A control method for improving the utilization rate of renewable energy in real time

Abstract

The present invention discloses a control method for improving the utilization rate of renewable energy in real time, specifically relating to the technical field of voltage collapse protection in DC distribution networks, and is used to solve the problems of voltage collapse propagation and reduced utilization rate of renewable energy caused by photovoltaic sudden drops; the method is achieved by the following steps: real-time monitoring of photovoltaic output, bus voltage and load current; prediction of voltage collapse propagation path; identification of active migration sources based on the timing difference between voltage drop and load current mutation; calculation of the current increment upper limit of the active migration source cluster; generation of cross-regional compensation instructions based on current increment and energy storage adjustable power; issuance of compensation instructions to energy storage units and synchronous locking of current regulation authority of active migration sources; dynamic coupling of collapse path prediction and migration source identification to achieve coordinated control of cross-regional power reverse compensation and local load behavior freezing, thereby blocking the chain propagation of voltage collapse and improving the on-site photovoltaic absorption rate.

Description

Technical Field

[0001] The present invention relates to the technical field of voltage collapse protection of a DC power distribution network, and more particularly, to a control method for improving the utilization rate of renewable energy in real time. Background Art

[0002] Currently, industrial park energy systems are gradually adopting a PV-storage-DC-flexible architecture, integrating photovoltaic power generation, energy storage units, and DC loads through a DC distribution network. This PV-storage-DC-flexible architecture reduces AC-DC conversion losses and adapts to the needs of new DC loads such as electric vehicle charging stations and data centers. Existing technologies utilize energy management systems to implement photovoltaic forecasting, energy storage scheduling, and load control, aiming to increase the local consumption rate of renewable energy.

[0003] Then, there is currently a chain reaction problem of DC network voltage stability and load dynamic response. When the output of local photovoltaic power drops sharply due to shading, the DC bus voltage drops regionally. The constant power load will automatically increase the current to maintain the rated power, causing the voltage drop of adjacent lines to increase, causing the voltage collapse range to spread to non-fault areas. The existing control strategy relies on local voltage regulation and overload protection mechanisms, and cannot block the power migration effect caused by the self-regulation behavior of the load, resulting in the deterioration of system-level power supply quality and the decline of renewable energy utilization. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a control method for improving the utilization rate of renewable energy in real time to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A control method for improving the utilization rate of renewable energy in real time, comprising:

[0007] S1, real-time monitoring of photovoltaic output data, DC bus voltage data and constant power load current data in each area of ​​the park;

[0008] S2. When it is detected that the voltage data of the local DC bus drops beyond a preset threshold and is accompanied by a sudden drop in the photovoltaic output data, the voltage collapse propagation path is predicted;

[0009] S3. Compare the timing difference between the voltage drop start time and the current data mutation time of the constant power load on the voltage collapse propagation path. If the timing difference is less than the critical delay, determine that the corresponding constant power load is an active migration source.

[0010] S4, identifying the active migration source cluster on the voltage collapse propagation path, and calculating the upper limit of the current increment required to maintain the rated power based on the constant power load current data;

[0011] S5. Generate a cross-region active compensation power instruction based on the current increment upper limit and the real-time adjustable power of the energy storage unit;

[0012] S6. Send cross-region active compensation power instructions to the energy storage units on the voltage collapse propagation path, and at the same time lock the current regulation authority of the active migration source load.

[0013] Furthermore, real-time monitoring of photovoltaic output data, DC bus voltage data, and constant power load current data in each area of ​​the park is carried out, including:

[0014] Divide the park DC network into multiple monitoring areas based on electrical distance;

[0015] A voltage monitoring point is set at the DC bus in each monitoring area, an output monitoring point is set at the photovoltaic grid connection point, and a current monitoring point is set at the constant power load access point;

[0016] Using synchronous measurement technology based on the precise time protocol, the photovoltaic output data, DC bus voltage data and constant power load current data of each monitoring area are synchronously recorded with microsecond timestamps.

[0017] Furthermore, when it is detected that the voltage data of the local DC bus drops beyond a preset threshold and is accompanied by a sudden drop in photovoltaic output data, the voltage collapse propagation path is predicted, including:

[0018] When the DC bus voltage data of any monitoring area drops significantly within a preset time window and the PV output data of the same monitoring area drops sharply within the same time window, voltage collapse propagation path prediction is performed;

[0019] Based on the voltage data of the DC bus in each monitoring area, the voltage sag depth value is calculated. The voltage sag depth value is the ratio of the difference between the rated voltage and the actual voltage of the DC bus to the rated voltage;

[0020] Arrange the spatial dimensions from near to far according to the electrical distance of the monitoring area, and arrange the time dimension according to the time series to construct the spatiotemporal distribution tensor of the voltage sag depth;

[0021] A three-dimensional convolution kernel is used to perform convolution operation on the spatiotemporal distribution tensor to extract the gradient field characteristics of voltage sag propagation;

[0022] Generate a dynamic weight distribution matrix based on the electrical admittance matrix elements between monitoring areas to weight and enhance the spatial components of the gradient field characteristics;

[0023] According to the spatial gradient direction of the weighted gradient field characteristics, the voltage collapse propagation path prediction result is output, which spreads from the monitoring area with the deepest voltage sag to the monitoring area with adjacent electrical distance.

[0024] Furthermore, the weight of the 3D convolution kernel in the spatial dimension is negatively correlated with the electrical distance between the monitoring areas, and the weight in the temporal dimension decays as the time interval increases.

[0025] Furthermore, the timing difference between the voltage sag start time and the current data mutation time of the constant power load on the voltage collapse propagation path is compared. If the timing difference is less than the critical delay, the corresponding constant power load is determined to be an active migration source, including:

[0026] Extract the voltage drop starting time of the monitoring area with the deepest voltage sag from the voltage collapse propagation path prediction results;

[0027] Obtain constant power load current data for each monitoring area along the voltage collapse propagation path, and use a mutation point detection method based on phase-locked loop technology to identify the moment of current mutation;

[0028] Calculate the absolute time difference between the voltage drop start time and the current sudden change time in the same monitoring area as the timing difference;

[0029] Compare the timing difference with the critical delay, which is set based on the minimum opening time of the DC circuit breaker;

[0030] When the time sequence difference is less than the critical delay and the current change direction is increasing, the constant power load in the corresponding monitoring area is marked as an active migration source.

[0031] Furthermore, the active migration source cluster on the voltage collapse propagation path is identified, and the upper limit of the current increment required to maintain the rated power is calculated based on the constant power load current data, including:

[0032] According to the spatial gradient direction of the voltage collapse propagation path prediction results, the monitoring areas with adjacent electrical distances and marked as active migration sources are merged into active migration source clusters;

[0033] Obtain current data for each constant-power load in the active migration source cluster and calculate the average steady-state current before the fault occurs.

[0034] Based on the current minimum DC bus voltage and the load rated power, the minimum current increment required to maintain the rated power is calculated;

[0035] The minimum current increments of all constant power loads in the active migration source cluster are accumulated to serve as the upper limit of the cluster current increment.

[0036] Furthermore, based on the current increment upper limit and the real-time adjustable power of the energy storage unit, a cross-region active compensation power instruction is generated, including:

[0037] Multiply the current increment upper limit of the active migration source cluster by the current minimum DC bus voltage value to convert it into the power demand upper limit of the active migration source cluster;

[0038] Obtain the real-time adjustable power of each energy storage unit on the voltage collapse propagation path;

[0039] Compensation power is allocated to each energy storage unit in the order of the electrical distance between the monitoring area where each energy storage unit is located and the active migration source cluster, from closest to farthest. The allocated amount does not exceed the real-time adjustable power of the energy storage unit, and the cumulative allocated amount does not exceed the power demand limit.

[0040] Generate a cross-region active compensation power instruction including the energy storage unit identification and the allocated compensation power value.

[0041] Furthermore, the real-time adjustable power is the difference between the maximum allowable discharge power of the energy storage unit in the current operating state and the current actual output power.

[0042] Furthermore, cross-region active compensation power instructions are issued to energy storage units on the voltage collapse propagation path, while the current regulation authority of the active migration source load is locked, including:

[0043] Sending the allocated compensation power value contained in the cross-region active compensation power instruction to the corresponding energy storage unit power controller on the voltage collapse propagation path, so that the energy storage unit adjusts the output power according to the allocated compensation power value;

[0044] Sending a permission lock command to the current regulator of the active migration source load to prohibit the current regulator from automatically adjusting the load current based on the DC bus voltage change;

[0045] During the period when the authority is locked, the current value of the active migration source load is maintained at the current value at the locking moment until the system returns to steady-state operation.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. Based on the dynamic prediction of voltage collapse propagation paths and the sequential identification of active migration sources, an early intervention mechanism for collapse propagation was established. Through the coordinated control of cross-regional active compensation power commands and the locking of local current regulation authority, a dual defense mechanism of "remote power compensation + local behavior freezing" was established. Energy storage units injected compensation power in the reverse direction along the collapse path, directly offsetting the power shortfall caused by the sudden PV dip. Simultaneously, the current regulation authority of the active migration source load was locked, fundamentally interrupting the vicious cycle caused by load self-regulation. This effectively curbed the spread of regional voltage sags and compressed the fault impact range to the initial disturbance area.

[0048] 2. Accurate matching and compensation of power shortages are achieved through calculation of the upper limit of cluster current increment and dynamic allocation of energy storage power. The energy storage scheduling strategy based on electrical distance prioritizes the use of available energy storage capacity in adjacent collapse paths to minimize power transmission losses. The authority locking mechanism maintains constant current operation of the actively migrated source load, avoiding energy waste caused by load removal triggered by traditional overload protection. While maintaining system voltage stability, it ensures the continuous grid-connected operation of photovoltaic power generation units, significantly improving the local absorption rate and utilization rate of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a control method for improving the utilization rate of renewable energy in real time according to the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] Example: Figure 1 The present invention provides a control method for improving the utilization rate of renewable energy in real time, comprising:

[0052] S1, real-time monitoring of photovoltaic output data, DC bus voltage data and constant power load current data in each area of ​​the park;

[0053] S2. When it is detected that the voltage data of the local DC bus drops beyond a preset threshold and is accompanied by a sudden drop in the photovoltaic output data, the voltage collapse propagation path is predicted;

[0054] S3. Compare the timing difference between the voltage drop start time and the current data mutation time of the constant power load on the voltage collapse propagation path. If the timing difference is less than the critical delay, determine that the corresponding constant power load is an active migration source.

[0055] S4, identifying the active migration source cluster on the voltage collapse propagation path, and calculating the upper limit of the current increment required to maintain the rated power based on the constant power load current data;

[0056] S5. Generate a cross-region active compensation power instruction based on the current increment upper limit and the real-time adjustable power of the energy storage unit;

[0057] S6. Send cross-region active compensation power instructions to the energy storage units on the voltage collapse propagation path, and at the same time lock the current regulation authority of the active migration source load.

[0058] The specific implementation process of dividing the campus DC network into multiple monitoring areas based on electrical distance is as follows: first, obtain the complete topological structure information of the campus DC network, which includes the connection relationship and line parameters of all electrical nodes; calculate the equivalent impedance between any two nodes by establishing a node admittance matrix, where the element values ​​of the node admittance matrix are determined by the line resistance and line inductance values; define the modulus of the equivalent impedance as the electrical distance, which represents the strength of the mutual influence of the voltage between nodes; use a graph theory-based clustering algorithm to perform regional division, and use electrical distance as a similarity measurement indicator in the clustering process. When the electrical distance between two nodes is less than a dynamically set threshold, they are classified into the same monitoring area. The dynamically set threshold is adjusted according to the real-time operating status of the power grid, and the adjustment basis includes the ratio of the line rated current carrying capacity to the minimum short-circuit current. The specific implementation method is to extract the critical impedance value from the historical fault data in the line protection constant value management system as a benchmark reference.

[0059] The technical solution for deploying monitoring points within the monitoring area is as follows: the DC bus voltage monitoring point is set at the physical connection between the regional trunk line and the DC bus, and a closed-loop Hall voltage sensor is used to achieve non-contact measurement. Its working principle is to convert the bus voltage value into a proportional current signal based on the principle of magnetic field induction, and then output a standard voltage signal through a high-precision instrument amplifier; the photovoltaic output monitoring point is located in the photovoltaic array DC output junction box. By simultaneously collecting the instantaneous values ​​of the photovoltaic DC side voltage and current, the power value is calculated in real time using an analog multiplier circuit. The refresh cycle of the power calculation result is synchronized with the power frequency cycle of the power grid; the constant power load current monitoring point is integrated into the load control unit, using a combination of a shunt and an isolation amplifier. The shunt converts the load current into a millivolt voltage signal, and the isolation amplifier outputs the sampled value after eliminating common-mode interference.

[0060] Microsecond-level synchronization measurement is achieved through the following methods: deploying a distributed clock system based on the Precision Time Protocol, which consists of a master clock node and several slave clock nodes; the master clock node receives the atomic clock time signal sent by the Global Positioning System satellite and generates a reference clock pulse; a dedicated synchronization network built through single-mode optical fiber uses a two-way delay compensation mechanism to distribute the reference clock to the slave clock nodes in each monitoring area. The specific implementation process of delay compensation includes measuring the round-trip time of the optical fiber transmission path, calculating the offset caused by path asymmetry, and dynamically adjusting the slave clock phase; when each monitoring point collects data, the local slave clock generates a timestamp containing absolute time information. The timestamp generation mechanism uses the method of aligning the clock counter with the satellite second pulse to ensure that the time deviation between different monitoring points is controlled at the sub-microsecond level.

[0061] The complete process of data recording and transmission is as follows: Each monitoring point device completes signal acquisition and analog-to-digital conversion within a fixed sampling interval, and the sampling interval is set to match the characteristic harmonic frequency of the power grid; the converted digital quantity and the timestamp corresponding to the sampling moment are encapsulated into a data message. The message structure includes a message header, a timestamp segment, a data segment and a checksum; the message header identifies the monitoring area number and sensor type, the timestamp segment uses the coordinated universal time format to record the year, month, day, hour, minute, second and microsecond information, and the data segment contains floating-point values ​​of photovoltaic output, DC bus voltage, and constant power load current; the message is transmitted to the central processing unit via an industrial Ethernet switch, and the transmission process uses a priority queue mechanism to ensure the real-time performance of key data.

[0062] The verification method for monitoring area division involves injecting a small signal current perturbation of a specific frequency into a selected representative monitoring area under normal operating conditions; using a spectrum analyzer to measure the voltage response amplitude on the DC bus in other monitoring areas; and determining that the area division meets the electrical coupling characteristic requirements when the voltage response attenuation between monitoring areas whose electrical distance exceeds a set threshold reaches a preset decibel value. The optimization criteria for voltage monitoring point locations are as follows: temporary measurement points are placed at different locations on the DC bus, and the transient response consistency of the voltage waveform at each point is compared under load sudden change conditions; the point with the smallest transient oscillation amplitude and the fastest response speed is selected as the final monitoring location.

[0063] Specific measures for data quality control are as follows: real-time rationality verification of constant-power load current data is implemented by querying the load equipment archive to obtain the rated power value, and calculating the expected current range based on the current DC bus voltage. When the measured current value continuously exceeds this range, the sensor calibration process is triggered; a change rate threshold is set for photovoltaic output data. When the output change in adjacent sampling periods exceeds this threshold, the backup sensor is automatically enabled for data review; and digital filtering is applied to the DC bus voltage data. The filter design adopts a finite impulse response structure, and the cutoff frequency is determined according to the characteristic frequency of the power electronic switching device.

[0064] The monitoring system's dynamic maintenance mechanism includes the following: When a new distributed power source is connected, the electrical distance between the power source access point and the center point of the existing monitoring area is recalculated; if the minimum electrical distance is greater than the set ratio of the original division threshold, the regional reorganization process is initiated, and the continuity mark of historical monitoring data is retained during the reorganization process; all changes to monitoring configuration parameters are synchronized to the central database via a version control protocol to ensure the consistency of regional definitions during the fault analysis period. The reliability of the communication network is guaranteed by combining physical dual paths with logical redundancy protocols. When the signal quality index of the primary communication path falls below the threshold, it automatically switches to the backup path. The switching process ensures that the data packet sequence is not interrupted.

[0065] The central database's architectural design and implementation utilizes a time-partitioned storage strategy, creating data partitions based on fixed time periods. Within each partition, a secondary index based on monitoring area identifiers is established, using a balanced binary tree algorithm for the index structure. The data compression algorithm utilizes a hybrid lossy and lossless mode, employing lossless compression for timestamp fields and adaptive lossy compression for monitoring data fields, with the compression rate dynamically adjusted based on data type. The data access interface defines a standardized query language, supporting multi-dimensional searches based on time range, monitoring area, and data type.

[0066] The above technical solution enables high-precision synchronous acquisition and standardized processing of electrical parameters across the entire park. The electrical distance calculation method accurately reflects the electrical coupling characteristics of the DC network, providing a spatial topological foundation for subsequent voltage collapse analysis. A microsecond-level time synchronization mechanism eliminates time-scale errors in distributed measurements, making transient events at different locations comparable. A dynamic threshold setting mechanism ensures that monitoring area divisions adapt to changes in grid operation. A multi-level data quality control process effectively suppresses the impact of measurement noise and equipment anomalies. The optimized design of the database system meets the requirements for storing and rapidly retrieving massive amounts of real-time data.

[0067] When the DC bus voltage data of any monitoring area is monitored to have a significant drop within a preset time window and the PV output data of the same monitoring area also experiences a sudden drop within the same time window, the specific implementation process for performing voltage collapse propagation path prediction is as follows: the DC bus voltage data series and PV output data series for the target monitoring area within the preset time window are extracted from the central database; the deviation between the sliding average of the voltage data series and the rated voltage is calculated. The upper limit of the steady-state fluctuation range is determined by analyzing the historical operating data of the area for 30 consecutive days, taking the 95th percentile of the voltage fluctuation value as a benchmark, and then multiplying it by a safety factor. The safety factor is set as an adjustable parameter between 1.2 and 1.5 according to the grid operating regulations. A significant drop is determined when the deviation value continuously exceeds this upper limit. The first-order difference value of the PV output data series is simultaneously calculated. The normal operating fluctuation threshold is obtained by parsing the technical documentation provided by the PV inverter manufacturer to extract the maximum allowable power fluctuation parameter. A sudden drop is determined when the difference values ​​of multiple consecutive sampling points are all negative and the cumulative change exceeds this threshold.

[0068] The process for calculating the voltage sag depth based on the DC bus voltage data of each monitoring area is as follows: the actual DC bus voltage value measured synchronously in each monitoring area at the current moment is read; the pre-stored DC bus rated voltage value is called from the power grid parameter database. This rated voltage value is determined by the engineering design documents and entered into the database when the system is put into operation; the voltage sag depth value is calculated using a standardized process, specifically by dividing the difference between the rated voltage and the actual voltage by the rated voltage to obtain a dimensionless percentage value; this calculation process is performed synchronously in all monitoring areas throughout the entire park, generating a set of voltage sag depth values ​​as input for spatiotemporal analysis.

[0069] The method for constructing a spatiotemporal distribution tensor of voltage sag depth includes: in the spatial dimension, the monitoring areas are sorted from small to large according to the electrical distance value, and the sorting algorithm adopts an improved bubble sort method; in the temporal dimension, the time series is divided into fixed time steps, and the time step is set as an integer multiple of the sampling interval, and the specific multiple is determined according to the transient response characteristics of the power grid; a three-dimensional tensor data structure is constructed, in which the first dimension index corresponds to the sequence number of the sorted monitoring area, the second dimension index corresponds to the sequence number of the time series, and the third dimension stores the voltage sag depth value of the corresponding spatiotemporal point; in the data filling rule, the missing data is processed using a spatial interpolation algorithm based on electrical distance, which calculates the interpolation result according to the electrical distance weights of adjacent monitoring areas.

[0070] The implementation steps of using a three-dimensional convolution kernel to perform convolution operations on the spatiotemporal distribution tensor are as follows: the spatial dimension size of the three-dimensional convolution kernel is determined according to the electrical coupling strength of the monitoring area, and the coverage range is set to include the main electrically related areas; the weight distribution of the spatial dimension adopts the inverse electrical distance weighted model, which normalizes the inverse of the electrical distance as the weight coefficient; the weight distribution of the time dimension adopts an exponential decay model, and the attenuation coefficient is set based on the time constant of the transient process of the power system; the convolution operation uses a zero-filling mode to keep the feature map size unchanged, and the gradient field features extracted by the convolution layer contain information on the spatial propagation direction and time evolution rate.

[0071] The technical solution for generating a dynamic weight allocation matrix based on the electrical admittance matrix elements between monitoring areas is as follows: obtaining the node admittance matrix; extracting the absolute value of the non-diagonal elements of the admittance matrix as the electrical correlation strength index between areas; normalizing the correlation strength index using the maximum normalization method; constructing a square matrix with the same dimension as the gradient field feature space as the dynamic weight allocation matrix; the weighted enhancement operation is implemented using matrix element multiplication, and the enhancement coefficient is dynamically adjusted according to the tightness of the electrical connection.

[0072] The logic of the output voltage collapse propagation path prediction result is as follows: in the weighted enhanced gradient field characteristics, a two-dimensional difference algorithm is used to calculate the spatial gradient vector at each time point; the modulus of the gradient vector is calculated using the Euclidean norm formula; the direction angle is calculated using the inverse tangent function; the method for locating the collapse starting point is to search for the global maximum point of the voltage sag depth value in the entire network; the path search algorithm uses the nearest neighbor search method guided by the gradient direction, and the search step size is dynamically adjusted according to the electrical distance; the final output path sequence contains the monitoring area numbers arranged in chronological order.

[0073] The preset time window is set as follows: the lower limit is determined by the fastest operating time of the DC protection system, which is extracted from the relay protection setting sheet; the upper limit is set based on the duration of a typical voltage collapse process, which is calculated through power system simulation; and the actual window length is determined by optimizing historical fault data analysis. The electrical admittance matrix is ​​updated as follows: when the grid topology changes, the node admittance matrix is ​​recalculated based on the change notification sheet; regular verification is performed using a test signal injection method, with the test signal amplitude set proportionally to the system capacity.

[0074] The optimization process for the 3D convolution kernel parameters includes: determining the kernel size using a grid search method that covers typical electrical correlation scales; optimizing the weight decay coefficient using gradient descent, with the objective function defined as the spatiotemporal overlap between the predicted and actual paths; and handling exceptions in the dynamic weight allocation matrix using a weighted average of neighboring regions based on electrical similarity for missing data and a median filter with an adaptive threshold for outlier handling.

[0075] The verification scheme for path prediction results involves building an equivalent model of the campus power grid on a real-time digital simulation platform. A gradual short-circuit model matching the measured data is used as the fault injection model. The coincidence index is calculated using a spatiotemporal path matching algorithm that considers both the matching of regional sequences and time intervals. The output data structure follows the power grid data exchange standard and includes a time-series list of regional numbers.

[0076] Post-processing of voltage sag depth values ​​involves setting the limiting boundary according to power system safety operation regulations and determining the time constant of inertia delay based on the response characteristics of the measurement system. Optimization measures for constructing the spatiotemporal distribution tensor include: data compression using the Tucker decomposition algorithm based on eigenvalue preservation; and a sliding window update mechanism implemented using a ring buffer. Spatial gradient direction calculation uses an eight-directional Sobel operator group to improve directional identification accuracy.

[0077] The above technical solution enables prediction of voltage collapse propagation paths: the parameters for triggering conditions are derived from actual grid operating data; voltage sag depth calculations are standardized to ensure comparability; space-time tensor construction fully preserves fault propagation characteristics; the three-dimensional convolution kernel design integrates electrical distance and time decay characteristics; a dynamic weighting mechanism strengthens critical propagation paths; and the gradient tracking algorithm has clear physical meaning. The time complexity of the prediction process has been optimized to meet real-time requirements, providing accurate input for active control.

[0078] The specific implementation process for extracting the voltage sag start time in the monitoring area with the deepest voltage sag from the voltage collapse propagation path prediction results is as follows: based on the voltage collapse propagation path sequence, which is a list of monitoring area numbers arranged in chronological order, the first monitoring area in the sequence is located as the area with the deepest voltage sag; the DC bus voltage historical data of this area during the voltage collapse event is retrieved from the central database, and the data time range covers the entire voltage sag process; the voltage signal is processed using the variational mode decomposition algorithm, which decomposes the signal into a finite number of intrinsic mode functions by iteratively solving a constrained variational problem; the intrinsic mode component containing the main sag characteristics is selected, and the voltage sag start time is determined by detecting the zero crossing point where the first-order derivative of this component changes from positive to negative.

[0079] The technical solution for obtaining constant-power load current data of each monitoring area on the voltage collapse propagation path and identifying the mutation moment is as follows: according to the voltage collapse propagation path sequence, the constant-power load current data stream of each monitoring area is accessed in sequence, and the data comes from the deployed load current monitoring points; a mutation detection method based on digital phase-locked loop technology is adopted, and the specific implementation includes: constructing a digital phase-locked loop system, which consists of a digital phase detector, a digital loop filter and a digitally controlled oscillator; the phase detector adopts a multiplier structure to multiply the input current signal with the local orthogonal signal output by the digitally controlled oscillator; the loop filter adopts a second-order infinite impulse response structure, and its bandwidth parameter is set according to the maximum change rate of the load current; when the current suddenly changes, the instantaneous value of the error voltage output by the phase detector exceeds the dynamic judgment threshold. The dynamic threshold is set by calculating the root mean square value of the current signal in the sliding time window and multiplying it by an adjustable coefficient. The adjustable coefficient ranges from 2.5 to 3.5; the first sampling point when the error voltage exceeds the threshold is locked as the current mutation moment.

[0080] The process of calculating the absolute time difference between the voltage drop start time and the current mutation time in the same monitoring area is as follows: for each monitoring area in the voltage collapse propagation path sequence, the voltage drop start time and current mutation time of the area are extracted from the timestamp database, and both times are stored in the coordinated universal time format; the time difference is calculated by direct subtraction operation, and the calculation result is stored in milliseconds; when multiple current mutation points are detected in the same area, the mutation point with the smallest time difference with the voltage drop time is selected for calculation using the nearest neighbor priority principle; the storage structure of the calculation result is a tuple sequence containing the monitoring area number and the time difference.

[0081] The judgment logic for comparing the time difference with the critical delay is as follows: the setting of the critical delay is determined based on the minimum opening time of the DC circuit breaker, which is specifically implemented as follows: query the opening time parameter in the technical manual of the circuit breaker equipment, which is measured under standard test conditions; consider the inherent action delay of the protection relay and increase the action time margin of the protection device; the final critical delay value is the sum of the opening time and the action margin, and the action margin is set to 15% to 25% of the opening time; it is determined to be an active migration source when the following two conditions are met at the same time: Condition 1 is that the time difference is less than the critical delay, and Condition 2 is that the direction of current change is increasing; the method for determining the direction of current change is: take the current average value of the three sampling cycles before the current mutation point as the reference value, and take the current average value of the three sampling cycles after the mutation point as the comparison value. When the comparison value is greater than 105% of the reference value, it is determined to be in an increasing direction.

[0082] The parameters of the digital phase-locked loop system include: the center frequency of the numerically controlled oscillator is set to 1.2 to 1.8 times the fundamental frequency of the load, with the specific value determined by the parameters in the load device library; the loop filter cutoff frequency is calculated as the maximum current rate of change divided by twice the rated current value; and the adaptive adjustment mechanism of the adjustable coefficient in the dynamic threshold automatically reduces the coefficient value when the system detects high-frequency interference and increases the coefficient value when the load is operating smoothly. When multiple candidate sudden change points are detected, the optimization strategy is to calculate the absolute time difference between each candidate point and the voltage drop moment and select the candidate with the smallest time difference; if multiple minimum time difference points exist, the point with the largest current change slope is selected.

[0083] The dynamic adjustment mechanism for the critical delay is as follows: corrections are made based on the actual operating life of the circuit breaker, with the critical delay increasing by 2% for every five years of operation; compensation is made based on ambient temperature, with the temperature compensation coefficient increasing by 1% for every 10°C above the standard ambient temperature; and adjustments are made based on recent maintenance records, with a 5% reduction if maintenance has been performed within three months. A secondary verification method for the direction of current change is to verify that the current value remains above the pre-change level for ten sampling periods after the mutation point, with the fluctuation range not exceeding 10% of the rated value.

[0084] The active migration source tagging process is as follows: For constant power loads that meet the criteria, an active migration source tag is added to their device profile; a migration event record is generated, containing parameters such as the device number, time difference, current change, and timestamp; this record is pushed to the voltage control decision system in real time via a message queue. The principle for handling critical delay boundary conditions is as follows: when the time difference falls between 95% and 105% of the critical delay, a secondary verification process is initiated. This secondary verification uses the voltage drop acceleration indicator. If the voltage change rate exceeds the set threshold, the original judgment is maintained.

[0085] The data anomaly processing mechanism includes: when the starting time of the voltage drop is missing, linear interpolation is performed using the time of two adjacent areas in the propagation path; when the current mutation detection fails continuously, an alternative detection algorithm based on wavelet transform is enabled, which identifies mutations by detecting the maximum value point of the wavelet coefficient modulus; clock synchronization compensation for time difference calculation is: when the two moments come from different clock sources, an additional clock synchronization error compensation value is added, which is taken as twice the clock synchronization accuracy as a conservative estimate.

[0086] The verification scheme for the judgment results is as follows: construct a test scenario with active migration load on a real-time digital simulation platform. The test scenario includes typical operating conditions such as batch startup of data center servers and centralized charging of electric vehicle charging stations; record the correspondence between the judgment results and the actual migration source; calculate the false positive rate index, and automatically adjust the detection parameters when the false positive rate exceeds 5%; the output interface uses the IEC 61850 protocol to communicate with the control system, and the transmission delay is controlled within 100 microseconds.

[0087] The above technical solution enables precise identification of active migration sources: signal decomposition technology is used to detect the onset of voltage sags, eliminating noise interference; current mutation detection is modified with phase-locked loop technology to adapt to the characteristics of DC systems; time difference calculation establishes a unified time-scale benchmark; critical delay settings integrate device parameters and environmental factors; direction determination uses a multi-verification mechanism; and boundary condition handling incorporates a secondary verification process. The average execution time of the identification process in the industrial controller is 1.5 milliseconds, meeting the requirements for rapid voltage collapse control.

[0088] According to the spatial gradient direction of the voltage collapse propagation path prediction results, the monitoring areas with adjacent electrical distances and marked as active migration sources are merged into an active migration source cluster. The specific implementation process is as follows: spatial gradient direction data is extracted from the voltage collapse propagation path sequence, which contains the gradient vector direction angle information of each monitoring area during the propagation process; the criterion for determining electrical distance proximity is that the electrical distance between two monitoring areas is less than a specific ratio of the area division threshold, and the area division threshold is directly taken from the dynamically set clustering threshold; a breadth-first search algorithm is used to traverse all monitoring areas marked as active migration sources. When two areas simultaneously meet the electrical distance proximity conditions and the gradient direction angle difference is less than the set angle tolerance, they are classified into the same cluster; the cluster expansion termination rule is: when the electrical distance between the newly added area and any area in the cluster exceeds a set multiple of the cluster average electrical distance, the expansion is stopped; and finally a cluster topology structure containing cluster number, member area list, and center coordinates is generated.

[0089] The method for obtaining the current data of each constant-power load in the active migration source cluster and calculating the average steady-state current before the fault occurs is as follows: extract the current data series of each constant-power load in the cluster in a specific time window before the voltage collapse event from the real-time database, and the length of the time window is set according to the dynamic characteristics of the load; preprocess the current data series of each load, which includes outlier removal and limiting processing. Outliers are defined as data points that exceed the historical statistical range; the steady-state current average value is calculated using the time-weighted average method, and the weight distribution principle is: the weight of the recent data in the time window is greater than that of the remote data, and the specific weight ratio is adjusted according to the load response speed; the steady-state current average values ​​of all loads in the cluster are comprehensively calculated to obtain the overall steady-state current baseline value of the cluster.

[0090] The technical solution for calculating the minimum current increment required to maintain the rated power based on the current minimum DC bus voltage and the load rated power is as follows: obtain the minimum DC bus voltage in the monitoring area where the active migration source cluster is located in real time. This value is taken from the minimum value of the voltage monitoring point in the most recent multiple sampling cycles; call the rated power value of each constant power load from the equipment asset database. The rated power value is determined by the manufacturer's technical parameters and entered into the system when the equipment is put into operation; the calculation process of the minimum current increment is: divide the load rated power value by the current minimum DC bus voltage value, and then subtract the average steady-state current of the load; the calculation process performs a unified unit conversion process to ensure the consistency of power, voltage, and current units; impose a lower limit constraint on the calculation result of each load, and set it to zero when the calculated value is negative.

[0091] The implementation steps for accumulating the minimum current increments of all constant-power loads in the active migration source cluster as the cluster current increment upper limit are as follows: establish a sequential access mechanism for loads in the cluster; perform minimum current increment calculation on each load, and temporarily store the calculation results in a memory array; use a high-precision numerical accumulator for the accumulation operation, and the accumulation step is the minimum current increment value of a single load; implement dynamic verification during the accumulation process: when the increment value of a single load exceeds a set proportion of its rated current value, a review mechanism is triggered; the accumulation result is output as the cluster current increment upper limit, which represents the total current increase required to maintain the rated operation of all loads in the cluster under the current voltage conditions; the final output data structure contains the cluster number, the current increment upper limit value and the calculation timestamp.

[0092] The dynamic adjustment mechanism for the electrical distance proximity threshold is as follows: when the number of cluster areas exceeds a set value, the proximity threshold is proportionally relaxed; when the cluster includes a critical load area, the threshold is proportionally tightened. The adaptive rule for the time window for calculating the steady-state current average value is to adjust the window length based on the load current fluctuation characteristics, extending the window time as the fluctuation rate increases.

[0093] Exception handling for the minimum current increment calculation includes: when the DC bus minimum voltage is abnormally low, a valid historical voltage value is used as a substitute; when rated power data is missing, it is estimated based on typical parameters of similar equipment. The accumulation process is designed to be fault-tolerant: when a single load increment is detected, the historical operating data of that load is used as a substitute calculation basis.

[0094] The clustering results are verified by simulating active migration source cluster events in a simulation environment and measuring the degree of match between the clustering results and the actual distribution. The current increment upper limit is used as input to the voltage control module as a basis for load regulation decisions.

[0095] The above technical solution is used to identify active migration source clusters and calculate the upper limit of current increments: cluster division integrates spatial gradient and electrical distance constraints; steady-state current calculation adopts time-weighted method; minimum current increment calculation follows the power conservation principle; and multiple checks are implemented in the accumulation process.

[0096] For example, specific parameter settings include: the electrical distance adjacency determination ratio is set at 120% of the region division threshold, the gradient direction angle tolerance is set at 15 degrees, and the cluster expansion termination multiplier is set at 1.5 times; the steady-state calculation window is set at 60 seconds for data center loads and 10 seconds for industrial motor loads; the weight distribution is 1.5 times that of the more recent data; the current increment verification threshold is set at 50% of the rated current value; unit conversion is standardized to watts for power, volts for voltage, and amperes for current; the voltage replacement value for exception handling is the effective average over the last five minutes; and the cluster size threshold is set at five regions. The entire calculation process typically executes in 4 milliseconds on an industrial controller, meeting the requirements for rapid voltage collapse control.

[0097] The specific implementation process of converting the current increment upper limit of the active migration source cluster into the power demand upper limit is as follows: extract the current value from the cluster current increment upper limit data, which represents the total current increase required to maintain the rated operation of all constant power loads in the cluster; obtain the DC bus minimum voltage value in the monitoring area where the cluster is located in real time, and the voltage value comes from the minimum value record of the real-time monitoring system in the most recent multiple sampling cycles; calculate the power demand upper limit by multiplying the current increment upper limit value by the DC bus minimum voltage value; the calculation process performs unit unification and dimension conversion: the current unit is ampere, the voltage unit is volt, and the calculation result power unit is converted to kilowatt, and the conversion factor is divided by 1000; the calculation result is stored as a power demand upper limit record with a timestamp, which contains the cluster number and power value. For example, when the current increment upper limit is 50 amperes and the DC bus minimum voltage is 750 volts, the calculated power demand upper limit is 37.5 kilowatts.

[0098] The technical solution for obtaining the real-time adjustable power of each energy storage unit on the voltage collapse propagation path is as follows: extract three key parameters from the real-time database of the power grid energy management system: the rated maximum discharge power value of the energy storage unit, the current actual output power value of the energy storage unit, and the operating status flag of the energy storage unit; the rated maximum discharge power value is determined by the manufacturer's technical specification when the equipment is put into operation and entered into the system database; the current actual output power value is collected and updated at a fixed frequency through the data acquisition and monitoring system, for example, 10 times per second; the real-time adjustable power is calculated by subtracting the current actual output power value from the rated maximum discharge power value; the calculation process adds an operating status constraint: only when the operating status flag is in the ready state and the charge state is higher than the safety lower limit threshold is included in the calculation; the calculation result is truncated to the lower limit, and negative value results are forced to be set to zero; generate a data set containing the globally unique identifier of the energy storage unit and the real-time adjustable power value.

[0099] The implementation steps for allocating compensation power in order from near to far electrical distance are as follows: extract the distance value from the center position of the active migration source cluster to the monitoring area where each energy storage unit is located from the electrical distance; use the quick sort algorithm to sort the energy storage unit set in ascending order of distance value; establish a power allocation accumulator with an initial value set to zero; traverse the energy storage unit sequence in sorted order: for the current energy storage unit, calculate the allocable power margin, which is the power demand upper limit minus the current accumulator value; take the smaller value of the real-time adjustable power value and the allocable power margin as the actual allocation amount; accumulate the actual allocation amount to the power allocation accumulator; set two-level termination conditions for the allocation operation: condition one is that the accumulator value reaches more than 99% of the power demand upper limit to avoid floating-point calculation errors, and condition two is that all energy storage units are traversed; generate a temporary record containing the allocated power value for each energy storage unit.

[0100] The complete process for generating cross-region active compensation power instructions is as follows: creating an instruction data structure containing an instruction number, a coordinated universal time timestamp, and an expiration date field; encoding the energy storage unit allocation records generated during the allocation process in a specified format: each record contains a globally unique identifier for the energy storage unit, an allocated power value, and a priority tag; the priority tag is automatically set based on the electrical distance, with closer distances giving higher priority; the instruction validity period is set based on the duration of the voltage collapse process, taking 120% of the preset time window length; the final instruction is published via a message bus and transmitted to each regional energy storage control system using a publish-subscribe model; and an instruction log is generated and stored in a historical database for subsequent analysis.

[0101] The power conversion process handles exceptions as follows: When the DC bus minimum voltage falls below 30% of the rated value, a valid voltage value that hasn't triggered an alarm in the last five minutes is used as a replacement. When the current increment limit exceeds 200% of the historical maximum, a manual confirmation process is initiated. To ensure real-time adjustable power, if data from the data acquisition and monitoring system times out and is not updated, the current value is replaced with the average value of a sliding time window. The sliding window length is the 10 most recent valid sampling points.

[0102] The optimized design of the power allocation algorithm includes: Given the same electrical distance, energy storage units with shorter response times are prioritized; a single allocation limit of 20% of the energy storage unit's rated power is set to prevent sudden power fluctuations; and during the allocation process, the deviation between the power allocation accumulator and the power demand limit is monitored in real time, initiating a supplementary allocation process if the deviation exceeds 5%. The verification mechanism for the instruction generation phase involves injecting a typical voltage collapse scenario into a real-time digital simulation platform; verifying the matching of the total instruction amount with the demand limit, and triggering an alarm if the deviation exceeds a set threshold.

[0103] Key parameter setting basis: Power demand conversion uses the International System of Units to ensure dimensionality correctness; real-time adjustable power update frequency is synchronized with the control system scan cycle; the time complexity of the distance sorting algorithm is optimized to the logarithmic level; the allocation termination condition is set with a 99% threshold to avoid floating-point calculation errors; the instruction validity period is extended by 20% to cover the control signal transmission delay; the effective voltage replacement value in exception handling takes the average value of the non-alarm data in the last 5 minutes.

[0104] The above technical solution achieves precise control of cross-regional power compensation: power demand conversion strictly follows the physical laws of electric power; real-time adjustable power calculation reflects the actual operating capacity of the equipment; the allocation strategy integrates electrical distance and the dynamic response characteristics of the equipment; and command generation meets the communication standards of industrial control systems.

[0105] Specific implementation example: In the power conversion calculation, the upper limit of the current increment is 50 amperes, and the minimum voltage of the DC bus is 750 volts. The power demand upper limit calculation process is 50 amperes × 750 volts = 37,500 watts, which is converted to 37.5 kilowatts; in the real-time adjustable power calculation, the maximum discharge power of a certain energy storage unit is 200 kilowatts, and the current output is 80 kilowatts, so the real-time adjustable power is 120 kilowatts; after the allocation process is sorted by electrical distance, the nearest energy storage unit is allocated 30 kilowatts, and the next closest unit is allocated the remaining 7.5 kilowatts; the generation instruction contains two energy storage unit identifiers and their respective allocation values, and the validity period is set to 240 milliseconds based on 120% of the preset time window of 200 milliseconds.

[0106] The specific implementation process of sending the allocated compensation power value contained in the cross-region active compensation power instruction to the power controller of the energy storage unit corresponding to the voltage collapse propagation path is as follows: extracting the energy storage unit's globally unique identifier and the allocated compensation power value field from the generated cross-region active compensation power instruction data structure; establishing an instruction transmission channel, which uses the manufacturing message specification format for data encapsulation and converts the allocated compensation power value into a control signal frame format; the control signal frame includes a frame header check sequence, an energy storage unit address field, a power setting value field, and an effective time window field, where the power setting value field is in watts according to the International System of Units; transmitting the control signal frame to the local controller of the area where the target energy storage unit is located via an optical fiber communication network; after receiving the signal frame, the energy storage unit power controller performs three levels of verification: the first level verifies the integrity of the frame header, the second level verifies the address matching, and the third level verifies whether the power value is within the device's allowable adjustment range; if the verification passes, power adjustment is immediately performed, using a ramp control strategy that sets the power change rate cap at a specific proportion of the device's rated power per second, such as 10% of the rated power per second, to prevent sudden power changes; and after the adjustment is completed, sending an execution confirmation signal to the central control system, which contains the actual output power value and a timestamp.

[0107] The implementation method of sending an authority locking instruction to the current regulator of the active migration source load is as follows: extracting the device identifier of the target load from the active migration source tag list; generating an authority locking instruction data structure, which includes an operation type code, a load identifier, and a locking effective time field; the operation type code is defined as a specific value to represent the current regulation authority locking operation; the locking effective time is set to take effect within a specific number of milliseconds after the instruction is issued, for example, within 10 milliseconds; sending an instruction data packet to the current regulator of the target load through the control bus; after receiving the instruction, the current regulator performs an authority switching operation: first, the current current value is stored in a non-volatile memory as a locking reference value, and then the authority flag bit in the control register is modified, which controls whether to respond to changes in the DC bus voltage; after the authority is locked, the control logic of the current regulator switches to a fixed current mode, that is, the voltage sampling signal is ignored and the locking reference value current is continuously output; the authority locking status is represented by a specific bit of the status register, and the status bit is fed back to the monitoring system in real time.

[0108] The operating mechanism for maintaining the current value during the period of authority lock is as follows: a judgment standard for the system to restore steady state is established, which includes two parallel conditions: the first condition is that the fluctuation amplitude of the DC bus voltage does not exceed a specific percentage of the rated value within a specific number of seconds, for example, the fluctuation does not exceed 1% of the rated value within 60 seconds; the second condition is that the voltage collapse propagation path prediction result shows no propagation risk for multiple consecutive scanning cycles; during the authority lock period, the current regulator performs current closed-loop control at a fixed frequency: the actual load current value is collected, compared with the stored lock reference value, and a regulation signal is generated through the proportional-integral controller to drive the power device; when it is detected that the current deviation exceeds a specific proportion of the lock reference value, for example, more than 2%, the current calibration procedure is started, and the current value is resampled during calibration to update the lock reference value; multiple protections are implemented during the maintenance process: when the load temperature exceeds the safety threshold, it automatically switches to safety mode, and when communication interruption is detected, the last valid command state is maintained; after the system restores steady state, the central control system sends an authority unlock command, and the current regulator resumes the voltage-based automatic regulation function after verification.

[0109] Redundancy in the command transmission process includes the addition of error-checking codes to the control signal frame, initiating a retransmission mechanism when verification fails, with a retransmission interval of a specific number of milliseconds, such as 5 milliseconds, and a maximum number of retries. During power adjustment, power feedback verification is implemented, triggering a local closed-loop correction when the actual output power deviates from the set value by more than a specific percentage, such as 5%. Exception handling in the permission locking process involves automatically selecting the closest similar load as a backup lock target when the target load goes offline. If the lock command transmission times out, it is repeated via broadcast.

[0110] The dynamic compensation mechanism of the current maintenance link is to fine-tune the output current in real time according to the change of load impedance. The fine-tuning amount is calculated by multiplying the impedance change rate by the compensation coefficient. The compensation coefficient ranges from 0.1 to 0.3. The adaptive adjustment rule of the steady-state recovery criterion is to relax the voltage fluctuation threshold to a specific proportion of the rated value during the load low period, for example, to 1.5%.

[0111] Key parameter settings are based on the following: the control signal frame format complies with industrial equipment communication specifications; the ramp control change rate is set based on the thermal characteristics of the equipment; the permission lock effective time is synchronized with the system scan cycle; the current calibration threshold is twice the equipment accuracy level; the steady-state judgment time covers typical transient processes; error checking uses a cyclic redundancy check algorithm; and the retransmission mechanism parameters match the communication network delay characteristics.

[0112] The above technical solution realizes the coordinated control of instruction issuance and permission locking: instruction transmission adopts standardized data format to ensure compatibility; permission locking realizes atomic control through register operation; current maintenance combines closed-loop control and dynamic compensation; steady-state recovery criterion integrates electrical quantity and topological state.

[0113] Specific implementation example: During the instruction sending phase, the compensation power value of 30,000 watts allocated to a certain energy storage unit is extracted and encapsulated into a control signal frame (specific address code, power value field 30,000); the permission locking instruction generates a specific operation code, and the locking effective time is set to 10 milliseconds after the instruction is issued; the current regulator stores the current current value of 152 amperes and modifies a specific bit of the control register to a locked state; during the maintenance phase, the actual current is compared with 152 amperes at a frequency of 100 times per second, and the reference value is updated when the deviation exceeds 3 amperes; in the steady-state recovery judgment, an unlocking instruction is sent when the DC bus voltage is within the range of 750±7.5 volts for 60 seconds and no propagation risk is predicted.

[0114] This embodiment solves the problem of coordinated control of cross-regional dynamic compensation and local authority locking through multi-step coordination. Unlike traditional single-region regulation or global unified scheduling, the voltage collapse propagation path prediction established in step S2 provides a dynamic boundary basis for identifying active migration sources in step S3, enabling cluster division in step S4 to incorporate both spatial gradient and electrical distance constraints. Step S5 incorporates real-time bus voltage parameters when converting the cluster current increment upper limit into power demand, and allocates energy storage compensation power based on electrical distance priority based on the propagation path topology, forming a power compensation flow inversely proportional to the direction of the collapse propagation. Step S6 innovatively synchronizes power command issuance with load authority locking: Firstly, precise cross-regional power injection is achieved through industrial communication protocols, while secondly, hardware-level authority locking blocks negative feedback regulation from local loads. This spatiotemporal coordination effectively curbs the chain propagation of voltage collapse. In particular, the dynamic calibration mechanism for locking the reference value and the dual criteria for steady-state recovery (combining voltage stability indicators and path prediction status) address the over- and under-compensation issues inherent in traditional solutions, often caused by fixed thresholds.

[0115] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to actual conditions.

[0116] It should be noted that the present invention can be deployed on the device itself to implement embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0117] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission. Wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission methods include infrared, microwave, etc. The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0118] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0119] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0120] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0121] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0122] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0123] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0124] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A control method for improving the utilization rate of renewable energy in real time, characterized in that: include: S1, real-time monitoring of photovoltaic output data, DC bus voltage data and constant power load current data in each area of ​​the park; S2. When it is detected that the voltage data of the local DC bus drops beyond a preset threshold and is accompanied by a sudden drop in the photovoltaic output data, the voltage collapse propagation path is predicted; S3. Compare the timing difference between the voltage drop start time and the current data mutation time of the constant power load on the voltage collapse propagation path. If the timing difference is less than the critical delay, determine that the corresponding constant power load is an active migration source. S4, identifying the active migration source cluster on the voltage collapse propagation path, and calculating the upper limit of the current increment required to maintain the rated power based on the constant power load current data; S5. Generate a cross-region active compensation power instruction based on the current increment upper limit and the real-time adjustable power of the energy storage unit; S6. Send cross-region active compensation power instructions to the energy storage units on the voltage collapse propagation path, and at the same time lock the current regulation authority of the active migration source load.

2. A control method for improving the utilization rate of renewable energy in real time according to claim 1, characterized in that: Real-time monitoring of photovoltaic output data, DC bus voltage data, and constant power load current data in each area of ​​the park, including: Divide the park DC network into multiple monitoring areas based on electrical distance; A voltage monitoring point is set at the DC bus in each monitoring area, an output monitoring point is set at the photovoltaic grid connection point, and a current monitoring point is set at the constant power load access point; Using synchronous measurement technology based on the precise time protocol, the photovoltaic output data, DC bus voltage data and constant power load current data of each monitoring area are synchronously recorded with microsecond timestamps.

3. The control method for improving the utilization rate of renewable energy in real time according to claim 2, characterized in that: When it is detected that the voltage data of the local DC bus exceeds the preset threshold and is accompanied by a sudden drop in PV output data, the voltage collapse propagation path is predicted, including: When the DC bus voltage data of any monitoring area drops significantly within a preset time window and the PV output data of the same monitoring area drops sharply within the same time window, voltage collapse propagation path prediction is performed; Based on the voltage data of the DC bus in each monitoring area, the voltage sag depth value is calculated. The voltage sag depth value is the ratio of the difference between the rated voltage and the actual voltage of the DC bus to the rated voltage; Arrange the spatial dimensions from near to far according to the electrical distance of the monitoring area, and arrange the time dimension according to the time series to construct the spatiotemporal distribution tensor of the voltage sag depth; A three-dimensional convolution kernel is used to perform convolution operation on the spatiotemporal distribution tensor to extract the gradient field characteristics of voltage sag propagation; Generate a dynamic weight distribution matrix based on the electrical admittance matrix elements between monitoring areas to weight and enhance the spatial components of the gradient field characteristics; According to the spatial gradient direction of the weighted gradient field characteristics, the voltage collapse propagation path prediction result is output, which spreads from the monitoring area with the deepest voltage sag to the monitoring area with adjacent electrical distance.

4. The control method for improving the utilization rate of renewable energy in real time according to claim 3, characterized in that: The weight of the three-dimensional convolution kernel in the spatial dimension is negatively correlated with the electrical distance between the monitoring areas, and the weight in the time dimension decays as the time interval increases.

5. The control method for improving the utilization rate of renewable energy in real time according to claim 3 is characterized in that: Compare the timing difference between the voltage sag start time and the current data mutation time of the constant power load on the voltage collapse propagation path. If the timing difference is less than the critical delay, the corresponding constant power load is determined to be an active migration source, including: Extract the voltage drop starting time of the monitoring area with the deepest voltage sag from the voltage collapse propagation path prediction results; Obtain constant power load current data for each monitoring area along the voltage collapse propagation path, and use a mutation point detection method based on phase-locked loop technology to identify the moment of current mutation; Calculate the absolute time difference between the voltage drop start time and the current sudden change time in the same monitoring area as the timing difference; Compare the timing difference with the critical delay, which is set based on the minimum opening time of the DC circuit breaker; When the time sequence difference is less than the critical delay and the current change direction is increasing, the constant power load in the corresponding monitoring area is marked as an active migration source.

6. The control method for improving the utilization rate of renewable energy in real time according to claim 5, characterized in that: Identify active migration source clusters on the voltage collapse propagation path and calculate the upper limit of the current increment required to maintain the rated power based on the constant power load current data, including: According to the spatial gradient direction of the voltage collapse propagation path prediction results, the monitoring areas with adjacent electrical distances and marked as active migration sources are merged into active migration source clusters; Obtain current data for each constant-power load in the active migration source cluster and calculate the average steady-state current before the fault occurs. Based on the current minimum DC bus voltage and load rated power, calculate the minimum current increment required to maintain the rated power; The minimum current increments of all constant power loads in the active migration source cluster are accumulated to serve as the upper limit of the cluster current increment.

7. The control method for improving the utilization rate of renewable energy in real time according to claim 6, characterized in that: Generate cross-region active compensation power instructions based on the current increment upper limit and the real-time adjustable power of the energy storage unit, including: Multiply the current increment upper limit of the active migration source cluster by the current minimum DC bus voltage value to convert it into the power demand upper limit of the active migration source cluster; Obtain the real-time adjustable power of each energy storage unit on the voltage collapse propagation path; Compensation power is allocated to each energy storage unit in the order of the electrical distance between the monitoring area where each energy storage unit is located and the active migration source cluster, from closest to farthest. The allocated amount does not exceed the real-time adjustable power of the energy storage unit, and the cumulative allocated amount does not exceed the power demand limit. Generate a cross-region active compensation power instruction including the energy storage unit identification and the allocated compensation power value.

8. The control method for improving the utilization rate of renewable energy in real time according to claim 7, characterized in that: The real-time adjustable power is the difference between the maximum allowable discharge power of the energy storage unit in the current operating state and the current actual output power.

9. The control method for improving the utilization rate of renewable energy in real time according to claim 7, characterized in that: Issue cross-region active compensation power instructions to energy storage units on the voltage collapse propagation path, and lock the current regulation authority of the active migration source load, including: Sending the allocated compensation power value contained in the cross-region active compensation power instruction to the corresponding energy storage unit power controller on the voltage collapse propagation path, so that the energy storage unit adjusts the output power according to the allocated compensation power value; Sending a permission lock command to the current regulator of the active migration source load to prohibit the current regulator from automatically adjusting the load current based on the DC bus voltage change; During the period when the authority is locked, the current value of the active migration source load is maintained at the current value at the locking moment until the system returns to steady-state operation.

Images