are connected to the corresponding windings (windings W1 to W4) in the multi-winding coupled transformer 321 via leads; the switching period T sw Take 20 microseconds (corresponding to a 50 kHz switching frequency) and divide it into N. slot Eight equal-length time slots (2.5 microseconds per slot) are used. The time slot allocation is sorted according to the product of rated power and instantaneous power command. Each 500 kW energy storage port occupies two time slots, each 250 kW photovoltaic port occupies one time slot, four 60 kW charging piles share one time slot, and the AC port is allocated several time slots within the AC half-cycle. The carrier phase of the six bridge arms differs by 60 degrees (360 degrees divided by 6), which increases the total magnetic flux ripple frequency to 300 kHz (i.e., 6 times the switching frequency) and reduces the iron loss of the magnetic core.

[0045] The topology multiplexing factor RF of the multi-port integrated topology core is not less than 1.5, and the topology multiplexing factor RF is defined by the following formula: Where N port N represents the total number of ports on the integrated energy router. armRF represents the total number of bridge arms in the shared switch bridge arm array. When RF is 1.0, it represents the traditional design where each port occupies one bridge arm exclusively; when RF is 1.5, it represents that every two bridge arms are shared by three ports in a time-sharing manner, which is the lower boundary design point for sharing switching devices. The physical significance of requiring RF to be no less than 1.5 is that: when RF is less than 1.5, the correspondence between ports and bridge arms is still close to the traditional one-to-one correspondence, and the hardware savings brought by sharing are insufficient to offset the control complexity cost introduced by time-sharing multiplexing; when RF is no less than 1.5, the hardware savings increase significantly, making the sharing scheme economically advantageous in engineering; through the dynamic magnetic circuit coupling coefficient tensor and closed-loop coupling control in the S3 stage control layer integration, the control complexity cost introduced by time-sharing multiplexing is limited to an acceptable range in engineering, thereby supporting stable operation of RF in the range of 1.5 to 3.0.

[0046] N in this embodiment port Equal to 12, N arm Equals 6, RF equals 12 divided by 6 equals 2.0, enters the topology integration range.

[0047] The integrated energy router has modular expansion capabilities, which are manifested in the following way: when expanding to the Nth... w When adding one port, a new winding is added to the same core of the multi-winding coupled transformer, wound symmetrically. Simultaneously, in the shared switch arm array, the decision to add a common arm depends on the existing number of arms and the rated power of the new port. The rule is: if the rated power of the new port is not greater than 0.5 times the existing maximum single-port rated power and RF is still not less than 1.5 after expansion, the new port uses the existing arm in a time-sharing manner, and no new arm is added; otherwise, one common arm is added. After expansion, short-circuit and open-circuit impedance tests are re-performed, and the magnetic coupling coefficient K is updated. mag The corresponding nominal values ​​are used; during the expansion process, the hardware wiring, control laws and port attribute tables of the existing ports do not need to be redesigned, and only the port attribute tables and time slot allocation tables are updated incrementally according to the rules.

[0048] Example of capacity expansion in this embodiment: When expanding the 9th port (e.g., adding a 100 kW energy storage port), the rated power of the new port of 100 kW is not greater than 0.5 times the existing maximum single port of 500 kW, i.e., 250 kW. After the expansion, the total number of ports becomes 13, the total number of bridge arms remains 6, and RF becomes 13 divided by 6, which is approximately equal to 2.17 and is still not less than 1.5. Therefore, the new port uses the existing bridge arms in a time-sharing manner and does not add any new bridge arms.

[0049] Magnetic circuit coupling coefficient K mag The nominal coupling coefficient K between the i-th winding and the j-th winding mag,ij,0 The calibration method is obtained through joint back-calculation of short-circuit impedance and open-circuit impedance tests. It constitutes the step-by-step output of the S1 topology layer integration in stage S1, specifically: Z short,ij Z is the input impedance measured from the i-th winding under the condition that the j-th winding is short-circuited and the other windings are open-circuited; open,ij This represents the input impedance measured from the i-th winding with all windings open; sqrt is the square root operation. Measurements are completed during the factory commissioning phase, yielding an N... w Line N w Column symmetric matrix K mag,0 , where N w K represents the total number of windings. The off-diagonal elements of this matrix characterize the initial magnetic circuit coupling between ports, while the diagonal elements are always 1. mag,0 The matrix serves as the output of stage S1 and provides reference data to step B2 in stage S3.

[0050] In this embodiment, K is measured according to the method. mag,0 The matrix is ​​an 8x8 symmetric matrix, with diagonal elements always equal to 1, off-diagonal elements ranging from 0.85 to 0.98, and an average value of 0.92. Measurements were performed using a vector network analyzer during the factory commissioning phase. Examples of submatrix values ​​between the first four windings are shown in the table below.

[0051] Phase S2: Access layer integration, i.e., multi-energy plug-and-play access. Phase S2 completes the interface standardization of heterogeneous energy through plug-and-play access interfaces, including steps A1 to A5.

[0052] Step A1: Sample the port voltage, port frequency, port polarity, and port impedance of the power input port via the plug-and-play access interface, and apply a small injected signal within the sampling window to measure the frequency response of the port impedance characteristics. The sampling window is 50 milliseconds to 200 milliseconds, and the injected signal amplitude is no greater than 1% of the rated voltage.

[0053] In this embodiment, the sampling window is 100 milliseconds; the amplitude of the injected signal is no greater than 1% of the rated voltage (for example, the peak-to-peak value of the injected signal at a 380-volt grid port is no more than 3.8 volts); and the frequency response of the port impedance is measured at three characteristic frequency points: 1 Hz, 10 Hz, and 100 Hz.

[0054] Step A2: Perform a coarse identification of the energy type. When the port frequency is no greater than 5 Hz and the port polarity is stable, the energy type is determined to be DC; when the port frequency is in the range of 45 Hz to 65 Hz, the energy type is determined to be AC ​​power frequency; when the port frequency exceeds the range of 45 Hz to 65 Hz and is time-varying, the energy type is determined to be AC ​​variable frequency.

[0055] In this embodiment: the photovoltaic port 11, energy storage port 12 and DC load port 42 are determined to be DC type because the port frequency is less than 5 Hz and the polarity is stable; the grid port 14 (frequency 50 Hz, polarity bipolar) is determined to be AC ​​power frequency type; the wind power port 13 (frequency range 20 Hz to 80 Hz time-varying) is determined to be AC ​​frequency converter type.

[0056] Step A2P further subdivides and identifies the ports classified as DC in Step A2. Based on the port voltage's volt-ampere characteristics, the port impedance characteristics in the 1 Hz to 100 Hz frequency band, and auxiliary parameters returned from the access side, DC ports are further subdivided into photovoltaic, energy storage, and DC load types: those with non-linear volt-ampere characteristics (current exhibits an inflection point at the maximum power point) and low impedance are classified as photovoltaic; those with nearly linear volt-ampere characteristics (current changes approximately proportionally to voltage) and the ability to return state-of-charge parameters from the access side are classified as energy storage; and those with monotonic volt-ampere characteristics that do not return state-of-charge parameters are classified as DC loads.

[0057] In this embodiment: For photovoltaic port 11, a small-signal sweep current of 1 mA (frequency from 1 Hz to 100 Hz) is injected. The measured impedance amplitude is 12 ohms at 10 Hz and the phase angle is approximately -15 degrees. The current-voltage characteristic is nonlinear under different DC biases, and it is determined to be photovoltaic. For energy storage port 12, the measured impedance amplitude is approximately 0.1 ohms at 10 Hz and the phase angle is close to 0 degrees. The current-voltage characteristic is close to linear, and the access side returns SOC parameters (value range 0 to 1), and it is determined to be energy storage. For DC load port 42, the measured impedance is monotonic resistive at 10 Hz and does not return SOC, and it is determined to be DC load.

[0058] Step A3: Based on the energy type determined in steps A2 and A2P, map the modulation method of the access port to the corresponding channel in the shared switch bridge arm array. When the energy type is photovoltaic, adaptive buck-boost modulation is used and maximum power point tracking is enabled; when the energy type is energy storage, adaptive buck-boost modulation is used and state-of-charge closed-loop modulation is enabled; when the energy type is AC power frequency, adaptive synchronous rectification is used and phase-locked loop is enabled; when the energy type is AC inverter, adaptive synchronous rectification is used and phase and frequency tracking is enabled. The input voltage adaptation range of adaptive buck-boost modulation is 100V to 1500V.

[0059] In this embodiment: the photovoltaic port 11 and the energy storage port 12 adopt adaptive buck-boost modulation (input voltage adaptation range of 100V to 1500V, which can cover the voltage range of 400V to 1200V for photovoltaic and 600V for energy storage); the grid port 14 adopts adaptive synchronous rectification and enables phase-locked loop; the wind power port 13 adopts adaptive synchronous rectification and enables phase and frequency tracking.

[0060] Step A4: Write the energy type, port voltage, port frequency, port polarity, port impedance, and port impedance frequency response characteristics into the port attribute table as a record to complete the port registration; at the same time, encode the above characteristics together into the access feature vector of the port and store it in the port access feature library.

[0061] The port attribute table uses an indexed value structure. Each record is indexed by the port's physical ID, and the value contains eight fields: Energy Type (enumerated type, values ​​are photovoltaic / energy storage / DC load / AC power frequency / AC frequency converter), Port Voltage (floating-point type, unit: volts), Port Frequency (floating-point type, unit: hertz), Port Polarity (enumerated type, values ​​are positive / negative / bipolar), Port Impedance Amplitude (floating-point type, unit: ohms), Phase Angle of Port Impedance at 1 Hz (floating-point type, unit: degrees), Phase Angle of Port Impedance at 10 Hz, and Phase Angle of Port Impedance at 100 Hz. The port attribute table is written every 100 milliseconds.

[0062] In this embodiment, the port attribute table is stored in the internal SRAM of the controller 5, occupying a total space of about 4 kilobytes, with each record occupying 22 bytes; the byte occupancy of the eight fields is as follows: energy type 1 byte (enumeration code 1 to 5), port voltage 32-bit floating point, port frequency 32-bit floating point, port polarity 1 byte, port impedance amplitude 32-bit floating point, and impedance phase angle at 1 Hz / 10 Hz / 100 Hz each 32-bit floating point.

[0063] The port access feature library is a fixed-length vector array. Each feature vector is a 64-dimensional floating-point vector, composed of eight normalized fields, with 8 redundant bits reserved for storing extended features. The feature library capacity is typically between 128 and 1024 entries, which are eliminated according to the least recently used strategy. When a feature vector is repeatedly hit during runtime re-identification, its access count is incremented, and those accessed more frequently are given priority for retention during elimination.

[0064] In this embodiment, the port access feature database has a capacity of 1024 entries; each feature vector consists of eight fields that have been standardized and normalized, and is eliminated according to the least recently used strategy, with those having higher access counts being retained first.

[0065] Step A5, Re-identification in Operational State. During the operation of the integrated energy router, the plug-and-play access interface performs secondary sampling on the port voltage, port frequency, port polarity, and port impedance of each port at a set operational state sampling period. The currently sampled feature vector is compared with the feature vectors already registered in the port access feature library, and the similarity is measured using Euclidean distance. If the distance is less than a set threshold (typically 0.15 per unit), the original registration result is maintained, and steps A1 to A4 are skipped; if the distance is greater than the set threshold, the re-identification process for that port is triggered, and steps A1 to A4 are re-executed in sequence, updating the port attribute table and the record of that port in the port access feature library.

[0066] In this embodiment, the sampling period for re-identification in running state is 1 second; the Euclidean distance threshold is a per-unit value of 0.15.

[0067] Phase S3: Control layer integration, i.e., power coordination and decoupling between ports. Phase S3 integrates the distributed port control into a unified coordination and decoupling control through the inter-port power coordination and decoupling controller, including steps B1 to C2.

[0068] Step B1, Desired Power Command Generation. Based on the upper-level dispatch command and at least one of the following: rated capacity of each port, energy storage state of charge, maximum power point tracking expectation, or grid dispatch command, generate the desired power command P for each port. ref,i , where i is the port number.

[0069] In this embodiment, the main loop cycle of controller 5 is 10 milliseconds; upper-layer scheduling instructions are received via the EtherCAT bus; the desired power instruction P... ref It is a 12-dimensional floating-point vector, corresponding to 12 ports.

[0070] Step B2: Online generation of the dynamic decoupling matrix. Based on the port attribute table and the magnetic circuit coupling coefficient K. mag Generate the decoupling matrix D(t) online. Specifically, the magnetic circuit coupling coefficient K... mag No longer using the factory-calibrated fixed value K mag,ij,0 Instead, it is rewritten as the dynamic magnetic circuit coupling coefficient tensor K, reflecting the operating point offset, according to the following formula. mag,ij (t): The meanings of the parameters are as follows: i and j are the port numbers, respectively; K mag,ij,0 The nominal magnetic coupling coefficient between port i and port j, obtained from step S1; ΔT ij (t) represents the real-time temperature rise difference between the winding corresponding to port i and the winding corresponding to port j, in degrees Celsius, which is sampled in real time by temperature sensors embedded near each winding; T ref For reference temperature rise, 40 degrees Celsius is taken as a typical value; P i (t) and P j (t) represents the measured instantaneous power of port i and port j at time t, in kilowatts (P). n,i With P n,j These are the rated power of the corresponding ports; SOC j (t) represents the energy storage charge state of port j at time t, taking a value from 0 to 1, with 0.5 as the neutral value for non-energy storage ports; α T α P α SThese represent the temperature correction weight, power correction weight, and state of charge correction weight, respectively; ψ is the asymmetric bivariate coupling correction function, defined as: Where sgn is the sign function, which reflects the influence of power directionality between ports on the coupling degree. It contributes positively when the per-unit power of port i is greater than that of port j, and negatively otherwise, giving the coupling correction direction-aware capability; η is the state-of-charge correction function, defined as: This function is bell-shaped, reaching a maximum value of 1 when the state of charge (SOC) is 0.5, and a minimum value of 0 when the SOC is 0 or 1. This reflects the actual characteristic that energy storage has the greatest impact on magnetic circuit coupling under intermediate charge states, and the least impact under saturated or depleted states. A typical value for the weighting parameter is α. T Take 0.05 to 0.20, α P Take 0.10 to 0.30, α S The value is taken from 0.02 to 0.10, and is determined based on the sensitivity of each port to changes in operating conditions and the measured sensitivity coefficient.

[0071] The derivation of the ψ function using the asymmetric form x multiplied by y multiplied by sgn(x min y) instead of other candidate forms (such as x multiplied by y, which is pure bilinear, min(x,y), which is taking the minimum value, and tanh(x min y), which is a smooth sign function) is based on the following: First, the x multiplied by y term reflects the product effect of the per-unit power of the two ports, corresponding to the product relationship of the coupling flux of the two ports under the small-signal bilinear approximation; Second, the sgn(x min y) term reflects the directionality of the power between the ports, which means that when the per-unit power of port i is greater than that of port j, the main magnetic flux contributed by port i in the magnetic circuit produces positive coupling enhancement to port j, and vice versa, it is negative decoupling. This asymmetry is ignored in the traditional symmetric bilinear approximation; Third, compared with the smooth sign function tanh, the discrete output of the sign function sgn is convenient for real-time calculation by field-programmable gate array hardware, and the engineering implementation cost is low.

[0072] The η function is expressed as a quadratic polynomial, which is 1 minus the square of (1 minus 2 times SOC). This is equivalent to 4 times SOC multiplied by (1 minus SOC). The derivation is based on the following: First, the equivalent internal resistance of the energy storage battery is the smallest and the charging and discharging efficiency is the highest when the battery is in the intermediate state of charge. It increases significantly when the state of charge is close to 0 or 1. The reciprocal of the equivalent internal resistance is proportional to the influence of magnetic circuit coupling. Therefore, the correction function should reach its maximum value when SOC is 0.5. Second, the advantage of the quadratic polynomial over other bell-shaped candidates (such as the sech square function and the negative cosine function) is that the operation is simple, involving only one subtraction, one multiplication, and one squaring, which can be completed in a single clock cycle on a field-programmable gate array. Third, the quadratic polynomial is strictly zero when SOC is 0 or 1, which strictly corresponds to the limiting case where the influence of magnetic circuit coupling is minimized when the battery is in a saturated or depleted state, with no approximation error.

[0073] The technical significance of this formula lies in dynamizing the factory-calibrated fixed coupling coefficient tensor, enabling the generation of the decoupling matrix to reflect in real time the effects of temperature drift, operating point offset, and changes in the energy storage state of charge on the magnetic circuit coupling, thereby maintaining the decoupling effect under large disturbance conditions. The derivation of this formula is based on the first-order approximate response of the magnetic circuit coupling coefficient to temperature, the asymmetric bilinear approximation of the power operating point, and the bell-shaped approximation of the state of charge. Each of these three approximations has a clear physical meaning and can be independently verified.

[0074] Based on the dynamic K mag,ij (t) Further construct the coupling relationship matrix M(t), where the element in the i-th row and j-th column M ij (t) is given by the following formula: Inverting the coupling matrix M(t) yields the decoupling matrix D(t): In this embodiment, the parameter value for step B2 is: temperature correction weight α. T Set to 0.10, power correction weight α P Take 0.20, and set the state of charge correction weight α. S Take 0.05, reference temperature rise T ref The temperature is set to 40 degrees Celsius; the temperature sensors of each port winding are K-type thermocouples with a sampling period of 100 milliseconds; the SOC of non-energy storage ports is set to 0.5 as the neutral value; the decoupling matrix D(t) has a dimension of 12 x 12 and an update period of 10 milliseconds.

[0075] An example illustrating the operation of the dynamic tensor formula: When the measured temperature rise at port 1 (photovoltaic port 11, channel 1) is 35 degrees Celsius and the measured temperature rise at port 2 (energy storage port 12, channel 1) is 30 degrees Celsius, ΔT 12 (t) equals 5 degrees Celsius, temperature correction term α T Multiply by ΔT 12 Divide by T ref =0.10 multiplied by 5 divided by 40 equals 0.0125; if the per-unit power of port 1 is 0.8 and the per-unit power of port 2 is 0.6 at that moment, then ψ(0.8,0.6) equals 0.8 multiplied by 0.6 multiplied by sgn(0.8 minus 0.6) equals 0.48, and the power correction term α P Multiplying by ψ equals 0.20, multiplying by 0.48 equals 0.096; if the SOC of port 2 is 0.5, then η(0.5) equals 1 minus (1 minus 1) squared equals 1, and the state of charge correction term α S Multiplying by η equals 0.05; the total of the three types of correction terms is approximately 0.158, therefore K mag,12 (t) equals K mag,12,0Multiplying by (1 plus 0.158) equals 0.92 multiplied by 1.158, which is approximately 1.07. This indicates that under this operating condition, the coupling between port 1 and port 2 is significantly enhanced, and a larger compensation weight needs to be given in the decoupling matrix.

[0076] This embodiment provides a 4x4 submatrix numerical example of M(t) and D(t) at a typical operating point (per-unit power of 0.8 at each port, temperature rise of 35 degrees Celsius at each port, and SOC of 0.5 at the energy storage port) (only the first four ports are shown). M(t) submatrix:

[0077] D(t) submatrix (obtained by inverting M(t)):

[0078] The complete D(t) matrix is ​​12 rows and 12 columns. The absolute values ​​of the diagonal elements are greater than 3.0 and the absolute values ​​of the off-diagonal elements are less than 3.0, forming a strictly diagonally dominated structure within the rows, which ensures that the matrix is ​​invertible.

[0079] Where M(t) 1 Let M(t) be the inverse matrix.

[0080] Step B3, the desired power command P is decoupled using the decoupling matrix D(t). ref,i The mapping is represented by the duty cycle and phase shift angle of each common arm in the shared switch arm array, and the magnetic circuit equalization compensation command ΔΦ is superimposed. bal To avoid magnetization of the magnetic core.

[0081] In this embodiment, the duty cycle resolution is 0.001, the phase shift angle resolution is 0.1 degrees, and the magnetic circuit equalization compensation command ΔΦ bal The duty cycle and phase shift angle of the D(t) output are superimposed and injected through an adder.

[0082] Step B4, Magnetic circuit equalization and collaborative compensation. Magnetic circuit equalization compensation command ΔΦ bal The generation depends not only on the flux integral residual, but also on the decoupling matrix D(t) and the desired power command P. ref Specifically, ΔΦ bal It is generated by the following formula: Where Φ meas (t) is the measured magnetic flux density vector, obtained in real time by magnetic flux sensors embedded at uniformly distributed locations in the magnetic core; Φ ref (t) represents the reference magnetic flux density vector; β1 and β2 are weighting coefficients, typically with β1 ranging from 0.5 to 1.0 and β2 ranging from 0.3 to 0.8; Φ targFor the target magnetic flux density, 50% of the core saturation magnetic flux density is taken as a typical value. The first term of the generative formula is the traditional flux integral residual feedback term, and the second term is β² multiplied by D(t) multiplied by P. ref (t) is the newly added coupling correction term. The introduction of the second term enables the magnetic circuit equalization compensation to proactively predict the impact of the desired power command on the magnetic circuit state through the decoupling matrix, avoiding the conflict between magnetic circuit equalization compensation and port power decoupling in traditional designs. Step B4 is coupled to step B2 through D(t), and to step B1 through P. ref Forming a coupling.

[0083] In this embodiment, the parameters for step B4 are: the weight β1 of the flux integration term is 0.8, and the weight β2 of the coupling correction term is 0.6; the flux density is sampled by four Hall flux sensors from four evenly distributed positions on the magnetic core 322, with a sampling period of 10 microseconds; the target flux density Φ targ Take 0.25 Tesla (i.e. 50% of the core saturation flux density).

[0084] Step C1: Periodically sample the actual power P at each port. meas,i (t) and port circulation i circ .

[0085] In this embodiment, the sampling period is 1 millisecond; port circulating current i circ Actual power P was measured using a circulating current transformer. meas It is obtained by integrating the product of the port voltage and the port current.

[0086] Step C2, virtual impedance adaptive feedback and reverse trigger re-identification. Determine port circulating current i. circ With rated current I n Is the ratio no greater than 3%, and determine the actual power P of each port? meas,i (t) and desired power command P ref,i Is the error between them within 2% of the expected power command? If the result is yes, maintain the current decoupling matrix D(t) and virtual impedance Z. v Enter steady-state operation; when the determination result is negative, dynamically adjust the virtual impedance Z according to the following adaptive mapping law. v And return to step B2: Among them, the circulating error signal e circ Defined as: Where i circ (t) represents the port circulating current measured at time t; i circ,target The target value for circulation is taken as 0 amperes under steady-state operation. ΔP ref (t) represents the rate of change of the desired power command, K P K I K FThese are the proportional gain, integral gain, and feedforward gain, respectively. When the virtual impedance is continuously N... cnt When the circulation is not suppressed to the target range in a sampling period (typically N) cnt (Take 50, corresponding to 50 milliseconds), triggering a reverse return to the running state re-identification in step A5. At this time, it is believed that the port may have experienced a sudden change in energy conditions, and it is necessary to enter the re-identification process. This makes steps A5, B2, B4, and C2 form a complete closed-loop coupling structure, and the working state of any link will affect other links in reverse.

[0087] In this embodiment, the parameter value for step C2 is: virtual impedance Z. v The initial value of the injection amplitude is 0.15 per unit; the proportional gain K P Set to 0.5; Integral gain K I Set to 0.05; Feedforward gain K F Take 0.3; circulation target i circ,target Set to 0 amperes; the number of consecutive sampling periods that fail to meet the threshold is N. cnt Take 50 (corresponding to 50 milliseconds) to trigger the reverse recognition of step A5 running state.

[0088] For example, to illustrate the reverse triggering mechanism: when the output of wind power port 13 fluctuates due to a sudden change in wind speed, controller 5 adjusts Z through the feedback law in step C2. v If the circulating current is not suppressed to the target range within 50 milliseconds, the controller 5 triggers step A5 in the reverse direction to re-identify the operating state of the wind power port 13, which may switch from the low-frequency subclass of AC frequency converter to the high-frequency subclass, thereby adjusting the modulation strategy of step A3.

[0089] 1.3 Test Data The operational test data of this embodiment shows that the overall efficiency is not less than 97% under full load conditions and not less than 95% under light load (20% load) conditions; the steady-state circulating current suppression (port circulating current i) is also high. circ With rated current I n The ratio of power consumption to output power consumption is no more than 1.0% (better than the traditional two-level control of 3.0%); the port power command tracking error is no more than 2%; the port first access identification time is no more than 200 milliseconds, the running state re-identification time is no more than 50 milliseconds; the operating condition change response time is no more than 20 milliseconds.

[0090] Example 2, A distributed power generation park access node with 4 source terminals and 2 negative terminals.

[0091] This embodiment provides another smaller-scale implementation of an integrated energy router, suitable for distributed generation nodes that include photovoltaic rooftops, small wind turbines, energy storage cabinets, and interfaces to the park's power grid. The method steps in this embodiment are the same as in Embodiment 1, with only supplementary explanations provided for parameter values ​​that differ from those in Embodiment 1; descriptions of the same method steps will not be repeated.

[0092] 2.1 Device Composition and Port Configuration The energy input ports of this embodiment include one photovoltaic port 11 (rated power 100 kW), one wind power port 13 (rated power 80 kW), one energy storage port 12 (rated power 150 kW), and one grid port 14 (380 volts power frequency); the load and output ports include one AC load port 41 and one DC load port 42.

[0093] 2.2 Complete Implementation Process Phase S1 Topology Layer Integration The steps of stage S1 are the same as in Example 1. The parameters for S1 in this example are: the multi-winding coupled transformer is configured as a four-winding coupled transformer, and the number of magnetic elements N. mag It consists of one unit, with a PM-type ferrite core; the shared switch arm array includes four common arms, with a total of N arms. arm There are 4; the switching frequency is 30 kHz; the total number of ports is N. port The value is 6; the topology reuse factor RF equals 6 divided by 4, which equals 1.5 (at the lower boundary of the topology integration interval); K is measured according to the calibration method. mag,0 The matrix is ​​a 4x4 symmetric matrix with an average value of 0.90.

[0094] Phase S2 Access Layer Integration The steps of stage S2 are the same as in Example 1. The parameters of S2 in this example are as follows: the sampling window in step A1 is 100 milliseconds; in step A3, photovoltaic port 11 and energy storage port 12 adopt adaptive buck-boost modulation (voltage adaptation range 100V to 1500V), grid port 14 adopts adaptive synchronous rectification and enables phase-locked loop, and wind power port 13 adopts adaptive synchronous rectification and enables phase and frequency tracking; the data structure of the port attribute table in step A4 is the same as in Example 1 (each record is 22 bytes); in step A5, the sampling period for re-identification of the running state is 1 second, and the Euclidean distance threshold is 0.15; the capacity of the port access feature library is 128 entries.

[0095] Stage S3 control layer integration The steps of stage S3 are the same as in Example 1. The parameters for S3 in this example are: Step B1: Controller main loop period 10 milliseconds; Step B2: Values ​​of the weight parameters α in the dynamic magnetic circuit coupling coefficient tensor formula. T Take 0.08, α P Take 0.15, αS Take 0.04, T ref The temperature is set to 40 degrees Celsius, the decoupling matrix D(t) has a dimension of 6x6 and an update period of 20 milliseconds; in step B4, the magnetic circuit equalization compensation weights β1 and β2 are set to 0.7 and 0.5 respectively; in step C2, the virtual impedance Z... v The initial value of the injection amplitude is taken as a per-unit value of 0.10, K. P Take 0.4, K I Take 0.04, K F Take 0.25, N cnt Take 50.

[0096] 2.3 Test Data The operational test data of this embodiment shows that: under full load conditions, the overall efficiency is not less than 97%, the steady-state circulating current suppression is not more than 1.2%, the port power command tracking error is not more than 2%, the initial access identification time is not more than 200 milliseconds, the running state re-identification time is not more than 60 milliseconds, and the response time to sudden changes in operating conditions is not more than 25 milliseconds.

[0097] Example 3, Variable operating condition robustness test scenario.

[0098] This embodiment has the same hardware topology as Embodiment 1 (8 source ports plus 4 negative ports, N...). port For 12, N arm The parameters (6, RF 2.0, eight-winding multi-winding coupled transformer, switching frequency 50 kHz) are used to verify robustness under large disturbance conditions. The method steps and parameter values ​​in this embodiment are the same as in Embodiment 1; the difference from Embodiment 1 is that the following sudden change scenario is added during operation.

[0099] 3.1 Complete Implementation Process The steps and parameter values ​​for stage S1 (topology layer integration), stage S2 (access layer integration), and stage S3 (control layer integration) in this embodiment are the same as in Embodiment 1 (see section 1.2 in Embodiment 1). This embodiment incorporates three concurrent mutation scenarios defined in Section 3.2 below during the execution of the complete method flow to verify robustness.

[0100] 3.2 Setting up scenarios for sudden changes in operating conditions Scenario 1 of sudden change in operating conditions: At time t = 0, the output of the first line of photovoltaic port 11 drops from 250 kW to 50 kW due to cloud cover. Scenario 2 of sudden change in operating conditions: At time t = 10 milliseconds, the frequency of wind power port 13 suddenly changes from 30 Hz to 45 Hz due to gusts (still belonging to AC frequency conversion, but the tracking parameters need to be switched in step A3). Scenario 3 of sudden change in operating conditions: When the first channel of energy storage port 12 is charged to 0.95 at t = 20 milliseconds, the constant voltage mode is switched.

[0101] 3.3 Control Response Process At time t=0, photovoltaic shading occurs. Controller 5 detects the difference between the measured power and the expected power at photovoltaic port 1 through multiple time slices within a 100-millisecond sampling period, triggering a correction to the dynamic magnetic circuit coupling coefficient tensor formula: the temperature correction term remains essentially unchanged, while the power correction term α... P Multiplying by ψ changes significantly due to the abrupt drop in P1(t) from 1.0 per unit to 0.2 per unit, decreasing from 0.096 to approximately -0.024 (note the directionality of the ψ function leading to sign reversal); the state-of-charge correction term remains unchanged. New K mag,1j The value of D(t) is reduced by about 12% from the original value, and the decoupling matrix D(t) is updated at time t = 10 milliseconds.

[0102] At time t = 10 milliseconds, a sudden change in wind power occurs, and the superimposed shading from the photovoltaic system has not fully recovered. Controller 5 simultaneously processes changes at both ports: the frequency change at wind power port 13 triggers the operational state re-identification step A5, which, after matching with the port's access feature library, maintains the AC frequency conversion class but updates the modulation parameters; changes at photovoltaic port 11 continue to be processed through dynamic K... mag,1j (t) is reflected in the decoupling matrix. Magnetic circuit equalization and collaborative compensation ΔΦ bal By multiplying β2 by D(t) and P ref The (t) term provides a forward-looking prediction of the combined effect of power changes at both ports on the magnetic circuit, thus avoiding core bias.

[0103] At time t = 20 milliseconds, the energy storage SOC switching occurs. Controller 5 updates η(SOC) accordingly. j Treatment of the term: When the SOC changes abruptly from 0.85 to 0.95, η(SOC) changes abruptly from 0.51 to 0.19, and the state-of-charge correction term changes significantly. Simultaneously, the virtual impedance Z... v If the circulating current is not suppressed to the target range within 50 milliseconds, a reverse return to step A5 is triggered for runtime re-identification.

[0104] At time t = 30 milliseconds, all control loops achieve steady-state convergence. Circulation suppression (i) circ with I n The steady-state value of the ratio decreased to 0.9%; the flux residual decreased to 1.3%; and the port power command tracking error decreased to 1.8%.

[0105] 3.4 Control group and verification of unexpected effects To compare and verify the overall effect, this embodiment also designed a control group consisting only of the basic scheme (without introducing dynamic K). mag,ij (t), without introducing the D(t) and P in step B4 refThe coupling terms are omitted, and the feedforward term and reverse triggering mechanism in step C2 are not introduced. In the control group, under the same three-condition concurrent abrupt change scenario, the steady-state convergence time is approximately 65 milliseconds, the steady-state circulation suppression is approximately 3.2%, and the steady-state magnetic flux residual is approximately 5.1%. This embodiment (including the dynamic tensor formula and closed-loop coupling mechanism) achieves a steady-state convergence time of approximately 30 milliseconds (54% faster response), a steady-state circulation suppression of 0.9% (a 72% reduction), and a steady-state magnetic flux residual of 1.3% (a 75% reduction). This comparison clearly shows that the combined application of the dynamic tensor formula and closed-loop coupling mechanism achieves an overall effect far exceeding the simple summation of the basic scheme.

[0106] Example Parameters and Scope of Application The table below summarizes the key parameter values ​​for the three embodiments. The parameter values ​​may be further adjusted within the defined range in other embodiments, without constituting a limitation on the scope of protection.

[0107]

[0108] In the above methods and embodiments, 1. Synchronous integration of shared magnetic components and shared switch bridge arms Instead of configuring independent converters for multiple energy input ports and load output ports, they are all connected to the same multi-port integrated topology core. This topology core includes a shared switch arm array and shared magnetic components. The shared magnetic components use multi-winding coupled transformers, with each winding wound on the same magnetic circuit. The common arm in the shared switch arm array is time-division multiplexed by multiple ports. This "dual-sharing" structure is the core difference between this and existing multi-port energy routers.

[0109] 2. Multi-port hardware integration rules based on topology multiplexing factor (RF) A topology reuse factor RF is proposed, with a minimum value of 1.5, to quantify the reuse relationship between the number of ports and the number of shared bridge arms. This feature demonstrates that it is not simply a matter of paralleling multiple ports, but rather achieving hardware size compression and power density improvement through the reuse relationship between ports and bridge arms. The common bridge arm is divided into time slots according to the switching cycle, and time-division multiplexing is performed based on the port's rated power and instantaneous power commands.

[0110] 3. Self-identification and adaptive modulation mapping of heterogeneous energy ports By sampling port voltage, frequency, polarity, impedance, and impedance frequency response through a plug-and-play access interface, the system identifies whether the access port belongs to the photovoltaic, energy storage, DC load, AC power frequency, or AC inverter type, and automatically matches the corresponding modulation method according to the energy type. For example, photovoltaic ports are matched with maximum power point tracking, energy storage ports with state-of-charge closed-loop modulation, AC power frequency ports with phase-locked loop modulation, and AC inverter ports with phase and frequency tracking.

[0111] 4. Runtime re-identification mechanism based on port access feature library Instead of identifying the energy type only upon initial connection, the system continuously samples port features during operation and compares the current feature vector with the port connection feature database for similarity. When port operating conditions change significantly, a re-identification process is automatically triggered to update the port attribute table and modulation method. This approach addresses control mismatch issues caused by changes in operating conditions such as photovoltaic shading, wind power speed variations, and sudden changes in energy storage SOC.

[0112] 5. Dynamic Magnetic Circuit Coupling Coefficient Tensor and Closed-Loop Decoupling Control The core of the control layer is: instead of using a fixed coupling coefficient for decoupling, it generates a dynamic magnetic circuit coupling coefficient tensor K online based on the port temperature rise difference, port power state, and energy storage state of charge. mag,ij(t) Then, a decoupling matrix D(t) is generated based on this, and magnetic circuit equalization compensation and virtual impedance adaptive feedback are combined to suppress port circulating current and power coupling. When virtual impedance feedback cannot effectively suppress circulating current, it can also trigger reverse operation state re-identification, thereby forming a closed-loop control between access identification, dynamic decoupling, magnetic circuit equalization and circulating current suppression.

[0113] In this application, the beneficial effects of the multi-port integration method include the following four aspects, which correspond to the core contributions of stage S1 topology layer integration, stage S2 access layer integration, stage S3 control layer integration, and the overall gain generated by the three-stage closed-loop coupling.

[0114] First, regarding the beneficial effects of the S1 topology layer integration: sublinear hardware scaling, significantly improved power density, and feasible modular expansion. Magnetic component sharing and switching device sharing are simultaneously achieved within the same physical topology, with a topology reuse factor (RF) of no less than 1.5 as a quantifiable design target. In a typical embodiment with 12 ports (8 source ports and 4 negative ports), the total number of bridge arms N... arm Only 6, number of magnetic components N mag With only one topology multiplexing factor (RF) of 2.0, the total number of switching devices is reduced by no less than 50% compared to the traditional design of one bridge per port, and the number of magnetic components is reduced by no less than 80% compared to the traditional design of independent inductors or transformers per port. The overall power density is increased by more than 40% compared to the existing technology that only performs magnetic integration. At the same time, the modular expansion rule means that adding new ports does not require redesigning the existing topology, and the expansion increment time is reduced by more than 70% compared to the traditional solution.

[0115] Second, regarding the beneficial effects of the S2 access layer integration: heterogeneous energy sources truly achieve plug-and-play functionality and support operational re-identification. Distributed energy sources such as photovoltaics, energy storage, wind power, and the power grid can be directly connected to the same physical port, eliminating the need for dedicated power electronic converters for each energy source. Simultaneously, through the operational re-identification mechanism in step A5, the re-identification process can be automatically triggered and the modulation method adjusted when the port's energy operating conditions change. The initial access identification time is no more than 200 milliseconds, and the operational re-identification time is no more than 50 milliseconds.

[0116] Third, regarding the beneficial effects of the S3 control layer integration: A dynamic magnetic circuit coupling coefficient tensor is proposed for the first time under the integrated topology. Dynamic magnetic circuit coupling coefficient tensor K mag,ij The tensor (d(t)) integrates three physical quantities: port temperature rise difference, port per-unit power ratio, and energy storage state of charge. Formally, it is presented as a combination of an asymmetric bivariate sign function ψ and a bell-shaped function η, which is mathematically completely different from existing dq decoupling methods based on fixed coupling coefficients and fixed virtual impedance adjustment methods. The introduction of this tensor enables the decoupling matrix D(t) to follow the operating point shift in real time, maintaining decoupling effectiveness under large disturbance conditions.

[0117] Fourth, the three-stage closed-loop coupling generates an overall gain that goes beyond simple summation. Steps A5, B2, B4, and C2 utilize the port attribute table, decoupling matrix D(t), and magnetic circuit equalization compensation command ΔΦ. bal Virtual impedance Z v The mutual references and reverse triggers between these components form a complete closed-loop coupling structure, where the operating state of any component will affect the others. Specifically, when using the dynamic magnetic circuit coupling coefficient tensor formula alone (step B2), the port circulating current suppression is improved from the traditional 3% to approximately 1.8%; when using magnetic circuit equalization collaborative compensation alone (step B4), the magnetic flux residual is reduced from the traditional 5% to approximately 2.5%; when using virtual impedance adaptive feedback with a feedforward term alone (step C2), the response time to sudden changes in operating conditions is reduced from the traditional 50 milliseconds to approximately 30 milliseconds; when using operational state re-identification based on the access feature library alone (step A5, but without forming a reverse trigger with step C2), the time for rematching the port attribute table after a sudden change in operating conditions is approximately 50 milliseconds. When the above four steps are applied together to form a reverse trigger closed loop between step A5 and step C2, and between step B2 and step B4 through the decoupling matrix D(t) and the desired power command P... refAfter bidirectional coupling, the port circulating current suppression is further reduced to 1.0%, the magnetic flux residual is reduced to 1.5%, the response time to sudden changes in operating conditions is reduced to 20 milliseconds, and the time for re-identifying the running state is reduced to 10 milliseconds. The overall benefits far exceed the simple summation of the four steps when applied individually. The fundamental reason for this unexpected effect is that under closed-loop coupling, each sudden change in operating conditions is responded to in parallel by multiple paths (step B2 predicts the change in coupling degree on the feedforward path, step B4 synchronously corrects it on the magnetic flux path, step C2 follows it in real time on the feedback path, and step A5 determines whether topology reconstruction is needed on the state machine path). Small deviations on any path will not accumulate and amplify, but will instead be synchronously canceled out by other paths. It should be noted that the topology layer integration in stage S1 serves as the physical basis for runtime control (the topology structure and the nominal magnetic circuit coupling coefficient K). mag,0 The matrix does not change during runtime, therefore it does not participate in the reverse triggering closed loop during runtime; the closed loop coupling specifically refers to the four-stage closed loop formed between stage S2 and stage S3 during runtime, consisting of steps A5, B2, B4, and C2.

[0118] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The embodiments and features described in these embodiments can be arbitrarily combined without conflict. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A multi-port integration method for a comprehensive energy router suitable for multi-energy access, characterized in that, Includes the following steps: S1, topology layer integration, integrates multiple ports on the same set of shared switch bridge arm arrays and the same shared magnetic element at the physical topology level; S1-1, Design a multi-winding coupled transformer with shared magnetic elements; S1-2, configured with a shared switch bridge arm array, time-division multiplexing + carrier interleaving; S1-3, Design the topology reuse factor; S1-4, Calibrate K mag,0 Inverse calculation of open / closed circuit impedance in matrix; S2, access layer integration, standardizes heterogeneous energy interfaces into physical ports that can be uniformly connected at the access layer; S3, control layer integration, integrates distributed port controls into unified coordination and decoupling control at the control layer.

2. The multi-port integration method for a comprehensive energy router suitable for multi-energy access according to claim 1, characterized in that: In S1-1, the shared magnetic element is a multi-winding coupled transformer, which contains at least four windings. Each winding is connected to one of the energy input port and the load or output port, and all windings are wound on the same magnetic circuit. The multi-winding coupled transformer has a magnetic circuit coupling coefficient K. mag .

3. The multi-port integration method for integrated energy routers applicable to multi-energy access as described in claim 1, characterized in that: In S1-2, the shared switch arm array includes multiple common arms, which are shared by at least two ports among the energy input port and the load and output ports in a time-division multiplexing manner; the time-division multiplexing rule is the switching cycle T of the shared switch arm array. sw Divided into N slot Each time slot is assigned to a group of ports of equal length; the common bridge arm array is interleaved modulated on the carrier phase, making N arm The carrier phase difference between each arm is 360° / N. arm The degree is used to average the magnetic flux pulsation of the magnetic core.

4. The multi-port integration method for a comprehensive energy router suitable for multi-energy access according to claim 1, characterized in that: In S1-3, the topology reuse factor RF of the multi-port integrated topology core is not less than 1.5; the topology reuse factor RF, where N port N represents the total number of ports on the integrated energy router. arm This represents the total number of bridge arms in the shared switch bridge arm array. When RF is 1.0, it represents the traditional design where each port occupies one bridge arm exclusively; when RF is 1.5, it represents the lower boundary design point where every two bridge arms are shared by three ports in a time-sharing manner, which is the sharing of switching devices.

5. The multi-port integration method for a comprehensive energy router suitable for multi-energy access according to claim 1, characterized in that: In S1-4, the magnetic circuit coupling coefficient K mag The nominal coupling coefficient K between the i-th winding and the j-th winding mag,ij,0 The calibration method is obtained through joint back-calculation of short-circuit impedance and open-circuit impedance tests. It constitutes the step-by-step output of the S1 topology layer integration in stage S1, specifically: Z short,ij Z is the input impedance measured from the i-th winding under the condition that the j-th winding is short-circuited and the other windings are open-circuited; open,ij The input impedance measured from the i-th winding under the condition that all windings are open; sqrt is the square root operation; the measurement was completed during the factory debugging phase, resulting in an N. w Line N w Column symmetric matrix K mag,0 , where N w The total number of windings is represented by the off-diagonal elements of this matrix, which characterize the initial magnetic circuit coupling between ports, while the diagonal elements are always 1.

6. The multi-port integration method for integrated energy routers applicable to multi-energy access according to claim 1, characterized in that: Specifically, S2 includes: S2-1 samples the port voltage, port frequency, port polarity, and port impedance of the access energy input port through a plug-and-play access interface, and adds a small injected signal within the sampling window to measure the frequency response of the port impedance characteristics. S2-2, perform coarse identification of energy type; S2-3, based on the determined energy type, map the modulation method of the access port to the corresponding channel in the shared switch bridge arm array; S2-4: Write the energy type, port voltage, port frequency, port polarity, port impedance, and port impedance frequency response characteristics into the port attribute table as a record to complete the port registration; at the same time, encode the above characteristics into the access feature vector of the port and store it in the port access feature library. S2-5, Re-identification in runtime.

7. The multi-port integration method for integrated energy routers applicable to multi-energy access as described in claim 6, characterized in that: In S2-2, when the port frequency is no greater than 5 Hz and the port polarity is stable, the energy type is determined to be DC; when the port frequency is in the range of 45 Hz to 65 Hz, the energy type is determined to be AC ​​power frequency; when the port frequency exceeds the range of 45 Hz to 65 Hz and is time-varying, the energy type is determined to be AC ​​variable frequency.

8. The multi-port integration method for integrated energy routers applicable to multi-energy access according to claim 1, characterized in that: In S3, specifically including, S3-1, Desired power command generation; S3-2, Online generation of dynamic decoupling matrix; based on the port attribute table and the magnetic circuit coupling coefficient K. mag Generate the decoupling matrix D(t) online; S3-3, the desired power command P is obtained through the decoupling matrix D(t). ref,i The mapping is represented by the duty cycle and phase shift angle of each common arm in the shared switch arm array, and the magnetic circuit equalization compensation command ΔΦ is superimposed. bal To avoid magnetization of the magnetic core; S3-4, Magnetic circuit equalization and collaborative compensation; Magnetic circuit equalization compensation command ΔΦ bal The generation depends not only on the flux integral residual, but also on the decoupling matrix D(t) and the desired power command P. ref .

9. The multi-port integration method for a comprehensive energy router suitable for multi-energy access according to claim 8, characterized in that: S3 also includes, S3-5, periodically sample the actual power P at each port. meas,i (t) and port circulation i circ ; S3-6, Virtual Impedance Adaptive Feedback and Reverse Trigger Re-identification.

10. The multi-port integration method for a comprehensive energy router suitable for multi-energy access according to claim 9, characterized in that: In S3-6, determine the port circulating current i circ With rated current I n Is the ratio no greater than 3%, and determine the actual power P of each port? meas,i (t) and desired power command P ref,i Check if the error between them is within 2% of the expected power command; if the result is yes, maintain the current decoupling matrix D(t) and virtual impedance Z. v Enter steady-state operation; when the determination result is negative, dynamically adjust the virtual impedance Z according to the adaptive mapping law. v And return to S3-2.

Images