A method and system for distributed power coordinated control of four-port converters for traction power supply in areas with abundant renewable energy sources and weak power grids.

By using a distributed power coordinated control method for four-port converters, the problem of voltage and frequency instability caused by rapid changes in the traction load of electrified railways and fluctuations in the output of new energy sources in weak power grid areas was solved, achieving power balance and stable operation, and improving the scalability and reliability of the system.

CN123121840APending Publication Date: 2026-11-17WUHAN UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611321263.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In weak power grid areas with high renewable energy penetration, rapid changes in the traction load of electrified railways and fluctuations in renewable energy output lead to sudden changes in DC bus voltage and voltage frequency shifts. Existing control methods are difficult to achieve power balance and stability, and centralized control relies on global information, which can easily lead to priority trade-offs.

Method used

A distributed power collaborative control method using four-port converters is adopted. Through information exchange with neighboring nodes, the flexible power range of the entire network is determined. Control is carried out in the order of traction power balance, minimization of renewable energy curtailment, and optimization of the operating status of adjustable resources. By combining the distributed coordination layer and the local fast control layer, DC bus voltage regulation, fast and slow power distribution of hybrid energy storage, and frequency support for weak grids are realized.

Benefits of technology

It improves the scalability and reliability of the system, achieves stable operation under weak grid conditions, balances economic optimization and dynamic response, and is suitable for bidirectional power adaptability of traction energy harvesting and regenerative braking feedback.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121840A_ABST
    Figure CN123121840A_ABST
Patent Text Reader

Abstract

This invention discloses a distributed power collaborative control method and system for four-port converters in traction power supply in areas with abundant renewable energy in weak power grids. The method includes: in each coordination cycle, firstly, calculating the power envelope of each device based on the battery state of charge, supercapacitor energy state, and prediction window; then, sequentially solving for the feasibility of traction hard power balance, minimization of renewable energy curtailment, and optimization of adjustable resource operation status according to a strict lexicographical order; in the operation status optimization, performing distributed mirror near-end correction iteration to obtain power reference values ​​for each port; using a local control cycle faster than the coordination cycle, superimposing the power reference values ​​with DC bus voltage correction, grid-type frequency-active power, and reactive power-voltage control quantities, and sending them to the corresponding ports, and feeding back the measured quantities to the next cycle. This invention takes into account traction power supply reliability, renewable energy absorption, energy storage state constraints, and dynamic support for weak power grids, and is applicable to bidirectional operating conditions of traction and regenerative feedback.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of electrified traction power supply collaborative control, specifically to a four-port converter distributed power collaborative control method and system for traction power supply in areas with abundant renewable energy sources in weak power grids. Background Technology

[0002] In regions with high renewable energy penetration, wind power, photovoltaic power, and other renewable energy sources are typically connected to the grid centrally via power electronic interfaces. When the AC system at the connection point has a small short-circuit capacity and a low short-circuit ratio, the system inertia and damping support capabilities are limited, exhibiting characteristics of a weak grid. Electrified railway traction loads are characterized by high power, rapid changes, and bidirectional switching between traction and regenerative braking, leading to impacts. Under conditions of fluctuating renewable energy output, sudden increases in traction load, or concentrated feedback of regenerative braking energy, source-load power imbalances can propagate rapidly between ports via the common DC bus and AC grid connection point, easily causing sudden changes in DC bus voltage and voltage and frequency deviations at the weak grid connection point. In severe cases, this can even cause port power exceeding limits or system oscillations. Connecting renewable energy ports, hybrid energy storage ports, weak grid ports, and traction ports to the common DC bus via a four-port converter can create an energy exchange channel for local renewable energy consumption and regenerative braking energy utilization. However, existing control systems typically focus on power tracking or energy balance, lacking a unified support mechanism to address the aforementioned coupled disturbances. Therefore, there is an urgent need for a four-port converter control method that combines the functions of stabilizing the voltage of the common DC bus, supporting the AC voltage and frequency at the grid connection point in weak grids, and coordinating the power constraints of multiple ports, in order to suppress voltage and frequency fluctuations caused by new energy fluctuations and impulsive traction loads, and improve the stable operation capability, new energy absorption capability, and regenerative braking energy utilization level of the traction power supply system under weak grid conditions.

[0003] Existing traction power supply energy management methods mostly employ centralized day-ahead optimization, intraday rolling optimization, or threshold-based operation mode switching. While these methods can coordinate the power of new energy sources and energy storage to some extent, they struggle to sequentially ensure traction power demand, new energy consumption, and the optimization of adjustable resource operation status during rapid changes in traction load, all within the same online control process. Furthermore, when various control objectives are processed simultaneously through weighted sum objective functions, trade-offs may arise between different control objectives due to weight selection.

[0004] Existing distributed control methods for multi-port converters typically employ droop control or consistency control to achieve power sharing. Consistency control can bring adjacent control nodes into a state of uniformity, but uniform node states do not necessarily correspond to optimal power allocation that satisfies the power boundaries of the equipment and the energy boundaries of the energy storage. For hybrid energy storage consisting of batteries and supercapacitors, the power limits, energy state, charge / discharge efficiency, and response speed of both must be considered simultaneously. For grid-connected ports connected to weak grids, active power capacity must be reserved for fast frequency support, and capacity constraints for both active and reactive power must be met.

[0005] Therefore, it is necessary to provide a distributed power cooperative control method applicable to the above-mentioned four-port traction power supply system, which can obtain the flexible power range of the entire network under the condition of only exchanging information between neighboring nodes, handle traction power balance, new energy curtailment and adjustable resource operation status according to a determined priority order, and make the power reference formed by the coordination layer cooperate with the port local fast control. Summary of the Invention

[0006] This invention addresses the problems of varying power constraints and dynamic characteristics at each port in traction power supply systems in renewable energy-rich areas of weak power grids, reliance on global information for centralized control, and the tendency for priority trade-offs to occur among multiple control objectives. It provides a distributed power collaborative control method and system for four-port converters in traction power supply systems in renewable energy-rich areas of weak power grids.

[0007] The present invention achieves the above objectives through the following technical solution. Firstly, the present invention provides a distributed power collaborative control method for a four-port converter in a region with abundant renewable energy sources and a weak power grid for traction power supply. The four-port converter includes a renewable energy port, a hybrid energy storage port, a grid-connected port, and a traction port coupled via a common DC bus. The hybrid energy storage port includes a battery branch and a supercapacitor branch. Each port is equipped with a control node, and each control node forms a connected communication graph through neighbor communication links. The method includes the following steps:

[0008] Step S1: In each coordination cycle, read the DC bus voltage, grid frequency, grid connection point voltage, active power at each port, battery state of charge, supercapacitor state of energy, available power of new energy sources, power curtailment, and traction power command from the previous coordination cycle.

[0009] Step S2: Determine the feasible power range of the battery branch and the supercapacitor branch based on the battery state of charge, the supercapacitor energy state, the energy storage power of the previous coordination cycle, and the preset prediction window; determine the feasible power range of the grid-type port based on the physical active power range and the fast frequency support reserve; each control node obtains the lower flexible power envelope and the upper flexible power envelope of the entire network through neighbor envelope consistency iteration.

[0010] Step S3: Taking traction power balance as the first priority, determine whether the traction power command is in the physical feasible domain based on the lower envelope and upper envelope of the entire network flexible power; when physically feasible, set the emergency relaxation diagnostic quantity to zero; when physically infeasible, output the emergency relaxation diagnostic quantity that represents insufficient power supply capacity or insufficient absorption capacity.

[0011] Step S4: Within the feasible set that satisfies the first priority, the minimum amount of abandoned renewable energy is taken as the second priority. The feasible range of renewable energy absorption power is determined based on the available renewable energy power and the flexible power envelope of the entire network, and the upper bound of the feasible range is selected as the reference value of renewable energy absorption power.

[0012] Step S5: After the first and second priorities are determined, the operation status optimization of the hybrid energy storage port and the grid-connected port is set as the third priority. Each control node performs distributed mirror near-end correction iteration based on the communication graph Laplace correction, node marginal quantity and local power constraints to obtain reference values ​​of battery power, supercapacitor power and grid port active power.

[0013] Step S6: Within a local control cycle faster than the coordination cycle, the power reference value is input to the local controller of each port respectively; the hybrid energy storage port corrects and distributes the battery power and supercapacitor power according to the DC bus voltage deviation; the grid-type port performs active power-frequency control, frequency recovery control, reactive power-voltage control and capacity limiting; and the traction port executes traction power command or regenerative feedback power command.

[0014] Preferably, the active power at each port is represented by a positive value indicating active power injected into the common DC bus, and by a negative value indicating active power absorbed from the common DC bus; the next predicted state of the battery in step S2 satisfies: The next predicted state of the supercapacitor satisfies: , , These represent the state of charge of the battery and the state of energy of the supercapacitor after a prediction window, respectively. , These represent the battery state of charge and the supercapacitor state of energy, respectively, during the current coordination cycle. , These represent the active power of the battery branch and the supercapacitor branch, respectively. , These represent the battery prediction coefficient and the supercapacitor prediction coefficient, respectively.

[0015] in and The power energy boundary is determined by the prediction window, the rated energy of the energy storage, and the charge / discharge efficiency frozen according to the power direction of the previous cycle, respectively. The control node is obtained by inverse solving of the SOC and SOE hard boundaries, and the intersection is taken with the rated power boundary of the strain gauge.

[0016] Preferably, the communication diagram is a four-node ring communication topology consisting of a new energy control node, a hybrid energy storage control node, a grid control node, and a traction control node. The hybrid energy storage control node and the grid control node respectively inject their local flexible power upper and lower bounds into the neighbor envelope consistency iteration. Each node uses the Metropolis hybrid matrix to perform a preset number of neighbor envelope consistency iterations. The new energy node recovers the entire network's flexible power upper and lower envelopes without obtaining the private objective function and state data of each port.

[0017] Preferably, the traction hard power balance satisfies:

[0018] ;

[0019] in, This indicates the actual power absorbed by the new energy source. This indicates the active power of the battery branch. This indicates the active power of the supercapacitor branch. Indicates the active power at the grid-type port. This indicates the traction port power command. This indicates an emergency slack measure in case of insufficient energy supply. Indicates an emergency relaxation dose for insufficient absorption;

[0020] Let the lower and upper bounds of the network's flexible power be respectively , Therefore, the feasible range for renewable energy absorption capacity is:

[0021] to ;

[0022] When the interval is not empty, let , and select to make It reaches the minimum value of the second level of strict lexicographical order.

[0023] Preferably, the objective function of the strict lexicographical third level includes a quadratic preference term for battery power, supercapacitor power, and active power at the grid port, a quadratic deviation term for battery state of charge and supercapacitor energy state relative to the target state, and a quadratic smoothing term relative to the power of the previous coordination cycle; after freezing the available power of new energy sources, traction power command, energy storage state, and power of the previous coordination cycle within a coordination cycle, the objective function is a strictly convex function for adjustable resource power.

[0024] Preferably, the adjacency matrix of the four-node ring communication topology is A, and the Metropolis hybrid matrix is... The mirror Laplace matrix is The step size parameters c and β of the distributed iteration satisfy... Positive definite, in each distributed inner iteration, the auxiliary variable y, the Laplace correction d, and the nodal marginal variable λ are updated sequentially, where Represents the identity matrix.

[0025] Preferably, in the distributed internal iteration, the hybrid energy storage node solves a two-dimensional box-constrained quadratic programming problem with battery power and supercapacitor power as variables, the grid node solves a scalar quadratic function and projects it onto the reserved active power range, and the new energy node and traction node maintain their active power unchanged after the first two levels of strict lexicographical order are determined; the two-dimensional box-constrained quadratic programming problem enumerates nine activity sets in which battery power and supercapacitor power are respectively in the lower bound, free state or upper bound, and selects the feasible optimal point according to the Karush-Kuhn-Tucker sign condition.

[0026] Preferably, the duration of the local control cycle is shorter than the duration of the coordination cycle; the hybrid energy storage port superimposes the optimized power reference with the DC bus voltage proportional-integral correction, and performs power allocation according to the slow dynamics of the battery and the fast dynamics of the supercapacitor; the grid-type port performs active-frequency dynamics, frequency proportional-integral recovery, reactive-voltage support, and other functions in parallel. Capacity circular limiting; a fast frequency support margin is reserved in the physical active power range of the grid-connected port; the strict lexicographical level 3 only generates a slow power reference in the active power range after deducting the fast frequency support margin, where P is... , represents the active power of the grid-type port, Q represents the reactive power of the grid-type port, and S represents the rated apparent power of the grid-type port.

[0027] Preferably, the battery state of charge has a hard lower bound, a soft lower bound, a soft upper bound, and a hard upper bound, and the supercapacitor state of energy has a hard lower bound, a soft lower bound, a soft upper bound, and a hard upper bound; the local security layer limits or reduces the charging and discharging power according to the boundaries; the neighbor envelope consistency iteration and the distributed mirror near-end correction inner iteration each have a preset number of iterations.

[0028] Secondly, the present invention provides a four-port converter distributed power coordination control system, including a new energy control node, a hybrid energy storage control node, a grid control node, a traction control node, a measurement and status feedback module, a distributed coordinator, and a local fast controller; the distributed coordinator is used to perform the above-mentioned three-level strict lexicographical order solution for traction power balance, minimization of new energy curtailment, and optimization of adjustable resource operating status; the local fast controller is used to perform DC bus voltage correction, hybrid energy storage fast and slow power allocation, weak grid frequency and voltage support, and port safety limiting.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] First, the physical feasibility of traction power commands is determined by first assessing the flexible power envelope of the entire network, and then the solution is performed in the order of traction power balance, minimizing the amount of abandoned renewable energy, and optimizing the operating status of adjustable resources, so as to avoid the trade-off between priorities of different control objectives when using a single weighted sum objective function.

[0031] Second, each control node exchanges envelope estimates, auxiliary variables, and node marginal quantities only through neighbor communication. The device power boundary, energy storage state, and local objective function are retained at the respective control nodes, eliminating the need for centralized collection of global information and improving the scalability and reliability of the system.

[0032] Third, by setting time-scales for the distributed coordination layer and the local fast control layer, it is possible to achieve DC bus voltage regulation, fast and slow power allocation of hybrid energy storage, and frequency and voltage support for weak grids while forming a feasible power reference, thus taking into account both economic optimization and dynamic response.

[0033] Fourth, the same control process is applicable to both traction power harvesting and regenerative braking feedback, exhibiting good bidirectional power adaptability. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the four-port converter traction power supply system and hierarchical control structure of the present invention;

[0035] Figure 2 A flowchart of a four-port converter distributed power collaborative control method for traction power supply in areas with abundant renewable energy in a weak power grid, provided by the present invention;

[0036] Figures 3 to 6 These are simulation diagrams of DC bus voltage, weak grid frequency, hybrid energy storage state, and four-port active power when the traction power command is +5 MW.

[0037] Figures 7 to 10These are simulation diagrams of DC bus voltage, weak grid frequency, hybrid energy storage state, and four-port active power when the regenerative feedback power command is -3 MW. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0039] like Figure 1 As shown, the four-port converter in this embodiment includes a new energy port, a hybrid energy storage port, a grid-connected port, and a traction port. The four ports exchange energy via a common DC bus. The new energy port is used to connect to wind or photovoltaic power generation devices along the line; the hybrid energy storage port includes a bidirectional battery power branch and a bidirectional supercapacitor power branch connected to the common DC bus; the grid-connected port connects to a 110 kV weak grid via a 10 kV converter side and a step-up transformer; and the traction port connects to a 27.5 kV traction network. The new energy control node, hybrid energy storage control node, grid control node, and traction control node are connected via neighbor communication links and are respectively connected to the local controller of their respective ports.

[0040] In this embodiment, the system base capacity is 10 MVA, the rated voltage of the common DC bus is 16 kV, the equivalent capacitance is 40 mF, the short-circuit ratio of the weak grid is 2.5, and the impedance-to-reactance ratio is 10. The rated active power of the new energy port is 3.6 MW; the rated power of the battery branch is 6 MW, and the rated energy is 12 MWh; the rated power of the supercapacitor branch is 4 MW, and the rated energy is 50 kWh; the rated apparent power of the grid-type port is 2.5 MVA, and the physical active power range is -2 MW to 2 MW; the rated apparent power of the traction port is 6.4 MVA. The charging efficiency and discharging efficiency of both the battery and the supercapacitor are taken as 0.95.

[0041] This embodiment defines active power injected into the common DC bus as positive, and active power absorbed from the common DC bus as negative. Traction power during traction energy harvesting. When it is positive, regenerative braking feedback Negative; when the battery branch and supercapacitor branch are discharging. , It is positive when charging and negative when the power grid supplies power to the common DC bus. When the common DC bus supplies power to the grid, it is positive. It is negative. The average power dynamics of the common DC bus satisfy:

[0042] ;

[0043] in, This represents the equivalent capacitance of the common DC bus. This indicates the voltage of the common DC bus.

[0044] This embodiment uses a port average power model to verify the distributed power coordination process, without expanding the switching transients and harmonics of the converter.

[0045] like Figure 2 As shown, the distributed power cooperative control method of this embodiment is implemented according to the following steps.

[0046] Step S1: Using a coordination period of 10 ms, collect the following data from the previous coordination period: DC bus voltage, grid frequency, per-unit voltage at grid connection point, actual power of new energy sources, total power of hybrid energy storage, actual active power of the grid, actual active power of traction, battery state of charge (SOC), supercapacitor state of energy (SOE), actual power of the battery, actual power of the supercapacitor, instantaneous power mismatch of the target, coordination status, available power of new energy sources, and power curtailment, forming a 15-dimensional measurement vector. This measurement vector is then input to the distributed coordinator after a delay of one coordination period.

[0047] Step S2: Determine the feasible power range of the hybrid energy storage port based on the energy storage status and the preset prediction window. Prediction window T h Take 0.5 s, and Efficiency coefficient in MW and Based on the charging and discharging direction determined in the previous coordination cycle, the prediction coefficients for the battery and supercapacitor, as well as the next predicted state, satisfy the following conditions:

[0048] ;

[0049] ;

[0050] In the formula, and These represent the rated energy of the battery and the rated energy of the supercapacitor, respectively. The power boundary corresponding to the energy storage capacity is obtained by inverse solving the predicted state boundaries of SOC and SOE, and then intersected with the rated power range of -6 MW to 6 MW for the battery branch and the rated power range of -4 MW to 4 MW for the supercapacitor branch, respectively. The hard lower bound, soft lower bound, soft upper bound, and hard upper bound of the battery SOC are 20%, 30%, 70%, and 80%, respectively; the hard lower bound, soft lower bound, soft upper bound, and hard upper bound of the supercapacitor SOE are 25%, 30%, 70%, and 75%, respectively. In this embodiment, the initial values ​​of SOC and SOE are both 50%.

[0051] Step S3: The new energy control node, hybrid energy storage control node, grid control node, and traction control node form a four-node ring communication topology without communication faults, with its adjacency matrix A, Metropolis hybrid matrix W, and mirrored Laplace matrix L. m They are respectively:

[0052] ;

[0053] ;

[0054] ;

[0055] Each coordination cycle first performs 30 neighbor envelope consistency iterations. The hybrid energy storage control node injects the flexible power upper and lower bounds of the battery branch and the supercapacitor branch into the iterations, while the grid control node injects the active power upper and lower bounds after deducting the fast frequency support reserve into the iterations. The new energy control node obtains the lower flexible power envelope of the entire network based on the envelope estimates transmitted by the neighbor nodes. Flexible power envelope across the entire network .

[0056] Step S4 involves performing a three-level strict lexicographical order solution in the order of traction power balance, minimizing renewable energy curtailment, and optimizing the operational status of adjustable resources. The first level of traction power balance is:

[0057] ;

[0058] in, This indicates the actual power absorbed by the new energy source. This indicates the active power of the battery branch. This indicates the active power of the supercapacitor branch. Indicates the active power at the grid-type port. This indicates the traction port power command. This indicates an emergency slack measure in case of insufficient energy supply. Indicates an emergency relaxation dose for insufficient absorption;

[0059] The feasible range for renewable energy absorption capacity is:

[0060] to ;

[0061] in, This represents the lower bound of the renewable energy absorption capacity. This represents the upper limit of the renewable energy absorption capacity. This indicates the available power of new energy sources.

[0062] then At MW, the traction power command is in the physically feasible region, which will and Set to zero; take the second level The amount of abandoned renewable energy is: The amount of abandoned renewable energy is: .

[0063] Thus, the minimum renewable energy curtailment power is obtained within the feasible set that satisfies traction power balance.

[0064] When the feasible range of renewable energy absorption capacity is empty, select the boundary power with the smallest distance from the physical feasible region, and then... or This indicates insufficient energy supply capacity or insufficient absorption capacity.

[0065] Step S5, determining the first and second levels and After that, with , and Establish a third-level objective function for the adjustable variables:

[0066] ;

[0067] Where i∈{bat, sc, grid} The corresponding power in the previous coordination cycle, For power preference weights, For power smoothing weights, , These represent the deviation weights of SOC and SOE, respectively. , The corresponding target values ​​are set. In this embodiment, the power preference weights are set to [0.20, 1.00, 0.30, 2.00, 0.20] and the power smoothing weights are set to [0.20, 1.50, 0.40, 4.00, 0.20], following the order of [new energy, battery, supercapacitor, power grid, curtailment]. Pick , Pick ,, , Both are taken as 50%. The grid-type port reserves 1 MW of fast frequency support capacity within the physical active power range of -2 MW to 2 MW, and the slow power reference range formed by the distributed coordinator is -1 MW to 1 MW.

[0068] Each coordination cycle executes 240 distributed mirror near-end correction iterations, with step size parameters c and β both set to 0.05. Let the node power vector... Let the marginal quantity of a node be λ, and the auxiliary variable be y. The ℓth inner iteration is updated according to the following formula:

[0069] ;

[0070] ;

[0071] ;

[0072] The hybrid energy storage control node is solved in each inner iteration. and A two-dimensional box-constrained quadratic programming problem is used, where the battery power and supercapacitor power are in the lower bound, free state, or upper bound, respectively. Nine activity sets are formed by enumerating these sets, and the feasible optimal point is selected based on the Karush-Kuhn-Tucker conditions. The grid control nodes solve a scalar quadratic function, and the resulting power is projected onto a slow power reference interval of -1 MW to 1 MW. The step size parameter satisfies: .

[0073] In the four-node ring communication topology of this embodiment, L m The largest eigenvalue is 2 / 3, therefore The minimum eigenvalue lower bound is 0.01667, which is greater than zero. The distributed coordinator also outputs the optimized power balance residual, the node marginal quantity range, the emergency relaxation diagnostic quantity, and the lexicographical order feasibility flag, which are used by the local controller to determine the operating status.

[0074] Step S6: Perform fast control of each port with a local control cycle of 10 μs. The hybrid energy storage local controller superimposes the power reference output from the distributed coordinator with the proportional-integral correction of the DC bus voltage, and the proportional gain of the DC bus voltage Kp is... vdc Pick Integral gain Ki vdc Pick The anti-saturation coefficient Kaw is set to 5; the slow-channel time constant of the battery is set to 0.20 s, and the fast-channel time constant of the supercapacitor is set to 0.01 s. The frequency recovery proportional gain Kp at the grid-type port is also set. f Pick Integral gain Ki f Pick Reactive power - voltage gain The dynamic time constant is taken as 0.01 s; the equivalent inertia constant of the weak grid is taken as 2.0 s, and the frequency damping coefficient is taken as... W / Hz. The active and reactive power at the grid port satisfy:

[0075] ;

[0076] In the formula, Indicates the reactive power at the grid port. This indicates the rated apparent power of the grid-type port. Traction ports are classified according to... The bidirectional average active power is executed, and no 100Hz active power ripple is injected into the traction main channel.

[0077] To verify the above control method, a fixed-step discrete average power model was used for simulation, with a base step size of 10 μs and a simulation duration of 8 s. At the initial moment of the simulation, the available power of the renewable energy source was 1.5 MW, and the state of charge (SOC) of the battery and the state of energy efficiency (SOE) of the supercapacitor were both 50%. At 1.0 s, the available power of the renewable energy source was increased from 1.5 MW to 2.5 MW. At 2.0 s, a 2MW local load with a power factor of 0.90 and exhibiting lag was introduced to the weak grid side. At 4.5 s, a +5 MW traction power command and a -3 MW regenerative feedback power command were applied, respectively. At 6.0 s, the available power of the renewable energy source was increased from 2.5 MW to 3.0 MW. At 7.0 s, the available power of the renewable energy source was decreased from 3.0 MW to 2.0 MW.

[0078] like Figures 2 to 5 The figures show simulation diagrams of DC bus voltage, weak grid frequency, hybrid energy storage state, and active power at four ports when the traction power command is +5 MW. After the local load on the weak grid side is connected at 2.0 s, the grid frequency drops to a minimum of 49.87249 Hz and recovers to 50 Hz. After the traction power command is applied at 4.5 s, the output power of the new energy port is approximately 2.5 MW, the hybrid energy storage port switches to discharge state, and the power balance of each port is re-established. Under this condition, the peak deviation of the common DC bus voltage is 0.19855%, and the battery SOC is 49.97416% and the supercapacitor SOE is 49.81216% at the end of the simulation.

[0079] Figures 6 to 9 These are simulation diagrams of DC bus voltage, weak grid frequency, hybrid energy storage state, and four-port active power when the regenerative feedback power command is -3 MW. At 4.5 s, P... trac When the power output drops to -3 MW, the hybrid energy storage port switches to charging mode. The supercapacitor branch handles rapid power absorption, while the battery branch handles slower energy absorption. Under this condition, the peak deviation of the common DC bus voltage is 0.11652%, and at the end of the simulation, the battery SOC is 50.03658% and the supercapacitor SOE is 50.49986%.

[0080] In this embodiment, the port rating, prediction window, control cycle, number of iterations, and control gain can be adjusted according to the actual traction load capacity, weak grid strength, energy storage capacity, and controller computing power; the communication topology can also adopt other neighbor communication topologies that meet the connectivity and hybrid matrix weight conditions.

[0081] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A distributed power collaborative control method for a four-port converter for traction power supply in areas with abundant renewable energy sources and weak power grids, wherein the four-port converter includes a renewable energy port, a hybrid energy storage port, a grid-connected port, and a traction port coupled via a common DC bus, the hybrid energy storage port including a battery branch and a supercapacitor branch, each port is configured with a control node with local computing capabilities and forms a connected communication graph through neighbor communication links, characterized in that, include: S1. In each coordination cycle, read the DC bus voltage, grid frequency, grid connection point voltage, active power at each port, battery state of charge, supercapacitor state of energy, available power of new energy sources, abandoned power and traction power commands of the previous coordination cycle. S2. Based on the battery state of charge, supercapacitor energy state, power of the previous coordination cycle, and prediction window, calculate the feasible power envelopes of the battery branch, supercapacitor branch, and grid-type port respectively, and obtain the flexible power envelope of the entire network through the neighbor iteration of the communication graph. S3. Solve the feasibility of traction hard power balance, the minimization of new energy curtailment, and the operation state optimization of hybrid energy storage port and grid-connected port in strict lexicographical order, wherein the emergency relaxation diagnostic quantity is only enabled when the traction hard power balance exceeds the physical feasible region. S4. In the operation state optimization, each control node performs distributed mirror near-end correction iteration based on the communication graph Laplace correction, node marginal quantity and local box constraint subproblem to obtain reference values ​​of new energy consumption power, battery power, supercapacitor power and grid port active power. S5. Using a local control cycle faster than the coordination cycle, the power reference value is superimposed with the DC bus voltage correction, the grid-type active-frequency control, the reactive-voltage control, and the port safety limit, and then sent to the corresponding port. The measured value after execution is fed back to the next coordination cycle.

2. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The active power at each port is represented by a positive value for the active power injected into the common DC bus, and by a negative value for the active power absorbed from the common DC bus. In step S2, the next predicted state of the battery satisfies The next predicted state of the supercapacitor satisfies , , These represent the state of charge of the battery and the state of energy of the supercapacitor after a prediction window, respectively. , These represent the battery state of charge and the supercapacitor state of energy, respectively, during the current coordination cycle. , These represent the active power of the battery branch and the supercapacitor branch, respectively. , These represent the battery prediction coefficient and the supercapacitor prediction coefficient, respectively. in and The power energy boundary is determined by the prediction window, the rated energy of the energy storage, and the charge / discharge efficiency frozen according to the power direction of the previous cycle, respectively. The control node is obtained by inverse solving of the SOC and SOE hard boundaries, and the intersection is taken with the rated power boundary of the strain gauge.

3. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The communication graph is a four-node ring communication topology without communication faults, consisting of new energy nodes, hybrid energy storage nodes, grid nodes and traction nodes. Each node uses the Metropolis hybrid matrix to perform a preset number of neighbor envelope consistency iterations. Hybrid energy storage nodes and grid nodes inject their respective adjustable power upper and lower bounds. New energy nodes restore the flexible power upper and lower envelopes of the entire network without obtaining the private objective function and state data of each port.

4. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 3, is characterized in that... The adjacency matrix of the four-node ring communication topology is A, and the Metropolis hybrid matrix is... The mirror Laplace matrix is The step size parameters c and β of the distributed iteration satisfy... Positive definite, in each distributed inner iteration, the auxiliary variable y, the Laplace correction d, and the nodal marginal variable λ are updated sequentially, where Represents the identity matrix.

5. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 4, is characterized in that... In the distributed internal iteration, the hybrid energy storage node solves a two-dimensional box-constrained quadratic programming problem with battery power and supercapacitor power as variables, the grid node solves a scalar quadratic function and projects it onto the reserved active power range, and the new energy node and traction node maintain their active power unchanged after the first two levels of strict lexicographical order are determined; the two-dimensional box-constrained quadratic programming problem enumerates nine activity sets in which battery power and supercapacitor power are respectively in the lower bound, free state or upper bound, and selects the feasible optimal point according to the Karush-Kuhn-Tucker sign condition.

6. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The traction hard power balance satisfies: ; in, This indicates the actual power absorbed by the new energy source. This indicates the active power of the battery branch. This indicates the active power of the supercapacitor branch. Indicates the active power at the grid-type port. This indicates the traction port power command. This indicates an emergency slack measure in case of insufficient energy supply. Indicates an emergency relaxation dose for insufficient absorption; Let the lower and upper bounds of the network's flexible power be respectively , Therefore, the feasible range for renewable energy absorption capacity is: to ; in, This represents the lower bound of the renewable energy absorption capacity. This represents the upper limit of the renewable energy absorption capacity. Indicates the available power of new energy sources; When the interval is not empty, let , and select to make It reaches the minimum value of the second level of strict lexicographical order.

7. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The objective function of the strict lexicographical third level includes a quadratic preference term for battery power, supercapacitor power, and active power at the grid port; a quadratic deviation term for battery state of charge and supercapacitor energy state relative to the target state; and a quadratic smoothing term relative to the power of the previous coordination cycle. After freezing the available power of new energy sources, traction power command, energy storage state, and power of the previous coordination cycle within a coordination cycle, the objective function is a strictly convex function for adjustable resource power.

8. The distributed power collaborative control method for four-port converters in traction power supply in areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The duration of the local control cycle is shorter than the duration of the coordination cycle; the hybrid energy storage port superimposes the optimized power reference with the DC bus voltage proportional-integral correction, and performs power allocation according to the slow dynamics of the battery and the fast dynamics of the supercapacitor; the grid-type port performs active-frequency dynamics, frequency proportional-integral recovery, reactive-voltage support, and other functions in parallel. Capacity circular limit; The physical active power range of the grid-connected port is reserved with a fast frequency support margin. The strict lexicographical level 3 only generates a slow power reference within the active power range after deducting the fast frequency support margin, where P is... , represents the active power of the grid-type port, Q represents the reactive power of the grid-type port, and S represents the rated apparent power of the grid-type port.

9. The distributed power collaborative control method for four-port converters in traction power supply for areas with abundant renewable energy in weak power grids, as described in claim 1, is characterized in that... The battery state of charge has a hard lower bound, a soft lower bound, a soft upper bound, and a hard upper bound; the supercapacitor state of energy has a hard lower bound, a soft lower bound, a soft upper bound, and a hard upper bound; the local security layer limits or reduces the charging and discharging power according to the corresponding boundaries; the neighbor envelope consistency iteration and the distributed mirror near-end correction internal iteration each have a preset number of iterations.

10. A four-port converter distributed power collaborative control system for traction power supply in areas with abundant renewable energy sources and weak power grids, characterized in that, include: New energy ports, hybrid energy storage ports, grid-connected ports, and traction ports are coupled via a common DC bus; new energy control nodes, hybrid energy storage control nodes, grid control nodes, and traction control nodes are respectively set up with each port and exchange coordination quantities through a neighbor communication diagram; The measurement and status feedback module is used to generate measurement vectors including DC bus status, weak grid status, port power, energy storage status, and available power of new energy sources. The distributed coordinator is used to perform three-level strict lexicographical ordering for traction hard power balance, minimization of new energy curtailment, and optimization of adjustable resource operation status. It generates power references for each port through distributed iteration based on neighbor Laplace correction, mirror near-end box constraint solution, and node marginal quantity correction. Local fast controllers are used to perform DC bus voltage correction, hybrid energy storage fast and slow distribution, grid-type frequency and voltage support, and port safety limiting in a cycle faster than that of distributed coordinators.