Multi-source electric energy collaborative scheduling method and system based on spatial topological constraint

Abstract

The present application relates to the technical field of power system dispatching, and discloses a multi-source electric energy collaborative dispatching method and system based on spatial topology constraints, comprising: collecting load operation data and spatial topology connection information, combining physical communication node distance to construct a safe voltage tolerance corridor and draw a spatial voltage drop straight line; determining the control parameters such as the load list that needs to be cut by judging the projection of the spatial voltage drop straight line in the safe voltage tolerance corridor; constructing a load physical topology multi-way tree, and calculating the cumulative inductive distance value of each leaf node; performing multi-source power compensation operation according to the set control parameters, performing spatial sequential load cutting on the load list that needs to be cut according to the order from large to small of the cumulative inductive distance value, and performing closed-loop calibration based on the actual voltage value of the bus after cutting; the present application realizes continuous voltage safety checking and load timing cutting in the spatial dimension, and avoids transient voltage overshoot caused by voltage out-of-limit of the power supply line and concentrated load cutting.

Description

Technical Field

[0001] This invention relates to the field of power system dispatching technology, and more specifically, to a multi-source power coordinated dispatching method and system based on spatial topology constraints. Background Technology

[0002] With the development of distributed energy and microgrid technologies, multi-source power supply networks are widely used in zero-carbon industrial parks and municipal infrastructure. To ensure the continuous operation of critical equipment under disturbances such as grid sags or power imbalances, power dispatching systems need to coordinate and dynamically switch the power of energy storage, photovoltaics, and various power-consuming terminals in real time. Reasonable power coordination control and topology timing scheduling are core elements for maintaining stable bus voltage and preventing system-level power outages.

[0003] Existing power dispatching methods have been researched in the areas of multi-energy coordination and topology timing control. For example, CN121440579A discloses a voltage control method and system for multi-energy storage in distribution networks based on topology timing coordination. This method, based on the distribution network topology and node distance matrix, calculates the priority and charging power of each energy storage node and determines the action sequence by combining action delays, thus achieving local coordinated control of multiple energy storage systems and voltage fluctuation mitigation. CN121663545A discloses a method, system, equipment, and medium for automatic power control of multiple energy types. It constructs a two-layer market optimization model based on the physical characteristics and operational constraints of multiple entities, solves the output dispatching command set, and drives various energy supply units to perform power regulation, realizing coordinated dispatching of generator sets, energy storage, and adjustable loads under physical constraints.

[0004] However, in long-distance, multi-branch DC power supply networks such as large industrial parks and municipal roads, when a deep voltage sag occurs in the power grid accompanied by insufficient power from multiple sources, existing dispatching methods typically prioritize and centrally cut off non-critical loads to maintain power supply to sensitive loads. When multiple high-power loads on the same power supply branch are simultaneously disconnected within a short period, the line load current changes rapidly, causing transient voltage fluctuations on local busbars. This can lead to abnormal restarts, protection activations, or device damage to sensitive loads that were originally prioritized for power supply. Therefore, how to reduce the transient electrical impact on remaining loads caused by concentrated load shedding while reducing the power shortage is a problem that needs to be solved in the dispatching of multi-source power supply networks. Summary of the Invention

[0005] The aforementioned transient electrical surges are not only related to power shortages but also to rapid changes in line current and the spatial topology of the power supply network. Cables in long-distance, multi-branch DC power supply networks have accumulated resistance and distributed inductance along their length. When multiple mid-to-long-range loads are centrally disconnected, the current in the branches and their upstream cables drops rapidly, and the distributed inductance generates an induced electromotive force, leading to a transient voltage rise or overshoot on the bus. Existing centralized load shedding methods primarily determine the shedding targets based on load priority and power shortages, with less consideration given to load location, cable branch relationships, and differences in inductive reactance accumulated along the power supply path. Uniform voltage compensation methods also struggle to accommodate the voltage withstand requirements of loads in different spatial locations. Therefore, it is necessary to incorporate load voltage withstand range, power supply distance, and cable topology into the scheduling constraints to reduce transient voltage surges caused by centralized load shedding.

[0006] To overcome the shortcomings of existing technologies that fail to consider the spatial topology of the power supply network and the cumulative inductive reactance of cables during concentrated load shedding, which can easily lead to transient voltage overshoot and affect the safe operation of sensitive loads, this invention provides a multi-source power coordinated dispatching method and system based on spatial topology constraints. By constructing a safe voltage withstand corridor and a multi-branch tree of the load physical topology, power dispatching decisions are transformed into spatial geometric judgments and sequential execution logic based on the cumulative inductive reactance distance value. This scheme achieves dynamic coordination between multi-source power compensation and spatial sequential load shedding. It utilizes spatial sequential shedding to reduce the intensity of the effect of current change rate on distributed inductance, and combines closed-loop calibration to maintain the voltage gradient across the entire line within a safe range, ensuring the stability of voltage distribution at each node of the multi-source system and the operational safety of sensitive equipment under complex power grid sag conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] Multi-source power cooperative scheduling methods based on spatial topology constraints include:

[0009] Collect load operation data, physical communication node distances of each load node, and cable branch connection methods between each load. Construct a safe voltage withstand corridor and draw a spatial voltage drop line based on the physical communication node distances of each load node and the load operation data. Perform multi-source power collaborative calculation by projecting the spatial voltage drop line into the safe voltage withstand corridor to determine the load list to be cut, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value.

[0010] Based on the physical communication node distance and cable branch connection method, a multi-branch tree of load physical topology is constructed and the cumulative inductive reactance distance value of each leaf node is calculated. Multi-source power compensation operation is performed according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value. The load list to be cut off is spatially sequentially cut off in order of cumulative inductive reactance distance value from large to small. Closed-loop calibration is performed based on the actual bus voltage value after the cut-off.

[0011] The multi-source power collaborative calculation includes:

[0012] The system collects grid sag characteristic data and multi-source power supply data through a power router. Based on the grid sag characteristic data, it determines the sag scenario type. It then calculates the total power deficit and identifies core risk points by combining the multi-source power supply data and load operation data. The load operation data includes at least the voltage tolerance level. The grid sag characteristic data includes at least the voltage amplitude change and the duration of the sag.

[0013] The method for determining the type of temporary landing scenario includes:

[0014] The sag depth coefficient is calculated based on the voltage amplitude change, and the sag depth threshold is set as Kth and the sag duration threshold is set as Tth.

[0015] When the descent depth coefficient is less than or equal to Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a shallow descent with a short duration; when the descent depth coefficient is greater than Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a deep descent with a short duration; when the descent depth coefficient is less than or equal to Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a shallow descent with a long duration; when the descent depth coefficient is greater than Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a deep descent with a long duration.

[0016] The method for constructing the safe voltage withstand corridor includes:

[0017] A two-dimensional geometric coordinate system is established with the physical communication node distance of each load node as the horizontal axis and the node voltage as the vertical axis. The zero point of the horizontal axis is set at the output end of the power router. A safe voltage withstand corridor is constructed in the two-dimensional geometric coordinate system.

[0018] The safe voltage withstand corridor includes an upper boundary line and a lower boundary line. The upper boundary line is a horizontal straight line, and the lower boundary line is a stepped broken line.

[0019] The vertical axis coordinate value of the upper boundary line is equal to the upper limit of the voltage that the power router is allowed to output;

[0020] The lower boundary line is set at the horizontal axis coordinate position of each load node, which is the lower limit of the voltage withstand level of the corresponding load terminal.

[0021] The lower boundary line between two adjacent load nodes is taken as the higher of the two load nodes' lower voltage withstand level values.

[0022] The method for drawing the spatial pressure drop line includes:

[0023] Calculate the voltage drop slope per unit distance, set the output voltage setting value of the power router, use the output voltage setting value of the power router as the y-intercept of the spatial voltage drop line, and use the negative value of the voltage drop slope per unit distance as the slope of the spatial voltage drop line. Plot the spatial voltage drop line in the two-dimensional geometric coordinate system where the safe voltage withstand corridor is located.

[0024] The method for determining the load list to be trimmed includes:

[0025] Determine whether the space voltage drop line falls completely inside the safe voltage withstand corridor. If the space voltage drop line does not fall completely inside the safe voltage withstand corridor, calculate the voltage withstand level lower limit of each load node in the section where the space voltage drop line passes through the lower boundary line of the safe voltage withstand corridor, and subtract the estimated node voltage value at the corresponding horizontal axis coordinate position of the space voltage drop line to obtain the voltage deficiency difference. Add the load node with the largest voltage deficiency difference to the list of loads to be cut.

[0026] The energy storage discharge rate is determined based on a differentiated configuration according to the type of transient scenario and the total power gap.

[0027] The method for constructing the load physical topology multi-way tree includes:

[0028] The power router output is set as the root node and each load terminal is set as a leaf node. A load physical topology multi-branch tree is constructed based on the physical communication node distance and cable branch connection method of each load node.

[0029] The method for executing the spatial sequential load shedding includes:

[0030] In the list of loads to be cut, loads are cut off one by one in descending order of cumulative inductive reactance distance value until the cumulative cut-off power reaches the cut-off power limit. When the current load node to be cut off shares the same parent node with the previous cut-off load node in the load physical topology multi-branch tree, the waiting time interval is set to the basic waiting time interval multiplied by the shortening factor. The shortening factor is equal to the inductive reactance distance value of the branch cable where the current load node to be cut off is located divided by the cumulative inductive reactance distance value of the current load node to be cut off. When the current load node to be cut off does not share the same parent node with the previous cut-off load node in the load physical topology multi-branch tree, the waiting time interval is set to the basic waiting time interval.

[0031] The method for performing the closed-loop calibration includes:

[0032] After the multi-source power compensation operation and spatial sequential load shedding are completed, the actual bus voltage value is monitored. The actual bus voltage value is compared with the power router output voltage setting value to calculate the voltage deviation. The energy storage discharge rate and the power router output voltage setting value are corrected according to the voltage deviation. During the grid sag recovery phase, the disconnected loads are restored one by one from the near end to the far end in the spatial order opposite to the spatial sequential load shedding order. After each load terminal is restored, the unit distance voltage drop slope is recalculated and it is re-determined whether the spatial voltage drop line still falls within the safe voltage withstand corridor.

[0033] A multi-source power cooperative scheduling system based on spatial topology constraints is used to implement the above-mentioned multi-source power cooperative scheduling method based on spatial topology constraints. The system includes:

[0034] Collaborative Decision Module: This module collects load operation data, the physical communication node distances between each load node, and the cable branch connection methods between each load. Based on the physical communication node distances between each load node and the load operation data, it constructs a safe voltage withstand corridor and draws a spatial voltage drop line. By projecting the spatial voltage drop line onto the safe voltage withstand corridor, it determines the multi-source power collaborative calculation, identifies the load list to be cut, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value.

[0035] The calibration module performs the following: it constructs a multi-branch tree of the physical topology of the load based on the distance between physical communication nodes and the cable branch connection method, calculates the cumulative inductive reactance distance value of each leaf node, performs multi-source power compensation operation according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value, performs spatial sequential load removal on the load list to be cut off in descending order of cumulative inductive reactance distance value, and performs closed-loop calibration based on the actual bus voltage value after removal.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] The multi-source power collaborative scheduling method based on spatial topology constraints provided by this invention integrates the dispersed node-by-node voltage constraints in a multi-source power supply system into a continuous geometric constraint space by collecting the distance between physical communication nodes and the cable connection method. It constructs a safe voltage withstand corridor and projects a spatial voltage drop line, enabling the system to simultaneously verify the voltage drop difference caused by the spatial impedance of the cable at both ends of the entire power supply line. This avoids the risk of near-end load overvoltage damage or far-end load undervoltage disconnection due to excessive voltage drop caused by using a single-point unified constant voltage compensation strategy. At the same time, it constructs a multi-branch tree of the load physical topology based on the cable branch connection method and extracts the cumulative inductive reactance distance value of each leaf node as the sorting basis for load shedding. This transforms the traditional disordered batch load shedding into a spatial sequential shedding operation executed from far end to near end. By smoothing the transient change rate of cable current, it suppresses the transient voltage overshoot phenomenon induced on the DC bus by the instantaneous release of back electromotive force due to the distributed inductance of long-distance cables, ensuring the operational safety and overall voltage stability of sensitive loads that are not shedding on the same power supply line during the voltage dip. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart of the multi-source power coordinated scheduling method based on spatial topology constraints in this invention;

[0040] Figure 2 This is a schematic diagram illustrating the classification of temporary landing scenarios in this invention;

[0041] Figure 3 This is a schematic diagram of the construction of the safety voltage withstand corridor in this invention;

[0042] Figure 4 This is a schematic diagram illustrating the load clipping triggered by the linear spatial pressure drop extending through the corridor in this invention.

[0043] Figure 5 This is a flowchart illustrating the topology construction and multi-source power execution process in this invention.

[0044] Figure 6 This is a schematic diagram of the spatial sequential load removal sequence in this invention;

[0045] Figure 7 This is a schematic diagram illustrating the reverse execution of load recovery access from the near end to the far end in this invention;

[0046] Figure 8This is a functional block diagram of the multi-source power coordinated scheduling system based on spatial topology constraints in this invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Example 1:

[0049] Please see Figure 1 As shown, this embodiment provides a multi-source power cooperative scheduling method based on spatial topology constraints, including:

[0050] Step S10: Collect load operation data, physical communication node distance of each load node, and cable branch connection method between each load. Construct a safe voltage withstand corridor and draw a spatial voltage drop line based on the physical communication node distance of each load node and load operation data. Perform multi-source power collaborative calculation by projecting the spatial voltage drop line in the safe voltage withstand corridor to determine the load list to be cut, energy storage discharge rate, photovoltaic output lock value, and rectifier control parameter adjustment value.

[0051] Traditional power dispatching methods perform voltage checks on each load node individually based on the power balance equation. This requires independently calculating the cable voltage drop for each load node and comparing it with the voltage withstand threshold. As the number of load nodes increases, the computational complexity multiplies, making it difficult to complete all checks and decisions within the transient response window of a grid sag. Step S10 integrates the dispersed node-by-node voltage constraints into a continuous geometric region by constructing a safe voltage withstand corridor. By drawing a spatial voltage drop line, the voltage distribution of all nodes is compressed into a single linear equation. By determining whether the line falls within the corridor, the multi-node voltage check is transformed into a single geometric position determination. Thus, the synchronous voltage safety check and pruning decision for all load nodes are completed without increasing computational complexity. After completing the projection determination of the spatial voltage drop line, the system further performs multi-source power collaborative calculation. That is, by combining the sag scenario classification, the calculated total power gap, and the identified core risk points, the "multi-source" components such as energy storage battery packs, photovoltaic power generation units, and grid rectifiers are treated as a whole and solved jointly. This collaborative computing provides decision inputs for differentiated configuration of multi-source power allocation strategies, enabling the energy storage discharge rate, photovoltaic output lock-in value, and rectifier control parameter adjustment value to be configured in a targeted manner according to the differences in sag depth and sag duration. This avoids secondary risks such as premature depletion of energy storage or amplification of rectifier ripple that may be caused by a unified strategy in different sag scenarios.

[0052] Further, step S10 includes:

[0053] Step S11: Collect grid sag characteristic data, multi-source power supply data and load operation data through the power router. Simultaneously collect the physical communication node distance from each load node to the output end of the power router along the power supply cable, as well as the cable branch connection method between each load, the physical length of each branch cable, and the inductance value and resistance value per unit length of each branch cable.

[0054] Specifically, the power router, as the core power distribution hub on the DC bus side of a multi-source power supply system, is responsible for managing the power flow and maintaining voltage between the AC side of the power grid, photovoltaic power generation units, energy storage battery banks, and various load terminals. Power grid sag characteristic data includes the voltage amplitude change of the AC input on the power grid side, the time of sag occurrence, the duration of sag, and the sag recovery slope. The power router acquires this power grid sag characteristic data through a periodic sampling method using its built-in AC side voltage transformer and high-speed sampling module. The sampling period is determined based on the control response bandwidth of the power router, and the value of the sampling period must meet the Nyquist sampling theorem's requirement for a multiple of the fundamental frequency of the power grid, ensuring that the sampling frequency of the power grid sag characteristic data can completely reproduce the transient waveform characteristics during the power grid voltage sag process. Multi-source power supply data includes the current output value and adjustable output range of the photovoltaic power generation units, the state of charge and sustainable discharge power of the energy storage battery banks, and the current operating mode and output power of the rectifier. The power router acquires this multi-source power supply data through the communication bus between the photovoltaic inverter, the battery management system, and the rectifier control board. Load operation data includes the rated power, real-time operating power, voltage withstand level, and load priority identifier for each load terminal. The voltage withstand level represents the lower voltage limit at which each load terminal can maintain normal operation during a bus voltage drop. For example, the lower voltage withstand level of precision instruments or controllers may be higher, at 210V. If a grid voltage dip causes the voltage distributed to that node from the bus to fall below 210V, the terminal will lose power, malfunction, or enter a shutdown protection state. Conversely, the lower voltage withstand level of ordinary water pumps or lighting fixtures may be lower, at 180V. Even if the node voltage drops to 190V, such terminals can still maintain basic operation using the lower voltage. The load priority identifier is used to distinguish between critical and non-critical loads. The power router obtains the above load operation data through reports from the local controllers of each load terminal.

[0055] The process of collecting the physical communication node distances from each load node along the power supply cable to the output of the power router is as follows: The power router sends a distance query command to the local controller of each load terminal. Each local controller sends back the pre-calibrated physical communication node distance values ​​to the power router. During the system installation and commissioning phase, the physical communication node distance values ​​are measured by on-site engineers along the cable and written into the parameter register of the local controller. The physical communication node distance values ​​represent the actual conductor path length from each load terminal along the power supply cable to the output of the power router, rather than the straight-line distance in space. This allows the voltage drop model subsequently built based on the physical communication node distances to accurately reflect the physical characteristics of cable impedance accumulating with path length. The data collection of cable branch connection methods between loads is also completed during the installation and commissioning phase. Engineers write the topology relationship of each branch node into the topology configuration table of the power router in the form of parent-child node number pairs. The topology configuration table records which lower-level load nodes are connected to each branch node and the physical length of each branch cable segment. The inductance and resistance per unit length of each branch cable are determined by the technical parameters provided by the cable manufacturer. They are obtained from the manufacturer's datasheet based on the cable model and cross-sectional area. The resistance per unit length is calculated based on the resistivity of the cable conductor material and the conductor cross-sectional area. The power router synchronously stores the inductance and resistance per unit length of each branch cable in the topology configuration table.

[0056] Step S11 collects data covering four dimensions: grid-side disturbance information, multi-source power reserve information, load demand information, and power supply network spatial topology information (including cable resistance and inductance electrical parameters). All input parameters required for power dispatch are collected and centrally stored in the power router within the same time window. This allows subsequent steps S12 to S14 to be calculated based on the complete dataset from the same moment, avoiding power gap estimation errors or inaccurate voltage withstand judgments caused by inconsistent data collection times across different dimensions. The synchronous collection of physical communication node distances and cable branch connection methods provides the necessary topology and parameter inputs for constructing the load physical topology multi-branch tree and calculating the cumulative inductive reactance distance in step S20. Without the collection of this spatial topology data in step S11, step S20 would be unable to establish the spatial positional relationship between load nodes and the power router output, and the spatial sequential load shedding would lose its sorting basis.

[0057] Step S12: Determine the type of transient scenario based on grid transient characteristic data, calculate the total power gap and identify core risk points by combining multi-source power supply data and load operation data;

[0058] See Figure 2 This is a schematic diagram illustrating the classification of temporary landing scenarios provided in an embodiment of the present invention. Figure 2A two-dimensional classification coordinate system is established with sag duration as the horizontal axis and sag depth coefficient as the vertical axis. The coordinate system is divided into four independent regions by two vertical dashed lines: the sag depth threshold Kth and the sag duration threshold Tth, each corresponding to one of four different sag scenario types. The sag scenario type is determined based on two dimensions: voltage amplitude change and sag duration in the power grid sag characteristic data. The voltage amplitude change is defined as the difference between the rated power grid voltage and the actual power grid voltage during the sag. The ratio of the voltage amplitude change to the rated power grid voltage is defined as the sag depth coefficient, which is equal to the difference between the rated power grid voltage and the actual power grid voltage during the sag, divided by the rated power grid voltage. A sag depth threshold of Kth and a sag duration threshold of Tth are set. Kth is determined as follows: The sag depth coefficient distribution for each sag event is statistically analyzed from the historical operating data of the multi-source power supply system. All historical sag depth coefficients are arranged in ascending order. The ratio of the sum of the residual grid power supply and the current output of the photovoltaic power generation unit to the total load demand is calculated at each sag depth coefficient level. When this ratio decreases to the point where the power share that the energy storage battery pack needs to independently compensate exceeds the time length corresponding to the sag duration threshold Tth, where the energy storage battery pack can continuously discharge at its current state of charge to maintain the normal operation of all critical loads, the corresponding sag depth coefficient is the value of Kth. "Independently maintaining the operation of all critical loads" means that the discharge power of the energy storage battery pack can compensate for the difference between the sum of the residual grid power supply and the photovoltaic output and the total demand of all critical loads, and the remaining available capacity of the energy storage battery pack is sufficient to continuously discharge at this compensation power until the time length corresponding to the sag duration threshold Tth without triggering the lower limit of the battery management system's state of charge protection. In a typical DC bus power supply system, the value of Kth usually ranges from 0.1 to 0.4, with the specific value varying depending on the capacity configuration of the energy storage battery pack and the power proportion of critical loads. The method for determining Tth is as follows: First, determine the total power demand of all critical loads. Then, select the most stringent sag depth considered during the system design phase as the benchmark sag depth** (for example, take the maximum value corresponding to the preset protection requirements as a typical benchmark during the system design phase). Calculate the power share that the sum of the residual grid power supply and the representative output value of the photovoltaic power generation unit evaluated during the system design phase can bear under this benchmark sag depth condition. Subtract this power share from the total power demand of all critical loads to obtain the power difference that needs to be independently compensated by the energy storage battery pack. Then, divide the rated available capacity of the energy storage battery pack by this power difference to obtain the duration for which the energy storage battery pack can continuously compensate. This duration is used as the design value of Tth. The current available capacity refers to the net available capacity after subtracting the remaining capacity corresponding to the current state of charge protection lower limit set by the battery management system from the remaining capacity corresponding to the current state of charge of the energy storage battery pack.In a typical DC bus power supply system, the value of Tth usually ranges from hundreds of milliseconds to several seconds, with the specific value depending on the ratio of the installed capacity of the energy storage battery pack to the power of the critical load. Figure 2 The classification coordinate system determines the landing scenario type as follows: when the landing depth coefficient is less than or equal to Kth and the landing duration is less than or equal to Tth, the landing scenario type is determined to be shallow landing with short duration; when the landing depth coefficient is greater than Kth and the landing duration is less than or equal to Tth, the landing scenario type is determined to be deep landing with short duration; when the landing depth coefficient is less than or equal to Kth and the landing duration is greater than Tth, the landing scenario type is shallow landing with long duration; and when the landing depth coefficient is greater than Kth and the landing duration is greater than Tth, the landing scenario type is deep landing with long duration.

[0059] The classification of temporary landing scenarios is shown in Table 1:

[0060] Table 1 Classification of Temporary Landing Scenarios

[0061] Shallow temporary drop, short duration Less than or equal to Kth Less than or equal to Tth Deep descent short-term Greater than Kth Less than or equal to Tth Shallow descent long duration Less than or equal to Kth Greater than Tth Deep descent long-term Greater than Kth Greater than Tth

[0062] The calculation process for the total power deficit is as follows: The total load demand power is obtained by summing the real-time operating power of each load terminal collected in step S11. The actual power output from the grid side via the rectifier during the sag is recorded as the residual grid power, determined based on the sag depth coefficient and the rectifier's power-voltage conversion characteristic curve. The current output value of the photovoltaic power generation unit is recorded as Ppv, and the sustainable discharge power of the energy storage battery pack is recorded as Pbat. The total power deficit equals the total load demand power minus the residual grid power, minus the current output value of the photovoltaic power generation unit Ppv, and minus the sustainable discharge power of the energy storage battery pack Pbat. When the total power deficit is greater than zero, it indicates that the available power of the multi-source power supply system is insufficient to maintain the operation of all loads, requiring load pruning. When the total power deficit is less than or equal to zero, it indicates that the available power of the multi-source power supply system can cover all load demands, requiring no load pruning, but the output distribution of each power source still needs to be adjusted according to the sag scenario type.

[0063] Figure 2The two-dimensional classification method simplifies the complex characteristics of grid voltage sag events into intuitive regional divisions, providing a unified judgment framework for the accurate identification of core risk points in different scenarios. The identification of core risk points is differentiated based on the type of voltage sag scenario: for shallow, short-term sags, the core risk point is the amplification of bus voltage ripple caused by rectifier input voltage fluctuations; for deep, short-term sags, the core risk point is the rapid drop in bus voltage below the critical load voltage tolerance level due to an instantaneous power shortfall; for shallow, long-term sags, the core risk point is the continuous decline in the state of charge of the energy storage battery pack during long-term recharging, leading to insufficient recharging capacity in the later stages; and for deep, long-term sags, the core risk point is the combined effect of power shortfall and insufficient energy storage capacity, resulting in the bus voltage remaining below the safe operating range. The identification of core risk points provides a targeted adjustment direction for determining the energy storage recharging / discharging rate, photovoltaic output lock-in value, and rectifier control parameter adjustment values ​​in step S14. This allows the multi-source power allocation strategy to be configured with the goal of eliminating core risks, rather than adopting a uniform and fixed power allocation scheme.

[0064] Step S12 categorizes grid sag events into four scenarios based on two dimensions: sag depth coefficient and sag duration. This allows subsequent power allocation and load shaving strategies to be configured differently according to the core risk characteristics of different scenarios. Without the scenario classification and risk identification in step S12, step S14, in determining the energy storage discharge rate and photovoltaic output lock-in value, would be unable to distinguish the differentiated needs of multi-source power allocation for different sag scenarios. This could lead to the use of the same energy storage discharge strategy in deep sag, long-duration scenarios as in shallow sag, short-duration scenarios, resulting in the energy storage being exhausted in the middle of the sag. The total power gap calculation in step S12 is closely linked to the multi-source power supply data and load operation data collected in step S11. The data collection in step S11 provides the numerical basis for the current available power of each power source and the power demand of the load in step S12. The total power gap calculated in step S12, in turn, provides a quantitative constraint for determining the power reduction amount of the load list to be shaving in step S14.

[0065] Step S13: Establish a two-dimensional geometric coordinate system with the physical communication node distance of each load node as the horizontal axis and the node voltage as the vertical axis. Construct a safe voltage withstand corridor in the two-dimensional geometric coordinate system based on the voltage withstand level of each load node in the load operation data.

[0066] See Figure 3 This is a schematic diagram of the construction of a safe voltage withstand corridor provided in an embodiment of the present invention. Figure 3A two-dimensional geometric coordinate system is established with the physical communication node distance as the horizontal axis and the node voltage as the vertical axis. The positions of the output terminal, load 1, load 2, load 3, and load 4 are marked from left to right on the horizontal axis, illustrating the safe voltage withstand corridor area enclosed by the upper and lower boundary lines. The two-dimensional geometric coordinate system is constructed with the physical communication node distance from each load node along the power supply cable to the power router output terminal as the horizontal axis and the node voltage as the vertical axis. Combined with... Figure 3 The coordinate system is marked with the zero point of the horizontal axis set at the output end of the power router. The horizontal axis extends along the cable path from the output end of the power router to each load terminal. The coordinate value of each load node on the horizontal axis is equal to the physical communication node distance value collected in step S11. Figure 3 The horizontal axis shows the distribution of loads 1, 2, 3, and 4. The vertical axis values ​​are determined based on the rated voltage level of the DC bus system and the voltage withstand level of each load terminal. The upper limit of the vertical axis is set to the maximum allowable output voltage of the power router, and the lower limit is set to the lower voltage limit corresponding to the load terminal with the lowest voltage withstand level among all load terminals. For example... Figure 3 As shown, the safe voltage withstand corridor is represented in a two-dimensional geometric coordinate system as a strip-shaped region enclosed by an upper boundary line and a lower boundary line. The upper boundary line of the safe voltage withstand corridor is a horizontal straight line, and its vertical axis coordinate value is equal to the upper limit of the voltage allowed to be output by the power router. The lower boundary line of the safe voltage withstand corridor is a stepped broken line, and its horizontal axis coordinate value at each load node is the lower limit of the voltage withstand level of the corresponding load terminal. Figure 3 The lower boundary line exhibits a stepped change shape at load 1, load 2, load 3, and load 4.

[0067] The specific construction process of the lower boundary line of the safe voltage withstand corridor is as follows: arrange all load nodes according to the physical communication node distance value from smallest to largest, corresponding to... Figure 3The arrangement of the output terminal, load 1, load 2, load 3, and load 4 is such that, in the resulting load node sequence, the lower boundary line between any two adjacent load nodes is taken as the higher of the two load nodes' lower voltage withstand limit values. This ensures that the value of the lower boundary line at any location meets the voltage withstand requirements of all load terminals within the corresponding segment. For example, suppose node A at a distance of 100 meters on the cable is a common fan with a lower voltage withstand limit of 180V; node B at a distance of 200 meters is a precision medical device with a lower voltage withstand limit of 210V. When constructing the lower boundary line for this spatial segment between 100 meters and 200 meters, the system must "choose the higher value," uniformly setting the safety baseline for this segment to 210V. This is because there is a voltage drop in the power supply cable. If the reference line is incorrectly set to a lower 180V, when the cable voltage drops to 195V at 150 meters, the system will mistakenly judge it as "safe" (195V>180V) and will not trigger power compensation or load shearing. As a result, the voltage continues to decay along the cable, and when it reaches node B at 200 meters, it will inevitably be lower than 210V, directly causing the precision medical equipment to shut down due to undervoltage. Therefore, taking the higher value among adjacent nodes can force the system to detect the crisis in advance. The physical meaning of the safe voltage withstand corridor is: when the node voltage at any point on the power supply cable is between the upper and lower boundaries of the safe voltage withstand corridor, all load terminals from the power router output to that location can maintain normal operation; when the node voltage is lower than the lower boundary, at least one load terminal's operating voltage is lower than the lower limit of the voltage withstand level and faces the risk of shutdown.

[0068] Step S13 transforms the dispersed load voltage withstand constraints in a multi-source power supply system into a geometrically defined visual determination area. This allows the search process for the power router's output voltage setpoint in step S14 to be performed in a two-dimensional coordinate system using geometric relationships instead of calculating the voltage of each individual load node. The lower boundary line of the safe voltage withstand corridor uses a stepped polygonal line instead of a single horizontal line, reflecting the differentiated voltage withstand requirements of different load terminals at different locations on the power supply cable. This allows the subsequent determination of the position of the spatial voltage drop line within the safe voltage withstand corridor to simultaneously consider the spatial distribution characteristics of the cable voltage drop and the individual voltage withstand differences of each load terminal. Without the safe voltage withstand corridor constructed in step S13, step S14 would lack a unified spatial constraint benchmark when determining the power router's output voltage setpoint. It would require checking the voltage of each load node individually to ensure it meets the withstand requirements, resulting in a lengthy calculation process that is difficult to complete within the millisecond-level response window of the power grid sag. Step S13 integrates the physical communication node distances and voltage withstand levels of each load terminal collected in step S11 into a geometric tool called the safe voltage withstand corridor, providing a constraint boundary for step S14 that can be directly used for spatial geometric determination.

[0069] Step S14: Calculate the voltage drop slope per unit distance. Using the negative value of the voltage drop slope per unit distance as the slope, draw a spatial voltage drop line in the two-dimensional geometric coordinate system where the safe voltage withstand corridor is located. Determine whether the spatial voltage drop line falls completely inside the safe voltage withstand corridor. Based on the determination result, determine the load list to be cut and the power cut amount. Combine the determined sag scenario type, total power gap and identified core risk points to determine the energy storage discharge rate, photovoltaic output lock value and rectifier control parameter adjustment value.

[0070] See Figure 4 This is a schematic diagram of load clipping triggered by a straight line of spatial pressure drop through a corridor, provided in an embodiment of the present invention. Figure 4A two-dimensional geometric coordinate system is established with the distance between physical communication nodes as the horizontal axis and the node voltage as the vertical axis. The positions of the output terminal, load 1, load 2, load 3, and load 4 are marked from left to right on the horizontal axis, showing the relative positional relationship of the safe voltage withstand corridor area and the spatial voltage drop line before and after trimming. The slope of the voltage drop per unit distance is calculated based on the ratio of the total line voltage drop generated by the power supply cable under the current total load demand power to the total line length. The total equivalent impedance of the power supply cable is defined as Ztotal. Ztotal represents the total equivalent series impedance along the path from the output of the power router to the farthest load node, considering the current shunting effect at each branch node. Ztotal is the equivalent value of the sum of the resistive and inductive components of each branch cable segment. The resistive component of each branch cable segment is equal to the resistance value per unit length of the branch cable multiplied by the physical length. The inductive component of each branch cable segment is equal to the inductance value per unit length of the branch cable multiplied by the system operating angular frequency and then multiplied by the physical length. The system operating angular frequency is equal to twice pi multiplied by the system operating frequency. The system operating frequency is determined based on the switching frequency of the rectifier or the main frequency component of the bus side ripple. The specific method for determining the system operating angular frequency is further explained in step S21. The impedance value of each branch cable segment is equal to the sum of its resistive and inductive components. The specific calculation method is as follows: The impedance values ​​of each cable segment are accumulated segment by segment along the farthest path from the root node to the farthest leaf node. The impedance values ​​of shared cable segments before branch nodes are directly accumulated. The impedance values ​​of cable segments after branch nodes are weighted and calculated based on the ratio of the current flowing through that segment to the total current before being accumulated. The weighted calculation means multiplying the actual impedance value of a cable segment after a branch node by the ratio of the current flowing through that segment to the total current flowing out from the root node. This weighted impedance value reflects the equivalent contribution of that segment to the total equivalent impedance. The product Ztotal multiplied by the total current Itotal is the total equivalent voltage drop along the path from the power router output to the farthest load node. The current flowing through each cable segment is obtained by dividing the sum of the downstream load power carried by that segment by the bus rated voltage. Ztotal is calculated based on the physical length, inductance per unit length, and resistance per unit length of the cable conductor material of each branch cable collected in step S11, combined with the branch structure of the load physical topology multi-branch tree, using the weighted conversion method described above. The calculation method adopts the series and parallel impedance equivalent method in circuit theory. The total current flowing through the power supply cable is defined as Itotal, which equals the total power demand of the load divided by the rated voltage of the bus. The total length of the power supply cable from the output of the power router to the farthest load node is defined as Ltotal, which equals the sum of the physical lengths of all branch cables on the path from the root node to the farthest leaf node in the load physical topology multi-branch tree.The voltage drop slope per unit distance, Kslope, is equal to the total equivalent impedance Ztotal multiplied by the total current Itotal, then divided by the total line length Ltotal. The physical meaning of the voltage drop slope per unit distance is the voltage drop caused by the load current per unit physical distance along the power supply cable. The above calculation equates the voltage drop distribution of each branch cable in the multi-branch tree topology to a linear voltage drop distribution along the farthest path. This linear equivalence treatment has good approximation accuracy under conditions where the load power distribution of each branch is relatively uniform. When the load power distribution varies significantly, the deviation is corrected through the closed-loop calibration mechanism in step S24.

[0071] The process of plotting the spatial voltage drop line in the two-dimensional geometric coordinate system of the safe voltage withstand corridor is as follows: Using the power router's output voltage setpoint Vset as the ordinate of the spatial voltage drop line, and the negative value of the slope per unit distance as the slope of the spatial voltage drop line, a straight line extending from the ordinate point along the positive horizontal axis is plotted in the two-dimensional geometric coordinate system. The ordinate value of any point on the spatial voltage drop line represents the estimated node voltage value at the corresponding physical communication node distance on the power supply cable under the condition that the power router's output voltage is Vset. The equation of the spatial voltage drop line is: V(d) equals Vset minus Kslope multiplied by d, where V(d) is the estimated node voltage value at a physical communication node distance of d. Figure 4 As shown, the spatial voltage drop line is a straight line sloping from the upper left to the lower right. The negative value of the slope reflects the physical law that the node voltage gradually decreases along the power supply cable path. The load node farther away from the power router output end bears a greater cable voltage drop and the estimated node voltage value is lower.

[0072] The process of determining the output voltage setpoint Vset of the power router is as follows: Vset is assigned an initial value based on the upper limit of the voltage allowed to be output by the power router. During the load pruning decision-making phase, Vset remains unchanged at the upper limit of the voltage. A spatial voltage drop line with Vset as the vertical intercept is plotted in a two-dimensional geometric coordinate system. It is determined whether the spatial voltage drop line completely falls within the safe voltage withstand corridor. The determination condition is that the vertical coordinate value of each load node on the spatial voltage drop line is greater than or equal to the value of the lower boundary line of the safe voltage withstand corridor at the corresponding position, and the vertical intercept value of the spatial voltage drop line is less than or equal to the value of the upper boundary line of the safe voltage withstand corridor. When the spatial voltage drop line completely falls within the safe voltage withstand corridor, it indicates that under the condition that the output voltage setpoint of the power router is Vset, the estimated node voltage of all load terminals meets their respective voltage withstand level requirements. At this time, there is no need to perform load pruning, and the load list needs to be pruned to empty.

[0073] When the spatial voltage drop line fails to fall completely within the safe voltage withstand corridor, it indicates that the estimated node voltage of some load nodes is lower than the lower limit of the voltage withstand level. Load pruning is required to reduce the total load demand power and thus reduce the voltage drop slope per unit distance. Figure 4 The process is presented visually: before the trimming, the spatial voltage drop line extends beyond the lower boundary of the safety voltage withstand corridor in the section from load 3 to load 4, indicating that the load node voltage in this area cannot meet the withstand requirements; after the load trimming, the slope of the voltage drop per unit distance decreases, and the trimmed spatial voltage drop line shifts upward as a whole and falls completely into the interior of the safety voltage withstand corridor. The execution logic for load pruning is as follows: In the section where the spatial voltage drop line crosses the lower boundary of the safe voltage withstand corridor, the voltage withstand level lower limit of each load node is calculated, and the estimated node voltage value at the corresponding horizontal axis coordinate position of the spatial voltage drop line is subtracted from the value of the lower limit. The load node with the largest voltage deficiency difference in the corresponding section is marked as the first load to be pruned. The first load to be pruned is added to the list of loads to be pruned, and the real-time operating power of the first load to be pruned is deducted from the total load demand power. The voltage drop slope per unit distance is recalculated based on the deducted total load demand power, and the spatial voltage drop line is redrawn. It is then determined whether the updated spatial voltage drop line completely falls within the safe voltage withstand corridor. If it still does not fall within, the same logic is continued to add the next load node with the largest difference to the list of loads to be pruned. This process is repeated until the spatial voltage drop line completely falls within the safe voltage withstand corridor. At the end of the loop, the sum of the real-time operating power of all load terminals in the list of loads to be pruned is the pruning power allowance.

[0074] The energy storage replenishment rate is determined based on the type of sag scenario and the total power shortfall, with differentiated configurations: When the sag scenario is a shallow sag and short-term, the energy storage replenishment rate is set to one shallow short-term conventional discharge coefficient multiplied by Pbat, the sustainable discharge power Pbat of the energy storage battery pack. The shallow short-term conventional discharge coefficient is greater than 0 and less than or equal to 1. The coefficient is determined based on the ratio of the difference between the residual grid power supply and the total load demand in the shallow sag and short-term scenario, ensuring that the sum of the energy storage replenishment rate and the residual grid power supply covers the total load demand. When the sag scenario is a deep sag and short-term, the energy storage replenishment rate is set to one deep short-term coefficient multiplied by Pbat, the sustainable discharge power Pbat of the energy storage battery pack. The short-term transient discharge coefficient is multiplied by Pbat. The value of the deep short-term transient discharge coefficient is greater than the shallow short-term conventional discharge coefficient and less than or equal to 1. The method for determining the deep short-term transient discharge coefficient is based on the characteristic of a significant reduction in the residual power supply of the grid and a short duration of the transient. The ratio of the total power gap to the sustainable discharge power Pbat of the energy storage battery pack is used as the value of the deep short-term transient discharge coefficient, so that the energy storage battery pack can output power at a higher rate within a shorter duration of transient to make up for the instantaneous power gap. When the transient scenario is a shallow long-term transient, the energy storage compensation discharge rate is set to a shallow long-term average discharge coefficient multiplied by Pbat, which is the sustainable discharge power Pbat of the energy storage battery pack. The average discharge coefficient for the shallow-long type is smaller than that for the conventional shallow-short type. The method for determining the average discharge coefficient for the shallow-long type is to divide the current available capacity of the energy storage battery pack by the duration of the sag to obtain the average allowable discharge power, and then divide this by the sustainable discharge power Pbat of the energy storage battery pack to obtain the average discharge coefficient for the shallow-long type. This ensures that the energy storage battery pack maintains continuous power supply capability during a longer sag duration without premature depletion. The current available capacity refers to the net available capacity after subtracting the capacity corresponding to the lower limit of the state of charge protection set by the battery management system from the remaining capacity corresponding to the current state of charge of the energy storage battery pack. By deducting the capacity corresponding to the lower limit of the protection, it is ensured that the energy storage battery pack will not be forcibly cut off by the battery management system due to the state of charge dropping to the lower limit of the protection during the entire sag duration. When the scenario type is deep sag long duration, the energy storage replenishment discharge rate is set to a deep long conservative discharge coefficient multiplied by Pbat, which is the sustainable discharge power Pbat of the energy storage battery pack. The deep long conservative discharge coefficient is less than or equal to the shallow long average discharge coefficient, so that the energy storage battery pack will not drop to the protection lower limit of charge due to over-discharge during the duration of the sag. The deep long conservative discharge coefficient is determined by dividing the current available capacity of the energy storage battery pack by the estimated value of the duration of the sag to obtain the average allowable discharge power, and then dividing it by the sustainable discharge power Pbat of the energy storage battery pack to obtain the deep long conservative discharge coefficient. The definition of the current available capacity is the same as that of the shallow long average discharge coefficient, that is, the net available capacity after deducting the amount of charge corresponding to the lower limit of the state of charge protection.Because the sag depth is greater in deep sag scenarios and the residual power supply from the grid is lower, the energy storage battery pack needs to undertake a larger proportion of power compensation tasks. To prevent the energy storage battery pack from losing its power supply capacity due to over-discharge in the later stages of the sag, the conservative discharge coefficient for deep sag scenarios needs to be further lower than the average discharge coefficient for shallow sag scenarios in order to extend the continuous power supply time of the energy storage battery pack. When the deep-long conservative discharge coefficient calculated according to the above formula is greater than the shallow-long average discharge coefficient, the deep-long conservative discharge coefficient is forcibly set to the shallow-long average discharge coefficient to ensure that the constraint relationship of deep-long conservative discharge coefficient being less than or equal to shallow-long average discharge coefficient is met. The four discharge coefficients satisfy the relationship of deep-short transient discharge coefficient > shallow-short conventional discharge coefficient > shallow-long average discharge coefficient ≥ deep-long conservative discharge coefficient. The estimated value of the sag duration is calculated based on the sag recovery slope and the current sag depth in the grid sag characteristic data collected in step S11. The specific calculation method is to divide the voltage drop corresponding to the current sag depth by the sag recovery slope to obtain the estimated time length required for the grid voltage to recover to the rated value from the current moment. This estimated time length is the estimated value of the sag duration. The sag recovery slope characterizes the amount of voltage rise per unit time during the sag recovery process. Its value is extracted from the sampled data of the grid voltage recovery phase after the sag occurs. When sag recovery slope data is unavailable, twice the sag duration threshold Tth is used as the estimated sag duration. The basis for selecting twice Tth as the estimated value is that the sag duration in deep sag long-duration scenarios must be greater than Tth. Using twice Tth as a conservative estimate can reserve sufficient capacity margin for energy storage battery packs when accurate recovery slope data is lacking, and avoid premature depletion of energy storage battery packs due to underestimation of sag duration.

[0075] The photovoltaic (PV) output lock-in value is configured based on the type of sag scenario: during a sag, the output of the PV power generation unit is locked at the current output value Ppv at the time of the sag, preventing additional disturbances to the DC bus power supply caused by irradiance fluctuations. The rectifier control parameter adjustment values ​​are configured based on the core risk points in the sag scenario: when the core risk point is an increase in bus voltage ripple, the rectifier control parameter adjustment values ​​include an increment of the ripple suppression gain, which is determined based on the ratio of the change in bus voltage ripple amplitude before and after the sag. When the core risk point is a rapid drop in bus voltage, the rectifier control parameter adjustment values ​​include an increment of the voltage loop response bandwidth, improving the rectifier's ability to track and compensate for bus voltage drops by increasing the voltage loop response speed. When the core risk point is the continuous decline of the energy storage state of charge or the superposition of power shortage and insufficient energy storage capacity, the adjustment value of the rectifier control parameters includes the adjustment of the power limit value. The output power limit value of the rectifier is set to the maximum rectified output power corresponding to the residual power supply of the grid during the sag period, so as to prevent the rectifier from entering the overload protection state due to insufficient input voltage.

[0076] Step S14 combines the safe voltage withstand corridor constructed in step S13 with the unit distance voltage drop slope calculated in step S14 to determine the geometric positional relationship between the spatial voltage drop line and the safe voltage withstand corridor. The geometric criterion of whether the line completely falls within the strip area simultaneously completes two tasks: voltage safety verification and load pruning identification. The projection determination of the spatial voltage drop line in the safe voltage withstand corridor transforms the traditional node-by-node voltage calculation into a geometric relationship comparison. After pruning a load node in each iteration, only the slope needs to be updated and the relative position of the line and the corridor needs to be re-determined. The number of calculation steps is linearly related to the number of load nodes to be pruned, enabling the power router to complete load pruning decisions within the millisecond-level response window of the grid sag. The pruning logic in step S14 prioritizes pruning load nodes with the largest difference between the estimated node voltage and the lower limit of the voltage withstand level. This ensures that each pruning operation achieves the maximum shrinkage of the spatial voltage drop line with the minimum pruning power, reducing the total number of loads to be pruned and the maximum pruning power requirement. The energy storage discharge rate, photovoltaic output lock value, and rectifier control parameter adjustment value determined in step S14 are directly transmitted to step S22 as control commands for multi-source power compensation operation. The load list to be cut and the cut power quota determined in step S14 are directly transmitted to step S23 as the execution basis for spatial sequential load shedding. The power router output voltage setting value determined in step S14 is transmitted to step S24 as the comparison benchmark for closed-loop calibration, forming a complete parameter transmission link from decision output to execution input.

[0077] Step S10 transforms the power dispatch decision-making process of a multi-source power supply system during grid sag events from the traditional iterative numerical solution based on the power balance equation to a geometric determination based on the positional relationship between a straight line and a strip region in two-dimensional geometric space. This transformation compresses the voltage safety verification involving all load nodes into the calculation of the intersection of a linear equation and a stepped polygonal line, reducing the computational complexity from a product relationship with the number of load nodes to a linear relationship with the number of load nodes to be cut. This allows the power router to complete the calculation and output of all dispatch decisions within a millisecond-level time window after a grid sag occurs. Simultaneously, the stepped lower boundary line of the safe voltage withstand corridor integrates the differentiated voltage withstand levels of each load terminal into the spatial dimension. This ensures that the priority of load cutting depends not only on the load priority identifier but also on the magnitude of the voltage drop and voltage withstand margin experienced by each load node in the spatial distribution of the power supply cable, protecting more critical loads with smaller voltage margins under the same power cutting capacity. The introduction of physical communication node distance and safe voltage withstand corridor in step S10 encodes the spatial location information of load nodes into the power dispatch decision-making process by using a spatial voltage drop line. This ensures that the list of loads to be cut in step S10 naturally carries the spatial sequence information of each load node to be cut. The spatially sequential load shedding in step S20 can be performed directly according to the physical communication node distance of each load node in the list of loads to be cut, avoiding cable current surges and bus voltage oscillations caused by disordered spatial execution of shedding operations. Step S10 completes the classification of grid sag scenarios and spatial voltage safety verification in the same decision-making process, enabling the differentiated configuration of power allocation strategies and the spatial determination of load shedding decisions to be executed in parallel rather than serially, shortening the decision delay time from the occurrence of grid sag to the output of dispatch instructions.

[0078] Step S20: Construct a multi-branch tree of the physical topology of the load based on the distance between physical communication nodes and the cable branch connection method, and calculate the cumulative inductive reactance distance value of each leaf node. Perform multi-source power compensation operation according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value. Perform spatial sequential load removal on the load list to be cut off in descending order of cumulative inductive reactance distance value. Perform closed-loop calibration based on the actual voltage value of the bus after removal.

[0079] The output of step S10 is the numerical list of loads to be cut, the cut power limit, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value. These numerical results need to be converted into state switching instructions for each power source and cut-off-recovery timing instructions for each load terminal before they can be applied to the multi-source power supply system. Step S20 transforms the load list to be pruned into a cut-off execution sequence with spatial and temporal constraints by constructing a multi-branch tree of the load physical topology and calculating the cumulative inductive reactance distance value. Through multi-source power compensation operation, the energy storage discharge rate, photovoltaic output lock value, and rectifier control parameter adjustment value are transformed into the actual operating state of the three power sources. Through closed-loop calibration, the deviation between the actual bus voltage and the power router output voltage setting value is fed back to the correction of the energy storage discharge rate. Through the reverse-order recovery strategy, the spatial sorting information of the cut-off stage is reused to determine the access order in the recovery stage. This makes the multi-source power supply system form a complete technical link from spatial geometric determination to physical temporal execution throughout the entire process of grid sag occurrence, decision calculation, power compensation, load cut-off, closed-loop calibration to load recovery. Throughout the entire sag-recovery cycle, the bus voltage is maintained within the safe range defined by the safe voltage withstand corridor.

[0080] Further, step S20 includes:

[0081] Step S21, see Figure 5 The power router output is set as the root node and each load terminal is set as a leaf node. A load physical topology multi-branch tree is constructed based on the physical communication node distance and cable branch connection method of each load node. The inductive reactance distance value of each edge in the load physical topology multi-branch tree is calculated based on the physical length of each branch cable and the inductance value per unit length. The inductive reactance distance values ​​of all edges on the path from each leaf node to the root node are accumulated to obtain the cumulative inductive reactance distance value of each leaf node.

[0082] Specifically, the load physical topology multi-branch tree is a tree-like data structure with the output of the power router as the root node, each load terminal as the leaf node, and each branch cable connection point as the intermediate node. Each intermediate node in the load physical topology multi-branch tree can connect to two or more child nodes, thus representing the actual laying pattern of power cables branching from a main line to multiple branch lines. The construction process of the load physical topology multi-branch tree is as follows: the power router reads the parent-child node number pairs of each branch node from the topology configuration table collected in step S11, sets the number of the power router output as the root node number, and sequentially attaches each branch node and each load terminal to the corresponding parent node according to the hierarchical relationship in the parent-child node number pairs. The connection relationship between each pair of parent and child nodes constitutes an edge in the load physical topology multi-branch tree, and each edge corresponds to a branch cable segment. When all parent-child node number pairs in the topology configuration table have been traversed, the construction of the load physical topology multi-branch tree is completed. The number of leaf nodes in the tree is equal to the number of load terminals, and the number of edges in the tree is equal to the number of branch cable segments.

[0083] The inductive reactance distance value means the total inductive impedance accumulated along the physical length of a branch cable under power frequency or switching frequency excitation. The inductive reactance distance value couples the physical length of the cable with the electrical characteristics of the cable into a unified metric. The calculation process for the inductive reactance distance value of each edge in the multi-branch tree of the load physical topology is as follows: read the physical length and unit length inductance value of the branch cable corresponding to each edge from the topology configuration table, multiply the unit length inductance value by the system operating angular frequency to obtain the unit length inductive reactance value, and then multiply the unit length inductive reactance value by the physical length of the branch cable to obtain the inductive reactance distance value of each edge. The system operating frequency is determined based on the switching frequency of the rectifier or the main frequency component of the bus side ripple. When the switching frequency of the rectifier is a known fixed value, the switching frequency of the rectifier is used as the system operating frequency. When the rectifier adopts a frequency conversion modulation method, resulting in an unstable switching frequency, the main frequency component of the bus side ripple is used as the system operating frequency. The main frequency component of the bus side ripple is obtained by sampling and recording the bus voltage waveform when the rectifier is running under rated conditions during the system installation and commissioning phase. The frequency component with the largest amplitude is extracted by performing spectrum analysis on the sampled data as the main frequency component and the frequency value is written into the parameter register of the power router. When the modulation method of the rectifier changes, the above spectrum analysis process needs to be re-executed to update the values ​​of the main frequency components. The inductive reactance distance value, rather than the purely resistive distance value, is used as the metric because the cable voltage drop in the DC bus system is mainly dominated by the inductive component during transient processes. Although the resistive voltage drop accounts for a certain proportion under steady-state conditions, the rate of change of cable current is high during rapid load shedding caused by grid sags. The value of the inductive voltage drop component L multiplied by the rate of change of current is much larger than that of the resistive voltage drop component. Therefore, using the inductive reactance distance value as the basis for load shedding sorting can more accurately reflect the transient voltage impact amplitude experienced by each node at the moment of load shedding. The calculation process for the cumulative inductive reactance distance value of each leaf node is as follows: In the load physical topology multi-branch tree, starting from each leaf node, backtrack along the edges of the tree towards the root node, and accumulate the inductive reactance distance values ​​of all edges traversed along the backtracking path. The sum obtained after accumulation is the cumulative inductive reactance distance value of each leaf node. The cumulative inductive reactance distance value represents the total inductive impedance accumulated between each load terminal and the output terminal of the power router along the power supply cable. The larger the cumulative inductive reactance distance value, the stronger the transient inductive voltage drop impact on the bus voltage at the moment the load is disconnected.

[0084] Step S21 converts the physical communication node distance, cable branch connection method, physical length of each branch cable, and inductance per unit length collected in step S11 into a load physical topology multi-branch tree with inductive distance labeling. This provides a cut-off execution sequence based on the inductive distance value for the spatial sequential load cut-off in step S23. Without the construction of the load physical topology multi-branch tree and the calculation of the cumulative inductive distance value in step S21, step S23 will be unable to determine the cut-off order of each load terminal when performing load cut-off. This may result in unordered load cut-off in space, leading to alternating cut-off of far-end and near-end loads. Alternating cut-off will cause repeated abrupt changes in the current direction on the power supply cable. These repeated abrupt changes in the current direction, combined with the inductive component of the cable, will generate voltage oscillations with superimposed amplitudes on the bus side. Once the voltage oscillation causes the bus voltage to drop instantaneously below the lower limit of the voltage withstand level of the critical load, it will cause the critical load that should have been kept running to shut down unnecessarily. Step S10 establishes a spatial geometric judgment framework for voltage safety with the distance between physical communication nodes as the horizontal axis. Step S21 further introduces the electrical characteristic dimension of inductive distance on this basis, which refines the sorting basis for load disconnection from pure geometric distance to the equivalent impedance distance that includes the electrical parameters of the cable, thereby improving the correspondence between the disconnection sorting and the actual transient voltage impact.

[0085] Step S22: Dynamically adjust the rectifier control parameters according to the rectifier control parameter adjustment value to suppress the bus voltage ripple to below the preset ripple safety threshold, control the energy storage replenishment according to the energy storage replenishment and discharge rate, and control the photovoltaic output according to the photovoltaic output lock value to complete the multi-source power replenishment operation.

[0086] Specifically, see [link to relevant document] Figure 5In step S22, the multi-source power compensation operation regulates the rectifier, energy storage battery pack and photovoltaic power generation unit respectively according to the three sets of control commands determined in step S14: rectifier control parameter adjustment value, energy storage compensation discharge rate and photovoltaic output lock value. The dynamic adjustment process of rectifier control parameters is as follows: The power router sends the rectifier control parameter adjustment value to the rectifier control board through the communication bus. The rectifier control board modifies the proportional gain and integral gain of the rectifier's internal voltage loop controller according to the ripple suppression gain adjustment increment in the rectifier control parameter adjustment value. The modified proportional gain is equal to the original proportional gain plus the proportional component in the ripple suppression gain adjustment increment. The modified integral gain is equal to the original integral gain plus the integral component in the ripple suppression gain adjustment increment. The method for determining the proportional and integral components in the ripple suppression gain adjustment increment is as follows: Calculate the ripple amplification factor based on the ratio of the bus voltage ripple amplitude before the sag to the bus voltage ripple amplitude after the sag. Subtract 1 from the ripple amplification factor to obtain the ripple overshoot factor. Multiply the ripple overshoot factor by the original proportional gain and the original integral gain to obtain the proportional component and integral component, respectively. The proportional and integral components are determined according to the above formula based on the following: For a linear PI voltage loop controller, the controller's ability to suppress disturbances is approximately linearly related to the gain. Therefore, when the bus voltage ripple amplitude is amplified by a factor of k, the proportional gain and integral gain need to be increased by a factor of (k-1) to restore the ripple suppression effect to the level before the sag. Increasing the proportional gain and integral gain synchronously by the same proportion helps to maintain the phase margin of the PI controller and prevents control loop oscillations caused by asynchronous changes in proportional and integral gains. The rectifier control board operates the voltage loop controller with modified control parameters to suppress the bus voltage ripple to below the preset ripple safety threshold after the rectifier control parameters are adjusted. The ripple safety threshold is determined by taking the difference between the lower limit of the voltage withstand level of each load terminal and the rated voltage of the bus, and multiplying the minimum value of all differences by the ripple margin coefficient gamma to obtain the ripple safety threshold. The value of gamma is determined based on the steady-state accuracy of the bus voltage control system. Specifically, the bus voltage control system has an inherent voltage fluctuation range during steady-state operation. This inherent fluctuation range is determined by the adjustment accuracy and sampling quantization error of the voltage loop controller. The value of gamma needs to ensure that the ripple safety threshold is greater than the inherent fluctuation range to avoid the voltage loop controller frequently triggering adjustment actions due to the ripple safety threshold being set too low during normal operation. The typical range of gamma is 0.3 to 0.7. When the steady-state accuracy of the bus voltage control system is high, i.e., the inherent fluctuation range is small, a larger value of gamma is taken to relax the ripple safety threshold. When the steady-state accuracy is low, i.e., the inherent fluctuation range is large, a smaller value of gamma is taken to tighten the ripple safety threshold. This ensures that the setting of the ripple safety threshold retains sufficient margin while not excessively constraining the control bandwidth of the rectifier.

[0087] The control process for energy storage replenishment is as follows: The power router sends the energy storage replenishment / discharge rate determined in step S14 to the battery management system via the communication bus. The battery management system sets the target discharge power value of the energy storage battery pack according to the energy storage replenishment / discharge rate. The DC-DC converter of the energy storage battery pack operates with the target discharge power value as the power loop setpoint, stabilizing the discharge power of the energy storage battery pack at the level specified by the energy storage replenishment / discharge rate. The control process for photovoltaic output is as follows: The power router sends the photovoltaic output lock value determined in step S14 to the photovoltaic inverter via the communication bus. The photovoltaic inverter switches from maximum power point tracking control mode to constant power control mode. In constant power control mode, the photovoltaic inverter operates with the photovoltaic output lock value as the power output setpoint, no longer adjusting the output power according to changes in sunlight, thereby eliminating the power disturbance on the bus side caused by sunlight fluctuations during the photovoltaic power generation unit's voltage dip. The three operations of rectifier control parameter adjustment, energy storage power replenishment start-up, and photovoltaic output locking are executed synchronously in time. The power router sends control commands to the rectifier control board, battery management system and photovoltaic inverter at the same time within the same control cycle. The three power sources complete the state switching within the same time window, so that the bus voltage is always under the joint support of multiple power sources during the transition phase after the sag occurs.

[0088] The multi-source power compensation operation in step S22 transforms the decision output of step S14 into the actual operating state switching of each power source. Without this operation, the energy storage discharge rate, photovoltaic output locking value, and rectifier control parameter adjustment values ​​determined in step S14 would remain numerical and ineffective on physical devices. The bus voltage would continuously decrease due to the power gap before load shedding in step S23, causing load shedding in step S23 to only begin when the bus voltage is already below the lower limit of the withstand voltage level for some critical loads, thus weakening the timeliness of the shedding operation. The photovoltaic output locking operation in step S22 eliminates the power fluctuation source of the photovoltaic power generation unit during the sag, ensuring that the actual bus voltage value monitored in step S24 during the closed-loop calibration phase is only affected by three deterministic variables: residual grid power supply, energy storage discharge power, and load shedding power. This reduces the number of uncertainties in the voltage deviation calculation in step S24, making the correction direction and amount of the closed-loop calibration clearer.

[0089] Step S23, see below. Figure 5 In the list of loads to be cut, spatial sequential load cuts are performed one by one in descending order of cumulative inductive distance value until the cumulative cut power reaches the cut power limit. The waiting time interval is dynamically adjusted according to whether the branch cable of the current load node to be cut and the previous cut load node share the same parent node in the load physical topology multi-way tree.

[0090] See Figure 6This is a schematic diagram of the spatial sequential load removal sequence provided in an embodiment of the present invention. Figure 6 The document illustrates the connection topology between the power router and loads A, B, C, and D, marking the positional relationship between the near and far ends. It presents the execution order of the first, second, and third load shedding operations, and highlights the characteristics of load A having a smaller cumulative inductive reactance distance value and load D having a larger cumulative inductive reactance distance value, clarifying that the shedding direction is from the far end to the near end. Specifically, the spatially sequential load shedding in step S23 follows the order of cumulative inductive reactance distance values ​​from largest to smallest. That is, starting from the leaf node with the largest cumulative inductive reactance distance value, load terminals in the load list to be trimmed are shedding one by one. After each load terminal is shedding, the real-time operating power of the shedding load terminals is accumulated. Shedding stops when the accumulated shedding power reaches or exceeds the trimming power limit determined in step S14. The technical basis for starting shedding from the load terminal with the largest cumulative inductive reactance distance value is that the load terminal with the largest cumulative inductive reactance distance value is located at the farthest end of the power supply cable, corresponding to... Figure 6 Load D in the power supply cable carries current flowing through all branch cable segments along its entire length. Disconnecting the furthest load terminal reduces the current across the entire cable by an amount equal to the operating current of the furthest load terminal. Consequently, the inductive voltage drop across each branch cable segment decreases synchronously due to the reduced current. This causes the bus voltage recovery effect resulting from disconnecting a furthest load terminal to propagate along the entire power supply cable towards the root node, causing the voltages of all intermediate and near-end nodes to recover synchronously. If disconnection begins from the near-end load terminal with the smallest cumulative inductive reactance distance, the corresponding... Figure 6 In the case of load A, the operating current of the near-end load terminal only flows through a few branch cables near the root node. After the near-end load terminal is disconnected, the voltage recovery effect is limited to the near-end section, and the voltage improvement at the far-end node is weak. Under the same power reduction capacity, the recovery of the bus voltage is lower than that of the scheme that disconnects from the far end.

[0091] The dynamic adjustment mechanism for the waiting time interval is as follows: The base waiting time interval is defined as Twait_base. Twait_base is determined by multiplying the inductive time constant of the single branch cable with the largest inductive reactance distance by the time constant multiplier eta. The inductive time constant is equal to the inductive reactance of the single branch cable divided by its resistance. The inductive reactance of the single branch cable is equal to the inductance per unit length of that branch cable multiplied by the system operating angular frequency and then by the physical length of that branch cable. The resistance of the single branch cable is equal to the resistance per unit length of that branch cable multiplied by its physical length. The value of the time constant multiplier eta is determined based on the number of time constants required for the cable current to decay to a steady-state value after the load is disconnected. Typically, it is the number of time constants required for the deviation in current decay to the steady-state value to be less than or equal to the allowable deviation ratio epsilon. The value of epsilon is determined based on the bus voltage control accuracy requirements, with a typical range of 0.02 to 0.05. When the current load node to be removed and the previous removed load node share the same parent node in the load physical topology multi-way tree, the two load nodes are mounted on different branch cables below the same parent node. The current change caused by removing the current load node to be removed only affects the local section below the parent node, and the disturbance range of the bus voltage is small. At this time, the waiting time interval is set to Twait_base multiplied by the shortening factor mu. The value of mu is greater than 0 and less than 1. The method of determining mu is to divide the inductive reactance distance of the branch cable where the current load node to be removed is located (from the shared parent node to the current load node to be removed) by the cumulative inductive reactance distance of the current load node to be removed. The physical basis of this method is that the propagation range of the current transient change caused by the removal operation on the power supply cable and the amplitude of the voltage disturbance are both proportional to the inductive reactance length of the affected cable. When two load nodes share the same parent node, the current change caused by the disconnection operation only produces an inductive voltage drop change on the local branch cable segment from the shared parent node to the current load node to be disconnected. The smaller the proportion of the inductive reactance distance value of this local branch to the total cumulative inductive reactance distance value of the current load node to be disconnected, the narrower the range of disturbance and the smaller the disturbance amplitude of the disconnection operation on the bus side voltage. Therefore, the time required to wait for the current transient decay can be shortened by the same proportion.A smaller mu indicates a smaller proportion of the local branch in the total path, a weaker impact of the removal operation on the global bus voltage, and a greater reduction in the waiting time. When the current load node to be removed and the previous removed load node do not share the same parent node's branch cable in the load physical topology multi-way tree, the two load nodes are distributed on different branch paths. The current change caused by removing the current load node to be removed will affect all branch cable segments on the entire path from the current load node to the root node, and the disturbance range of the bus voltage will cover the root node. In this case, the waiting time interval is set to Twait_base and is not shortened.

[0092] The dynamic adjustment mechanism of the waiting time interval in step S23 shortens the disconnection interval between adjacent load terminals sharing the same parent node branch cable, while maintaining the intact disconnection interval between load terminals distributed on different branch paths. The physical basis of this differentiated waiting strategy is as follows: When two load terminals sharing a branch cable are disconnected, the current changes caused by the two disconnection operations are superimposed on the local cable segment below the shared branch cable. However, on the cable segment above the shared branch cable, the current changes from the two disconnections are in the same direction and their amplitudes are predictable, preventing the superposition of currents in opposite directions. Conversely, when two load terminals distributed on different branch paths are disconnected, the current changes caused by the two disconnections propagate along different paths from their respective leaf nodes to the root node. If the time interval between the two disconnections is too short, the superposition of the two current changes at the root node may cause overshoot of the bus voltage. Step S23 combines the cumulative inductive reactance distance value calculated in step S21 with the list of loads to be cut and the cut-off power limit determined in step S14. The cumulative inductive reactance distance value provides the basis for the disconnection order, the list of loads to be cut limits the disconnection range, and the cut-off power limit provides the disconnection termination condition; all three are indispensable. Figure 6 This intuitively reflects the sorting logic. The first cut-off, the second cut-off, and the third cut-off are executed in descending order of the cumulative inductive reactance distance value. If the spatial sequential load cut-off in step S23 is missing, the list of loads to be cut off determined in step S14 cannot be converted into actual cut-off actions with spatial and temporal constraints. The load cut-off will degenerate into disordered batch disconnection. The batch disconnection will release a large amount of cable energy at the same time, which will cause the bus voltage to have an uncontrollable transient overshoot.

[0093] Step S24: After the multi-source power compensation operation and spatial sequential load removal are completed, monitor the actual bus voltage value, compare the actual bus voltage value with the power router output voltage setting value to calculate the voltage deviation, and correct the energy storage discharge rate and the power router output voltage setting value according to the voltage deviation and the sag scenario type. During the grid sag recovery phase, recalculate the unit distance voltage drop slope of the spatial voltage drop line and determine whether the updated spatial voltage drop line still falls within the safe voltage withstand corridor. Reconnect the removed loads one by one from the near end to the far end in the spatial order opposite to the spatial sequential load removal order.

[0094] Specifically, in step S24, the monitoring of the actual bus voltage value is obtained through continuous sampling using the DC bus voltage sensor built into the power router. The sampling period is consistent with the sampling period of the grid sag characteristic data in step S11, ensuring that the voltage comparison accuracy of the closed-loop calibration is at the same level as the accuracy of the voltage safety verification in step S10. The voltage deviation is calculated as follows: the actual bus voltage value is subtracted from the power router output voltage setting value Vset determined in step S14 to obtain the voltage deviation deltaV. When the voltage deviation deltaV is greater than zero, it indicates that the actual bus voltage value is higher than the power router output voltage setting value Vset; when the voltage deviation deltaV is less than zero, it indicates that the actual bus voltage value is lower than the power router output voltage setting value Vset. The correction process for the energy storage discharge rate is as follows: When the voltage deviation deltaV is less than zero, regardless of whether the sag scenario type is shallow sag short-time, deep sag short-time, shallow sag long-time, or deep sag long-time, the energy storage discharge rate is increased by a correction increment delta_Pbat based on the original value determined in step S14. delta_Pbat is equal to the absolute value of the voltage deviation deltaV divided by the total equivalent impedance Ztotal of the power supply cable and then multiplied by the actual bus voltage value. The physical meaning of delta_Pbat is to convert the voltage deviation into equivalent power. The compensation amount after the gap; when the voltage deviation deltaV is greater than zero, the energy storage discharge rate is reduced by a correction reduction based on the original value determined in step S14. The calculation method of the correction reduction is the same as the calculation method of the correction increment, that is, the correction reduction is equal to the absolute value of the voltage deviation deltaV divided by the total equivalent impedance Ztotal of the power supply cable and then multiplied by the actual voltage value of the bus; when the energy storage discharge rate after reducing the correction reduction is less than zero, the energy storage discharge rate is set to zero to prevent the energy storage battery pack from entering the charging mode and drawing power from the bus side in reverse, which would cause the bus voltage to drop. The correction process for the power router's output voltage setpoint Vset is as follows: When the absolute value of the voltage deviation deltaV is greater than the voltage correction trigger threshold delta_Vth, to avoid excessive deviation between the power router's output voltage setpoint and the actual bus voltage value leading to control loop output saturation, half of the voltage deviation deltaV is added to the power router's output voltage setpoint Vset as the corrected power router output voltage setpoint. This causes the corrected Vset to adjust towards the actual bus voltage value by half the deviation. The selection of half the deviation as the basis for a single correction is as follows: If each correction directly adjusts Vset to the actual bus voltage value (i.e., full correction), when the bus voltage fluctuates continuously due to power changes over multiple consecutive control cycles, Vset will frequently and significantly jump with the fluctuations, causing the voltage control loop to overshoot due to rapid changes in the setpoint. If each correction is too small, for example, only one-tenth of the deviation is adjusted, the correction speed of Vset will be too slow, and the voltage deviation will not be effectively converged for a long time.A balance is struck between the correction speed and control stability by taking half of the deviation amount, allowing Vset to gradually converge to near the actual bus voltage value within several control cycles without oscillation. Simultaneously, the power output is increased by the correction increment delta_Pbat of the energy storage discharge rate in step S24 to drive the actual bus voltage value back towards the corrected Vset. The voltage correction trigger threshold delta_Vth is determined by multiplying the ripple safety threshold by the correction trigger proportional coefficient rho. The value of rho is determined based on the steady-state error bandwidth of the voltage control loop. The steady-state error bandwidth refers to the normal fluctuation range of the output voltage around the set value during steady-state operation of the voltage control loop, which is determined by the proportional gain and integral gain of the voltage loop controller. The value of rho needs to ensure that the voltage correction trigger threshold delta_Vth is greater than the steady-state error bandwidth to avoid normal steady-state fluctuations being misjudged as voltage deviations requiring correction. The typical value range of rho is 1.5 to 3.0.

[0095] During the grid voltage dip recovery phase, the AC input voltage on the grid side gradually recovers, and the residual power supply of the grid increases accordingly. The power router recalculates the difference between the total load demand and the available power of each power source to update the total power gap. It then recalculates the total current Itotal flowing through the power supply cables based on the updated total power gap, and recalculates the voltage drop slope per unit distance based on the updated total current Itotal. Finally, it plots the updated spatial voltage drop line with the corrected power router output voltage setpoint Vset as the y-axis intercept, and determines whether the updated spatial voltage drop line still falls within the safe voltage withstand corridor. When the updated spatial voltage drop line completely falls within the safe voltage withstand corridor and there is a voltage margin between the updated spatial voltage drop line and the lower boundary line of the safe voltage withstand corridor, it indicates that the system is ready to restore the disconnected load. (See also...) Figure 7 This is a schematic diagram of load recovery access being executed in reverse order from the near end to the far end, provided in an embodiment of the present invention. Figure 7 The diagram illustrates the connection topology between the power router and loads A, B, C, and D, marking the positional relationship between the near and far ends. It presents the execution order of the first, second, and third restorations, clearly indicating that the restoration direction is from near to far. The restoration of disconnected loads follows the reverse spatial order of the spatially sequential load disconnection in step S23, i.e., restoring access one by one starting from the disconnected load terminal with the smallest cumulative inductive reactance distance value, restoring sequentially from near to far. Figure 7The process begins by restoring the connection to load A, which has the smallest cumulative inductive reactance distance. This is followed by restoring the connection to load B, and then load C. After each load is restored, the voltage drop slope per unit distance is recalculated, and it is re-evaluated whether the spatial voltage drop line still falls within the safe voltage withstand corridor. If the spatial voltage drop line crosses the lower boundary of the safe voltage withstand corridor after a load is restored, the restoration is paused, and the grid voltage is allowed to recover further.

[0096] The technical basis for restoring access sequence from near end to far end is as follows: the cumulative inductive reactance distance value of load A at the near end is the smallest, while that of load D at the far end is the largest. The cumulative inductive reactance distance value of the near-end load terminal is relatively small, resulting in a shorter inductive cable segment through which the increased current flows during restoration. This leads to a smaller transient impact amplitude on the bus voltage and a faster attenuation rate. Restoring the near-end load first allows the bus voltage to gradually increase under relatively small disturbances, establishing a power gradient buffer for the subsequent restoration of the far-end load. The closed-loop calibration in step S24 feeds back the actual execution results of the multi-source power compensation operation in step S22 and the spatial sequential load removal in step S23 to the geometric judgment framework of the spatial voltage drop line and the safe voltage withstand corridor established in step S10 for verification, forming a complete closed-loop link from decision-making to execution to verification to correction. If the closed-loop calibration in step S24 is missing, the power router output voltage setting value Vset and the energy storage discharge rate determined in step S14 may deviate from the actual bus voltage due to factors such as cable impedance parameter errors and load power fluctuations after actual execution. The deviation will continue until the sag ends and cannot be corrected. The disconnected loads will also lack the judgment conditions for orderly recovery during the grid sag recovery phase and may be fully connected at once. The full connection at once will cause the bus voltage to drop below the lower boundary of the safe voltage withstand corridor again due to the load power step jump.

[0097] Step S20 transforms the scheduling decision parameters output from step S10 into a physical execution sequence arranged in an ordered manner on the time axis and following the inductive reactance distance gradient on the spatial axis. This ensures that the power scheduling of the multi-source power supply system during grid sags maintains spatial consistency and temporal coordination as it extends from the decision-making stage to the execution stage. Spatially sequential load shedding is performed in descending order of cumulative inductive reactance distance values. This ensures that the current changes induced by each shedding operation on the power supply cable propagate gradually from the far end to the near end in the same direction, rather than overlapping spatially. The direction of current change in the cable remains consistent throughout the shedding process, avoiding repeated abrupt changes in current direction that could induce oscillating voltage fluctuations in the inductive component of the cable. The bus voltage exhibits a monotonically increasing trend rather than oscillating fluctuations during the shedding process. The dynamic adjustment of the waiting time interval is based on the branch structure of the multi-branch tree of the load physical topology. The waiting time is shortened for load terminals on shared branch cables, while maintaining a complete waiting time for load terminals on different branch paths. This compresses the total time of the shedding operation to the necessary length while ensuring the stability of the bus voltage, shortening the scheduling execution window from the occurrence of a sag to the completion of all shedding operations. In step S20, the restoration access phase adopts the opposite spatial order to the cut-off phase, restoring one by one from the near end to the far end. This reverse restoration strategy and the far end priority strategy of the cut-off phase form a mirror-symmetric spatial sequence. The cable energy released in the cut-off phase gradually decreases from the far end to the near end in space, and the cable energy added in the restoration phase gradually accumulates from the near end to the far end in space. The energy storage change directions of the two phases are completely opposite in space and consecutive in time, so that the energy storage state change of the power supply cable in the entire sag-recovery cycle always advances in a single spatial direction, without the jump situation of energy storage in a local section being released first and then suddenly accumulating in large quantities. Thus, the monotonic change trend of the bus voltage is maintained in the entire sag-recovery cycle, eliminating the risk of secondary voltage drop caused by improper load access sequence in the restoration phase.

[0098] Example 2:

[0099] This embodiment, based on Embodiment 1, provides a multi-source power collaborative scheduling system based on spatial topology constraints, such as... Figure 8 As shown, it includes:

[0100] Collaborative Decision Module: This module collects load operation data, the physical communication node distances between each load node, and the cable branch connection methods between each load. Based on the physical communication node distances between each load node and the load operation data, it constructs a safe voltage withstand corridor and draws a spatial voltage drop line. By projecting the spatial voltage drop line onto the safe voltage withstand corridor, it determines the multi-source power collaborative calculation, identifies the load list to be cut, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value.

[0101] The calibration module performs the following: it constructs a multi-branch tree of the physical topology of the load based on the distance between physical communication nodes and the cable branch connection method, calculates the cumulative inductive reactance distance value of each leaf node, performs multi-source power compensation operation according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value, performs spatial sequential load removal on the load list to be cut off in descending order of cumulative inductive reactance distance value, and performs closed-loop calibration based on the actual bus voltage value after removal.

[0102] Furthermore, in the collaborative decision-making module, the multi-source power collaborative calculation includes:

[0103] The system collects grid sag characteristic data and multi-source power supply data through a power router. Based on the grid sag characteristic data, it determines the sag scenario type. It then calculates the total power deficit and identifies core risk points by combining the multi-source power supply data and load operation data. The load operation data includes at least the voltage tolerance level. The grid sag characteristic data includes at least the voltage amplitude change and the duration of the sag.

[0104] The method for determining the type of temporary landing scenario includes:

[0105] The sag depth coefficient is calculated based on the voltage amplitude change, and the sag depth threshold is set as Kth and the sag duration threshold is set as Tth.

[0106] When the descent depth coefficient is less than or equal to Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a shallow descent with a short duration; when the descent depth coefficient is greater than Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a deep descent with a short duration; when the descent depth coefficient is less than or equal to Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a shallow descent with a long duration; when the descent depth coefficient is greater than Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a deep descent with a long duration.

[0107] The method for constructing the safe voltage withstand corridor includes:

[0108] A two-dimensional geometric coordinate system is established with the physical communication node distance of each load node as the horizontal axis and the node voltage as the vertical axis. A safe voltage withstand corridor is constructed in the two-dimensional geometric coordinate system.

[0109] The safe voltage withstand corridor includes an upper boundary line and a lower boundary line. The upper boundary line is a horizontal straight line, and the lower boundary line is a stepped broken line.

[0110] The vertical axis coordinate value of the upper boundary line is equal to the upper limit of the voltage that the power router is allowed to output;

[0111] The lower boundary line is set at the horizontal axis coordinate position of each load node, which is the lower limit of the voltage withstand level of the corresponding load terminal.

[0112] The lower boundary line between two adjacent load nodes is taken as the higher of the two load nodes' lower voltage withstand level values.

[0113] The method for drawing the spatial pressure drop line includes:

[0114] Calculate the voltage drop slope per unit distance, set the output voltage setting value of the power router, use the output voltage setting value of the power router as the y-intercept of the spatial voltage drop line, and use the negative value of the voltage drop slope per unit distance as the slope of the spatial voltage drop line. Plot the spatial voltage drop line in the two-dimensional geometric coordinate system where the safe voltage withstand corridor is located.

[0115] The method for determining the load list to be trimmed includes:

[0116] Determine whether the space voltage drop line falls completely inside the safe voltage withstand corridor. If the space voltage drop line does not fall completely inside the safe voltage withstand corridor, calculate the voltage withstand level lower limit of each load node in the section where the space voltage drop line passes through the lower boundary line of the safe voltage withstand corridor, and subtract the estimated node voltage value at the corresponding horizontal axis coordinate position of the space voltage drop line to obtain the voltage deficiency difference. Add the load node with the largest voltage deficiency difference to the list of loads to be cut.

[0117] The energy storage discharge rate is determined based on a differentiated configuration according to the type of transient scenario and the total power gap.

[0118] Furthermore, in the calibration module, the method for constructing the load physical topology multi-way tree includes:

[0119] The power router output is set as the root node and each load terminal is set as a leaf node. A load physical topology multi-branch tree is constructed based on the physical communication node distance and cable branch connection method of each load node.

[0120] The methods and systems of the present invention may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of the present invention are not limited to the order specifically described above, unless otherwise specifically stated.

[0121] In addition, the parts of the technical solutions provided in the embodiments of the present invention that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0122] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-source power coordinated dispatch method based on spatial topology constraints, characterized in that, The method includes: Collect load operation data, physical communication node distances of each load node, and cable branch connection methods between each load. Construct a safe voltage withstand corridor and draw a spatial voltage drop line based on the physical communication node distances of each load node and the load operation data. Perform multi-source power collaborative calculation by projecting the spatial voltage drop line into the safe voltage withstand corridor to determine the load list to be cut, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value. Based on the physical communication node distance and cable branch connection method, a multi-branch tree of load physical topology is constructed and the cumulative inductive reactance distance value of each leaf node is calculated. Multi-source power compensation operation is performed according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value. The load list to be cut off is spatially sequentially cut off in order of cumulative inductive reactance distance value from large to small. Closed-loop calibration is performed based on the actual bus voltage value after the cut-off.

2. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 1, characterized in that, The multi-source power collaborative calculation includes: The system collects grid sag characteristic data and multi-source power supply data through a power router. Based on the grid sag characteristic data, it determines the sag scenario type. It then calculates the total power deficit and identifies core risk points by combining the multi-source power supply data and load operation data. The load operation data includes at least the voltage tolerance level. The grid sag characteristic data includes at least the voltage amplitude change and the duration of the sag.

3. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 2, characterized in that, The method for determining the type of temporary landing scenario includes: The sag depth coefficient is calculated based on the voltage amplitude change, and the sag depth threshold is set as Kth and the sag duration threshold is set as Tth. When the descent depth coefficient is less than or equal to Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a shallow descent with a short duration; when the descent depth coefficient is greater than Kth and the descent duration is less than or equal to Tth, the descent scenario type is determined to be a deep descent with a short duration; when the descent depth coefficient is less than or equal to Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a shallow descent with a long duration; when the descent depth coefficient is greater than Kth and the descent duration is greater than Tth, the descent scenario type is determined to be a deep descent with a long duration.

4. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 2, characterized in that, The method for constructing the safe voltage withstand corridor includes: A two-dimensional geometric coordinate system is established with the physical communication node distance of each load node as the horizontal axis and the node voltage as the vertical axis. The zero point of the horizontal axis is set at the output end of the power router. A safe voltage withstand corridor is constructed in the two-dimensional geometric coordinate system. The safe voltage withstand corridor includes an upper boundary line and a lower boundary line. The upper boundary line is a horizontal straight line, and the lower boundary line is a stepped broken line.

5. The multi-source power coordinated dispatch method based on spatial topology constraints according to claim 4, characterized in that, The vertical axis coordinate value of the upper boundary line is equal to the upper limit of the voltage that the power router is allowed to output; The lower boundary line is set at the horizontal axis coordinate position of each load node, which is the lower limit of the voltage withstand level of the corresponding load terminal. The lower boundary line between two adjacent load nodes is taken as the higher of the two load nodes' lower voltage withstand level values.

6. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 4, characterized in that, The method for drawing the spatial pressure drop line includes: Calculate the voltage drop slope per unit distance, set the output voltage setting value of the power router, use the output voltage setting value of the power router as the y-intercept of the spatial voltage drop line, and use the negative value of the voltage drop slope per unit distance as the slope of the spatial voltage drop line. Plot the spatial voltage drop line in the two-dimensional geometric coordinate system where the safe voltage withstand corridor is located.

7. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 6, characterized in that, The method for determining the load list to be trimmed includes: Determine whether the space voltage drop line falls completely inside the safe voltage withstand corridor. If the space voltage drop line does not fall completely inside the safe voltage withstand corridor, calculate the voltage withstand level lower limit of each load node in the section where the space voltage drop line passes through the lower boundary line of the safe voltage withstand corridor, and subtract the estimated node voltage value at the corresponding horizontal axis coordinate position of the space voltage drop line to obtain the voltage deficiency difference. Add the load node with the largest voltage deficiency difference to the list of loads to be cut.

8. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 3, characterized in that, The energy storage discharge rate is determined based on a differentiated configuration according to the type of transient scenario and the total power gap.

9. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 6, characterized in that, The method for constructing the load physical topology multi-way tree includes: The power router output is set as the root node and each load terminal is set as a leaf node. A load physical topology multi-branch tree is constructed based on the physical communication node distance and cable branch connection method of each load node.

10. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 9, characterized in that, The method for executing the spatial sequential load shedding includes: In the list of loads to be cut, loads are cut off one by one in descending order of cumulative inductive reactance distance value until the cumulative cut-off power reaches the cut-off power limit. When the current load node to be cut off shares the same parent node with the previous cut-off load node in the load physical topology multi-branch tree, the waiting time interval is set to the basic waiting time interval multiplied by the shortening factor. The shortening factor is equal to the inductive reactance distance value of the branch cable where the current load node to be cut off is located divided by the cumulative inductive reactance distance value of the current load node to be cut off. When the current load node to be cut off does not share the same parent node with the previous cut-off load node in the load physical topology multi-branch tree, the waiting time interval is set to the basic waiting time interval.

11. The multi-source power coordinated scheduling method based on spatial topology constraints according to claim 10, characterized in that, The method for performing the closed-loop calibration includes: After the multi-source power compensation operation and spatial sequential load shedding are completed, the actual bus voltage value is monitored. The actual bus voltage value is compared with the power router output voltage setting value to calculate the voltage deviation. The energy storage discharge rate and the power router output voltage setting value are corrected according to the voltage deviation. During the grid sag recovery phase, the disconnected loads are restored one by one from the near end to the far end in the spatial order opposite to the spatial sequential load shedding order. After each load terminal is restored, the unit distance voltage drop slope is recalculated and it is re-determined whether the spatial voltage drop line still falls within the safe voltage withstand corridor.

12. A multi-source power cooperative scheduling system based on spatial topology constraints, used to implement the multi-source power cooperative scheduling method based on spatial topology constraints as described in any one of claims 1-11, characterized in that, The system includes: Collaborative Decision Module: This module collects load operation data, the physical communication node distances between each load node, and the cable branch connection methods between each load. Based on the physical communication node distances between each load node and the load operation data, it constructs a safe voltage withstand corridor and draws a spatial voltage drop line. By projecting the spatial voltage drop line onto the safe voltage withstand corridor, it determines the multi-source power collaborative calculation, identifies the load list to be cut, the energy storage discharge rate, the photovoltaic output lock value, and the rectifier control parameter adjustment value. The calibration module performs the following: it constructs a multi-branch tree of the physical topology of the load based on the distance between physical communication nodes and the cable branch connection method, calculates the cumulative inductive reactance distance value of each leaf node, performs multi-source power compensation operation according to the rectifier control parameter adjustment value, energy storage discharge rate and photovoltaic output lock value, performs spatial sequential load removal on the load list to be cut off in descending order of cumulative inductive reactance distance value, and performs closed-loop calibration based on the actual bus voltage value after removal.

Images