A method and system for suppressing response rebound effect of a heterogeneous flexible load group

Abstract

This invention discloses a method and system for suppressing the rebound effect of heterogeneous flexible load groups, belonging to the field of load dispatching and control in AC distribution networks. It includes: collecting real-time operating parameters of heterogeneous flexible loads to generate a load group phase state sequence; calculating global phase synchronization and generating a phase dispersion critical coefficient by combining the grid net load ramp rate; dividing phase sub-intervals to generate a power reduction depth instruction set; performing integral operations on the state deviation data during the reduction period to generate a demand deficit vector; generating a differentiated recovery allowable delay time sequence based on the product of the deficit distribution percentile and the ramp rate; predicting local power rebound; and performing topology isolation operations by disconnecting the upstream grid-connected switch and closing the local energy storage switch for the over-limit target feeder node. This invention employs a physical phase discretization and spatial topology recoupling mechanism, which can effectively disperse load synchronization restoration peaks and absorb rebound power locally, improving the transient stability of the distribution network.

Description

Technical Field

[0001] This invention relates to the field of load dispatching and control in AC distribution networks, and in particular to a method and system for suppressing the response rebound effect of heterogeneous flexible load groups. Background Technology

[0002] With the development of new power systems, demand response technology has been widely applied to load dispatching and control in AC distribution networks. After a demand response event ends, a large number of heterogeneous flexible loads need to be reconnected to the grid to resume operation. During this recovery process, the concentrated connection of a large number of loads can easily trigger a short-term surge in grid load power, i.e., a "load rebound effect," which causes a huge power flow impact on the distribution network and seriously threatens the transient stability and safe operation of the power system.

[0003] In related technologies, Chinese invention patent application CN121172772A discloses a method and system for evaluating the priority of orderly power load recovery to reduce the load rebound effect. This method calculates the load recovery priority parameters by monitoring the power grid in real time and collecting power load parameters and status parameters; introduces a dynamic penalty mechanism to detect power load violations and determine penalty items; dynamically allocates optimization weights according to the current power grid operating scenario type; and constructs a priority model by combining the priority parameters, optimization weights, and penalty items to finally obtain the power load recovery priority estimate, thereby measuring the priority of orderly power load recovery.

[0004] Regarding the aforementioned technologies, this solution mainly relies on multi-dimensional index scoring to determine the queuing order for load restoration. When facing the centralized restoration of massive heterogeneous flexible loads, this scoring-based queuing scheduling method cannot strictly ensure that the transient power during the restoration process matches the real-time ramp-up acceptance boundary of the power grid. At the same time, in extreme cases where the rebound power still exceeds the limit, it also lacks underlying physical and electrical protection and isolation measures, resulting in poor accuracy and reliability in suppressing the load rebound effect, making it difficult to effectively guarantee the safe and stable operation of the distribution network under the access of a large proportion of flexible loads. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a method and system for suppressing the rebound effect of heterogeneous flexible load groups. By employing a physical phase discretization and spatial topology recoupling mechanism, it can effectively disperse load synchronous power restoration peaks and absorb rebound power locally, thereby improving the transient stability of the distribution network.

[0006] The above objectives can be achieved through the following approach: A method for suppressing the response rebound effect of heterogeneous flexible load groups includes: collecting real-time operating parameters of the heterogeneous flexible loads; extracting temperature deviation characteristics and state-of-charge deviation characteristics; calculating the normalized operating phase value of each flexible load within its working cycle; and concatenating the normalized operating phase values ​​to generate a load group phase state sequence. Based on the load group phase state sequence, a global phase synchronization coefficient is calculated; the current grid net load ramp rate is collected; an algebraic mapping from the current grid net load ramp rate to the constraint boundary is performed to generate a phase dispersion critical coefficient. Based on the load group phase state sequence, the heterogeneous flexible loads are divided into phase sub-intervals; and based on the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, the power reduction depth allocation weight of each phase sub-interval is calculated to generate a power reduction... The system receives the power reduction deep instruction set, collects real-time state deviation data of each flexible load during the power reduction period, performs hourly time integration on the real-time state deviation data, and generates a demand deficit vector for each load. It extracts the demand deficit vector, calculates the percentile characteristic value of each flexible load in the demand deficit vector, calculates the product parameter of the percentile characteristic value and the current grid net load ramp rate, and generates a recovery permit delay duration sequence. Based on the recovery permit delay duration sequence, it calculates the predicted local power rebound value of each feeder node, extracts the target feeder node whose predicted local power rebound value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnects the upstream grid connection switch of the target feeder node, and closes the local energy storage connection switch.

[0007] Optionally, the step of splicing the normalized operating phase values ​​to generate the load group phase state sequence includes: parsing the real-time operating parameters, extracting the ambient temperature of the temperature control equipment and the set control dead zone to generate temperature deviation features, extracting the current remaining power of the energy storage equipment and the rated charge / discharge capacity to generate state of charge deviation features; calculating an algebraic ratio based on the temperature deviation features and the state of charge deviation features, linearly mapping the algebraic ratio to a standard circumferential phase angle interval to generate normalized operating phase values, and performing one-dimensional vector splicing according to the equipment physical address sequence to generate the load group phase state sequence.

[0008] Optionally, generating the phase dispersion critical coefficient includes: extracting the normalized operating phase value of the load group phase state sequence, calculating the Euler complex exponent of the normalized operating phase value, performing an arithmetic mean summation absolute modulus operation on the Euler complex exponent to generate a global phase synchronization coefficient; calculating the mathematical ratio parameter between the current grid net load ramp rate and the maximum allowable current conversion threshold of the distribution area, and performing scalar mapping calculation using the mathematical ratio parameter to generate the phase dispersion critical coefficient.

[0009] Optionally, the step of generating the reduction depth instruction set includes: extracting the maximum normalized operating phase value and the minimum normalized operating phase value of the load group phase state sequence, calculating the numerical range, performing equal-interval segmentation on the numerical range to construct phase sub-interval boundaries, mapping the heterogeneous flexible load to the phase sub-intervals; calculating the algebraic difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, using the algebraic difference to divide the basic allocation weights, statistically analyzing the device quantity density of the phase sub-intervals, performing scalar scaling calculation on the basic allocation weights using the device quantity density, generating power reduction depth allocation weights, and converting them into a reduction depth instruction set.

[0010] Optionally, generating the demand deficit vector for each load includes: parsing the reduction depth instruction set to extract the target reference baseline parameter, collecting the current physical feedback parameter of each flexible load during the power reduction period, calculating the algebraic subtraction difference between the target reference baseline parameter and the current physical feedback parameter, and constructing real-time state deviation data; extracting the discrete time step parameter of the control cycle, calculating the scalar product of the real-time state deviation data and the discrete time step parameter, performing a traversal accumulation and summation operation on the scalar product within the power reduction cycle, and concatenating the accumulation and summation results according to the load node index to generate the demand deficit vector for each load.

[0011] Optionally, the method further includes: extracting the individual deficit power value of each flexible load in the demand deficit vector, extracting the allowable delay seconds of each flexible load in the recovery allowable delay duration sequence, calculating the physical quotient of the individual deficit power value and the allowable delay seconds, and generating the cold start transient impact power equivalent.

[0012] Optionally, generating the recovery permit delay duration sequence includes: performing an ascending sort operation on the values ​​of the demand deficit vector to construct an ordered sequence; extracting the position index of each flexible load in the ordered sequence; calculating the algebraic ratio of the position index to the total number of loads to generate percentile feature values; calculating the scalar product of the percentile feature values ​​and the current grid net load ramp rate to generate product parameters; collecting the basic recovery time span value of the grid distribution area; calculating the physical product result of the product parameters and the basic recovery time span value; and recombining the values ​​according to the equipment identification code to generate the recovery permit delay duration sequence.

[0013] Optionally, disconnecting the upstream grid-connected switch of the target feeder node and closing the local energy storage connection switch includes: obtaining the physical access address of each flexible load; dividing the cold start transient impact power equivalent and the recovery permission delay time sequence to feeder nodes according to the physical access address; calculating the cold start transient impact power equivalent under the same feeder node and performing scalar accumulation operation to generate a local power rebound prediction value; calculating the algebraic subtraction difference between the local power rebound prediction value and the equivalent power threshold of the current grid net load ramp rate; extracting the target feeder nodes where the algebraic subtraction difference is greater than zero; generating an electrical isolation trigger pulse signal; and using the electrical isolation trigger pulse signal to disconnect the upstream grid-connected switch of the target feeder node and close the local energy storage connection switch.

[0014] Based on the same inventive concept, this invention also provides a system for suppressing the response rebound effect of heterogeneous flexible load groups. The system includes: a phase state sequence generation module, used to collect real-time operating parameters of the heterogeneous flexible loads, extract temperature deviation characteristics and state-of-charge deviation characteristics, calculate the normalized operating phase value of each flexible load within its working cycle, and concatenate the normalized operating phase values ​​to generate a load group phase state sequence; a phase dispersion critical coefficient generation module, used to calculate a global phase synchronization coefficient based on the load group phase state sequence, collect the current grid net load ramp rate, perform an algebraic mapping from the current grid net load ramp rate to the constraint boundary, and generate a phase dispersion critical coefficient; and a reduction depth instruction set generation module, used to divide the heterogeneous flexible loads into phase sub-intervals based on the load group phase state sequence, and calculate the power of each phase sub-interval based on the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient. The system employs a power reduction depth allocation weighting method to generate a power reduction depth instruction set. A deficit vector generation module receives the power reduction depth instruction set, collects real-time state deviation data of each flexible load during power reduction, performs hourly time integration on the real-time state deviation data, and generates a demand deficit vector for each load. A delay duration sequence generation module extracts the demand deficit vector, calculates the percentile characteristic value of each flexible load in the demand deficit vector, calculates the product parameter of the percentile characteristic value and the current grid net load ramp rate, and generates a recovery permit delay duration sequence. A rebound prediction and topology isolation module calculates the local power rebound prediction value of each feeder node based on the recovery permit delay duration sequence, extracts target feeder nodes whose local power rebound prediction value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnects the upstream grid-connected switch of the target feeder node, and closes the local energy storage connection switch.

[0015] Compared with the prior art, the present invention has the following advantages: 1. A nonlinear phase discretization mechanism is introduced. By extracting the normalized operating phase of various heterogeneous loads and setting a dispersion critical coefficient in conjunction with the real-time grid ramp rate, power reduction is dynamically guided. This breaks the high synchronicity of load states during demand response, achieving uniform dispersion of physical states during the reduction phase, eliminating the energy storage conditions that could trigger large-scale centralized power restoration, and improving the smoothness of the demand response process. 2. The traditional unified power restoration broadcast mechanism has been abandoned, and a purely data-driven causal blocking strategy has been adopted. By extracting percentile features from the load integral demand deficit and establishing a product mapping with the grid ramp rate, a differentiated recovery delay time is generated for each load. This transforms the originally concentrated "instantaneous bursts" into gradual "tiered recovery," cutting off the unified triggering chain that leads to synchronous rebound, and ensuring that the power changes throughout the entire power restoration process are strictly enveloped within the grid's safe ramp rate boundary. 3. Based on the advanced perception of local power rebound prediction values, when the predicted local rebound power exceeds the limit, an electrical isolation pulse can be actively generated to disconnect the upstream switch on the target feeder and seamlessly connect to local energy storage. Through spatial electrical decoupling and on-site energy absorption, a hardware firewall is built for the large power grid to protect it from extreme impacts, ensuring the transient safety of the distribution network under a high proportion of flexible load access.

[0016] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0017] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for suppressing the rebound effect of heterogeneous flexible load group response according to an embodiment of the present invention. Figure 2 This is a mapping diagram of the percentage eigenvalues ​​of demand deficit and the recovery permit delay time under different power grid ramping scenarios in this embodiment of the invention; Figure 3 This is a comparison diagram of the predicted local power bounce value and the equivalent power threshold of each feeder node in an embodiment of the present invention, and a relationship diagram for exceeding the limit. Figure 4This is a schematic diagram of a heterogeneous flexible load group response rebound effect suppression system according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0020] Reference Figure 1 One embodiment of the present invention proposes a method for suppressing the rebound effect of heterogeneous flexible load group response. It adopts a physical phase discretization and spatial topology recoupling mechanism, which can effectively disperse the load synchronous power restoration peak and absorb the rebound power locally, thereby improving the transient stability of the distribution network.

[0021] The method described in this embodiment specifically includes: Real-time operating parameters of heterogeneous flexible loads are collected, temperature deviation characteristics and state of charge deviation characteristics are extracted, normalized operating phase values ​​of each flexible load in the working cycle are calculated, and the normalized operating phase values ​​are spliced ​​together to generate a load group phase state sequence. Optionally, the step of splicing the normalized operating phase values ​​to generate the load group phase state sequence includes: The real-time operating parameters are analyzed to extract the temperature deviation characteristics between the ambient temperature of the temperature control device and the set control dead zone, and the current remaining power of the energy storage device and the rated charge / discharge capacity are extracted to generate the state of charge deviation characteristics. The load data analysis and calculation unit acquires real-time operating parameters of various heterogeneous flexible loads through an advanced measurement system. Since the physical operating mechanisms of temperature control equipment and energy storage equipment are fundamentally different, mathematical transformations are necessary to eliminate dimensional differences and unify them into dimensionless deviation characteristics. For flexible loads without temperature regulation attributes, their temperature deviation characteristics are set to zero; for flexible loads without energy storage charging attributes, their state-of-charge deviation characteristics are set to zero. A two-dimensional deviation feature vector is constructed based on the temperature deviation characteristics and the state-of-charge deviation characteristics. For temperature control equipment, the load data analysis and calculation unit reads the ambient temperature from the equipment's built-in thermistor, the target set temperature from the control memory of the user interface panel, and the set control dead zone parameters fixed to prevent frequent compressor start-stops from the equipment's factory nameplate database. The temperature deviation characteristic refers to the normalization degree of the current temperature's deviation from the set target, and the calculation relationship of the temperature deviation characteristic satisfies the following formula: , Among them, letters Represents temperature deviation characteristics; letters Represents ambient temperature; letters Represents the target set temperature; letters This represents the set control dead zone. Since the numerator represents the actual temperature difference and the denominator represents the set tolerance temperature difference, both are in degrees Celsius. Dividing them yields... This is a dimensionless parameter. For energy storage devices, the load data parsing and calculation unit reads the current remaining charge from the battery management system and obtains the rated charge / discharge capacity from the device's manufacturer's manual. To measure the optimal operating point for healthy battery charging and discharging, half of the rated charge / discharge capacity is defined as the physical intermediate equilibrium state. The state-of-charge (SOC) deviation characteristic refers to the relative distance of the current charge from this equilibrium state. The calculation relationship of the SOC deviation characteristic satisfies the following formula: , Among them, letters Represents the characteristics of the state of charge deviation; letters Represents the current remaining battery power; letters Represents rated charge / discharge capacity; digital This represents the battery's half-load midpoint constant. Both the numerator and denominator are dimensionless (kilowatt-hours), and the result after division is... Also being dimensionless parameters, this ensures dimensional consistency in spatial calculations.

[0022] Based on the temperature deviation characteristics and the state of charge deviation characteristics, the algebraic ratio is calculated, and the algebraic ratio is linearly mapped to the standard circumferential phase angle interval to generate a normalized operating phase value. A one-dimensional vector is then concatenated according to the equipment physical address sequence to generate a load group phase state sequence.

[0023] The phase sequence construction unit extracts dimensionless deviation data from each device and transforms it to a polar coordinate system to characterize the load's response potential distribution. The unit introduces a bivariate arctangent function to calculate the algebraic ratio of temperature deviation characteristics to state-of-charge deviation characteristics, extracting the native angle of this state point in a two-dimensional Cartesian coordinate system. To facilitate synchronization calculation, this is linearly mapped to a standard circular phase angle interval between zero and one, generating normalized operating phase values. The calculation relationship of the normalized operating phase values ​​satisfies the following formula: , Among them, letters Represents the normalized running phase value; letter Represents the characteristics of the state of charge deviation; letters Represents temperature deviation characteristics; function This represents a bivariate arctangent function capable of identifying orientation in four quadrants; its output value is the algebraic ratio of negative to positive pi in native radians. This represents the constant pi. The original radians, divided by twice pi, are compressed to the range of -0.5 to +0.5, and then added to the constant 0.5 to achieve a positive linear shift towards the standard circumferential phase angle range. After calculation, the phase sequence construction unit reads the medium access control addresses assigned to each heterogeneous flexible load at the distribution network communication layer and uses them as the device physical addresses. The bubble sort algorithm is used to strictly arrange all participating device physical addresses in ascending order, forming a device physical address sequence. According to this sorting result, the normalized operating phase values ​​corresponding to each device are sequentially concatenated along the column vector dimension to finally generate a load group phase state sequence containing global state information.

[0024] For example, the phase sequence construction unit obtains the deviation data of the aforementioned commercial air conditioner and energy storage battery cabinet. Upon reading, the physical address of the commercial air conditioner ends in 01, and the physical address of the energy storage battery cabinet ends in 02. For the commercial air conditioner ending in 01, its temperature deviation characteristic is 0.5, and its state of charge deviation characteristic is 0. Substituting 0 and 0.5 into the normalized operating phase value calculation formula, the bivariate arctangent function calculates the original radian result when the bivariate values ​​are 0 and 0.5, which is 0. Dividing the original radian value of 0 by twice pi and adding the constant 0.5, the normalized operating phase value of the commercial air conditioner is mathematically calculated to be 0.5. For the energy storage battery cabinet ending in 02, its temperature deviation characteristic is 0, and its state of charge deviation characteristic is 0.3. Substituting 0.3 and 0 into the same formula, the bivariate arctangent function calculates the result as half the radian of pi. Dividing half of pi in radians by twice pi yields 0.25. Adding a constant of 0.5, the normalized operating phase value of the energy storage battery cabinet is mathematically calculated to be 0.75. Subsequently, the phase sequence construction unit compares the last digit of the physical address to determine that 01 precedes 02, thus establishing the device's physical address sequence. Following this order, a one-dimensional vector concatenation operation is performed on the values ​​0.5 and 0.75 to generate the load group phase state sequence at that moment, mathematically expressed as a two-dimensional column vector. The above calculations eliminate the physical barrier between thermodynamics and electrochemistry, providing a basic digital array for calculating the consistency and dispersion of the global response.

[0025] Based on the load group phase state sequence, calculate the global phase synchronization coefficient, collect the current power grid net load ramp rate, perform the algebraic mapping from the current power grid net load ramp rate to the constraint boundary, and generate the phase dispersion critical coefficient. Optionally, the generation of the phase dispersion critical coefficient includes: Extract the normalized operating phase value of the load group phase state sequence, calculate the Euler complex exponent of the normalized operating phase value, perform arithmetic mean summation absolute modulus operation on the Euler complex exponent, and generate the global phase synchronization coefficient. The synchronization degree analytical calculation unit extracts the normalized operating phase value corresponding to each heterogeneous flexible load from the load group phase state sequence. Since the normalized operating phase value is one-dimensional, dimensionless linear data and cannot directly characterize the phase cancellation effect of periodic oscillations, the synchronization degree analytical calculation unit uses Euler's formula to wrap and map the one-dimensional normalized operating phase value onto a unit circle on the two-dimensional complex plane, transforming it into an Euler complex exponent with amplitude and direction vector attributes. The calculation relationship of the Euler complex exponent satisfies the following formula: , Among them, letters The Euler complex exponent represents the k-th heterogeneous flexible load; the letter represents the alphanumeric exponent. Represents the natural constant; letters Represents the imaginary unit; letters Represents the constant pi; the letter This represents the k-th normalized running phase value extracted from the load cluster phase state sequence. Subsequently, the synchronization degree analytical calculation unit performs an arithmetic mean summation of the absolute modulus of all generated Euler complex exponents. The physical mechanism of this operation originates from the Kulamoto oscillator model in nonlinear dynamics, which first calculates the arithmetic mean vector of all complex vectors in the complex plane, and then extracts the absolute geometric length of this mean vector, using it as an indicator to quantify the degree of concentration of the load cluster state. The calculation relationship of the global phase synchronization degree coefficient satisfies the following formula: , Among them, letters Represents the global phase synchronization coefficient; letters Represents the total number of heterogeneous flexible loads, and its value is directly derived from the vector length parameter of the load group phase state sequence; symbol This represents performing a step-by-step addition operation on the variables within the set; symbol This represents the extraction of the absolute modulus of the complex number result. The imaginary dimension is eliminated through modulo operation, and the output global phase synchronization coefficient is a dimensionless real number between zero and one. When the global phase synchronization coefficient approaches one, it indicates that the physical states of the entire network load are highly similar, making it extremely easy to trigger resonance rebound during power restoration; when it approaches zero, it indicates that the phases of each load are evenly distributed, and the risk of rebound is low.

[0026] For example, the synchronization degree analysis and calculation unit obtains the load group phase state sequence containing two flexible loads output from the previous processing stage. It extracts the normalized operating phase value of the first load as 0.5 and the normalized operating phase value of the second load as 0.75. Substituting the value 0.5 into the Euler complex exponent calculation formula, it calculates twice the constant of pi multiplied by 0.5 to obtain one radian of pi. The corresponding Euler complex exponent expands in the complex plane as a real part of -1 and an imaginary part of 0, i.e., -1. Substituting the value 0.75 into the formula, it calculates twice the constant of pi multiplied by 0.75 to obtain 1.5 radians of pi. The corresponding Euler complex exponent expands as a real part of 0 and an imaginary part of -1, i.e., a negative imaginary unit. Subsequently, the synchronization degree analysis and calculation unit substitutes these two Euler complex exponents into the global phase synchronization degree coefficient formula to perform an arithmetic mean summation of the absolute modulus. First, a step-by-step summation is performed to obtain a vector sum of -1 for the real part and -1 for the imaginary part. Dividing this by the total quantity of 2 yields an arithmetic mean vector of -0.5 for the real part and -0.5 for the imaginary part. Finally, the absolute magnitude of this mean vector is extracted, and the sum of the squares of the real and imaginary parts (0.25) is calculated, resulting in 0.5. Taking the square root of 0.5, the global phase synchronization coefficient is mathematically calculated to be a dimensionless value of 0.707. This value accurately quantifies the current degree of aggregation between the two loads, providing a physical basis for the reduction depth allocation.

[0027] Calculate the mathematical ratio parameter between the current net load ramp rate of the power grid and the maximum allowable current conversion threshold of the distribution area, and use the mathematical ratio parameter to perform scalar mapping calculation to generate the phase dispersion critical coefficient.

[0028] The critical state assessment unit acquires the current net load ramp rate of the power grid in real time through measurement terminals deployed on the low-voltage side of the distribution transformer, and reads the maximum allowable current conversion threshold of the current distribution area from the static equipment ledger of the distribution management system. The current net load ramp rate refers to the actual rate of change of power on the pure load side of the power grid over time after eliminating distributed source fluctuations; the maximum allowable current conversion threshold of the distribution area refers to the maximum power change rate that the distribution transformer windings and feeder cables can withstand under the physical boundary of not triggering thermal overload protection and voltage drop alarms. This threshold is calibrated through short-circuit impedance and thermal steady-state experiments during equipment grid connection testing and is fixed in the ledger. The critical state assessment unit calculates the quotient of the current net load ramp rate of the power grid and the maximum allowable current conversion threshold of the distribution area, generating a mathematical ratio parameter. The mathematical ratio parameter characterizes the percentage of the dynamic impact pressure currently borne by the power grid relative to its physical limit. To achieve a smooth transition from the physical pressure of the power grid to the tightness of load control, the critical state assessment unit uses the mathematical ratio parameter to perform scalar mapping calculations based on a negative exponential decay function to generate a phase dispersion critical coefficient. The relationship calculated by this scalar mapping satisfies the following formula: , Among them, letters Represents the critical coefficient for phase dispersion; the letter represents the critical coefficient for phase dispersion. Represents the natural constant; letters Represents the mapping sharpness adjustment coefficient; letters This represents a mathematical ratio parameter. The mapping sharpness adjustment coefficient is not a subjective assumption, but an empirical constant determined by nonlinear curve fitting of two years' historical peak load frequency drop waveform data and ramp-up events in the target distribution network. This coefficient is fixed at 4.5 to ensure that when the grid pressure exceeds 40%, the control response can converge exponentially and rapidly to the high-pressure state. Since the mathematical ratio parameter is the division of power rates with the same dimensions, and the mapping sharpness adjustment coefficient is a constant, the generated phase dispersion critical coefficient is a dimensionless value between zero and one.

[0029] For example, the critical state assessment unit obtains the current grid net load ramp rate of 2.0 MW per minute from the distribution area's measurement terminal. Simultaneously, it retrieves the maximum allowable current conversion threshold for this distribution area as 5.0 MW per minute from the ledger database. The critical state assessment unit divides 2.0 MW per minute by 5.0 MW per minute, canceling out the dimensions, and calculates a dimensionless mathematical ratio parameter of 0.4. Subsequently, the critical state assessment unit substitutes the mathematical ratio parameter 0.4 and the fixed mapping sharpness adjustment coefficient 4.5 into the scalar mapping calculation formula. After calculation, 4.5 multiplied by 0.4 yields a product constant of 1.8. Further calculation of the natural constant to the power of -1.8 yields an exponent term value of approximately 0.165. Finally, subtracting 0.165 from the constant 1 mathematically calculates the phase dispersion critical coefficient to be 0.835. This result indicates that, given the current grid ramping pressure reaching 40% of the physical limit, a distributed critical threshold of 0.835 must be established for numerical comparison with the global phase synchronization coefficient in the next stage.

[0030] Based on the load group phase state sequence, the heterogeneous flexible load is divided into phase sub-intervals. According to the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, the power reduction depth allocation weight of the phase sub-interval is calculated, and a reduction depth instruction set is generated. Optionally, the set of instructions for generating the reduction depth includes: Extract the maximum normalized operating phase value and the minimum normalized operating phase value of the load group phase state sequence, calculate the numerical range, perform equal interval division on the numerical range to construct the phase sub-interval boundary, and map the heterogeneous flexible load to the phase sub-interval; The interval segmentation calculation unit reads the load group phase state sequence and iterates through all values ​​in the sequence. Since the load group phase state sequence is a one-dimensional numerical array, the interval segmentation calculation unit directly selects the maximum and minimum values ​​in the array using a numerical comparison algorithm, defining them as the maximum normalized operating phase value and the minimum normalized operating phase value, respectively. To determine the global span of the load phase, the interval segmentation calculation unit performs a subtraction operation on these two extreme values ​​to generate the numerical range. The calculation relationship of the numerical range satisfies the following formula: , Among them, letters Represents the range of values; letters Represents the maximum normalized running phase value; letter This represents the minimum normalized operating phase value. Since the phase values ​​involved in the calculation are all within the dimensionless standard circular phase angle range of zero to one, the generated numerical range is also a dimensionless decimal. After obtaining the numerical range, the interval segmentation calculation unit needs to construct discrete grouping containers. To ensure the real-time operation of the control algorithm in the underlying microprocessor and the fixed memory allocation, a fixed number of segments with equal intervals is adopted. The segmentation number parameter is not set arbitrarily, but is directly read from the fixed thread number constant in the memory configuration table of the power distribution automation control terminal. The interval segmentation calculation unit divides the numerical range by the segmentation number parameter to generate the step size, and starting from the minimum normalized operating phase value, it successively accumulates the step size to calculate a series of increasing node coordinates, which are used as the boundaries of the phase sub-intervals. The boundaries of two adjacent phase sub-intervals form a left-closed and right-open numerical interval, which is the phase sub-interval. Finally, the interval segmentation calculation unit extracts the normalized operating phase value of each heterogeneous flexible load, determines the interval range it falls into based on the value, and writes its physical identification code into the index list of the corresponding phase sub-interval, thereby completing the mapping and classification of the device at the logical storage level.

[0031] For example, the interval segmentation calculation unit reads the phase state sequence of a load group containing four heterogeneous flexible loads, with sequence values ​​of 0.1, 0.2, 0.3, and 0.9, respectively. Through traversal comparison, the maximum normalized operating phase value is extracted as 0.9, and the minimum normalized operating phase value is 0.1. Substituting 0.9 and 0.1 into the numerical range calculation formula, subtracting 0.1 from 0.9 yields a numerical range of 0.8. At this point, the interval segmentation calculation unit reads the segmentation quantity constant of 4 from the memory configuration table. Dividing the numerical range of 0.8 by the constant 4, the segmentation step size is calculated to be 0.2. Starting from the minimum value of 0.1, the step size is accumulated, generating five boundary coordinates sequentially: 0.1, 0.3, 0.5, 0.7, and 0.9. This constructs four phase sub-intervals: Interval 1 (0.1 to 0.3), Interval 2 (0.3 to 0.5), Interval 3 (0.5 to 0.7), and Interval 4 (0.7 to 0.9). Based on the equipment values, the first load's 0.1 and the second load's 0.2 are mapped to interval one; the third load's 0.3 is exactly equal to the left boundary of interval two and is mapped to interval two; the fourth load's 0.9 is mapped to interval four. No equipment is mapped in interval three. This operation completes the spatial discretization loading of the physical state of the heterogeneous flexible load.

[0032] Calculate the algebraic difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, use the algebraic difference to divide the basic allocation weights, count the device number density of the phase sub-intervals, use the device number density to perform scalar scaling calculation on the basic allocation weights, generate power reduction depth allocation weights and convert them into reduction depth instruction sets.

[0033] The command distribution control unit receives the mapping list output by the interval segmentation calculation unit and the calculated control boundary parameters. The command distribution control unit first performs an algebraic subtraction operation, subtracting the global phase synchronization coefficient, which represents the load aggregation degree, from the phase dispersion critical coefficient, which characterizes the grid's capacity limit, to obtain the algebraic difference. This algebraic difference represents the overall urgency of the system needing to forcibly disperse load phases; a larger difference indicates a greater need for deeper reduction actions. The command distribution control unit directly extracts the absolute value of this algebraic difference as the global base allocation weight. Subsequently, to achieve targeted differentiated dispersion, the command distribution control unit calculates the device quantity density for each phase sub-interval. The calculation relationship for device quantity density satisfies the following formula: , Among them, letters The number density of devices representing the target phase sub-interval; letters Represents the total number of heterogeneous flexible loads mapped within the target phase sub-interval; the letter This represents the total number of heterogeneous flexible loads participating in the scheduling across the entire network. This division eliminates the units digit of the device count, generating a dimensionless ratio. After obtaining the density, the instruction distribution control unit performs a quadratic nonlinear scalar scaling calculation on the basic allocation weights to generate power reduction depth allocation weights. The physical mechanism of this scaling calculation is: for densely populated areas with highly concentrated devices, a more severe penalty reduction force must be applied to forcibly dispel the synchronization state. The composite relationship between scaling calculation and instruction conversion satisfies the following formula: , , Among them, letters The power reduction depth assignment weight represents the target phase sub-interval; letters Represents the global basic allocation weight; letters The representative issued a reduction depth instruction to a single device within the target phase sub-interval; the letter and The meaning is consistent with the previous formula. By multiplying the basic allocation weight by the device quantity density twice, the total weight allocated to the interval exhibits exponential scaling. Subsequently, the power reduction depth allocation weight of the interval is averaged and divided by the number of devices in the interval to generate a dimensionless reduction ratio instruction for a single device. Finally, the instruction distribution control unit sorts the physical devices according to their communication addresses, assembles the reduction depth instructions of all devices into a one-dimensional column vector, and formally generates and outputs the reduction depth instruction set.

[0034] Receive the reduction depth instruction set, collect real-time state deviation data of each flexible load during the power reduction period, perform time-integration operation on the real-time state deviation data step by step, and generate demand deficit vector of each load; Optionally, the target reference baseline parameter is extracted by parsing the reduction depth instruction set, the current physical feedback parameter of each flexible load is collected during the power reduction period, the algebraic subtraction difference between the target reference baseline parameter and the current physical feedback parameter is calculated, and real-time state deviation data is constructed. The deviation data construction unit receives the output reduction depth instruction set. This instruction set contains reduction instructions corresponding to each device in the entire network in the form of a column vector, and the instruction values ​​are dimensionless decimals. The deviation data construction unit reads the full-load rated operating power of each flexible load from the static ledger of the power distribution management system, multiplies the full-load rated operating power by the corresponding reduction instruction, and calculates the power value that should be reduced. Then, it subtracts the power value that should be reduced from the full-load rated operating power to calculate the target reference baseline parameter. During the power reduction control cycle, the deviation data construction unit polls and collects the current physical feedback parameters of each flexible load at a fixed frequency through the underlying data bus, that is, the real-time active power consumption value. The deviation data construction unit subtracts the current physical feedback parameter from the target reference baseline parameter, performs an algebraic subtraction operation, and calculates the real-time state deviation data of a single device. The calculation relationship of the real-time state deviation data satisfies the following formula: , Among them, letters Represents the real-time state deviation data at the target sampling time; letters Represents the full-load rated operating power read from the static ledger; letters Represents the corresponding reduction instruction; letters This represents the current physical feedback parameters collected at that moment. The units of the full-load rated operating power and the current physical feedback parameters are both kilowatts. The reduction command is a dimensionless decimal, and the result of the calculation within the parentheses is also dimensionless. Multiplying and subtracting ensures strict consistency of the physical units on both sides of the equation, confirming that the unit of measurement for the real-time status deviation data is kilowatts. When the real-time status deviation data is positive, it indicates that the actual operating power of the equipment is lower than the target baseline, and an energy deficit is accumulating.

[0035] For example, the deviation data construction unit processes commercial air conditioning unit number 01. It reads the reduction instruction for this unit as a dimensionless value of 0.1, and its full-load rated operating power is read from the ledger as 200.0 kW. Substituting 0.1 and 200.0 kW into the formula, it calculates that a reduction of 20.0 kW should be made, with a target reference baseline parameter of 180.0 kW. At a certain control moment, the deviation data construction unit reads the current physical feedback parameter of the air conditioner via the bus as 150.0 kW. Substituting this into the real-time state deviation data calculation formula, subtracting 150.0 kW from 180.0 kW, the calculated real-time state deviation data at that moment is positive 30.0 kW. This indicates that in response to grid peak shaving, the air conditioner's actual operating power is 30.0 kW lower than the specified baseline, and it is rapidly accumulating a cooling energy deficit.

[0036] Extract the discrete time step parameter of the control loop, calculate the scalar product of the real-time state deviation data and the discrete time step parameter, perform a traversal accumulation and summation operation on the scalar product within the power reduction cycle, and concatenate the accumulation and summation results according to the load node index to generate the demand deficit vector of each load.

[0037] The integral summation calculation unit obtains the control clock parameters issued by the distribution automation master station and extracts the control cycle time as the discrete time step parameter. The discrete time step parameter is the physical time interval between two adjacent samplings of the underlying data, and its set value comes from the preset frequency division constant of the system crystal oscillator, which is fixed at 2 seconds here. The integral summation calculation unit extracts the real-time state deviation data of a flexible load at the current moment, performs a scalar multiplication operation with the discrete time step parameter, and calculates the accumulated single-step energy deficit value within the discrete micro-element time period. During the entire power reduction cycle, the integral summation calculation unit traverses all single-step energy deficit values ​​in time sequence and performs an arithmetic accumulation and summation operation. This accumulation and summation operation is equivalent to continuous-time integration operation in discrete mathematics, and its purpose is to quantify the accumulated energy deficit value of the equipment during peak shaving events. In order to adapt to the industrial standard unit of electricity metering, the accumulation result needs to be divided by the time conversion constant. The calculation relationship of the individual elements in the demand deficit vector satisfies the following formula: , Among them, letters Represents the total demand deficit of the i-th flexible load; the letter Represents the total number of samples taken during the entire power reduction cycle; symbol This represents iterating and accumulating data across all discrete sampling times; the letter... Represents real-time state deviation data from a single sample; letters Represents the discrete time step parameter; digital This represents a constant used to convert seconds to hours. The dimension is kilowatt. The unit is seconds. Multiplying them gives kilowatt-seconds. Dividing by 3600 seconds per hour, the physical dimension of the total demand deficit on the left side of the equation strictly converges to kilowatt-hours. After obtaining the total demand deficit of all equipment, the integral summation calculation unit performs a one-dimensional vector concatenation operation according to the load node index order to generate the demand deficit vector of each load.

[0038] For example, the integral summation calculation unit processes the commercial air conditioning unit numbered 01 above. The discrete time step parameter is obtained as 2 seconds. Assume that during a power reduction cycle of up to 600 seconds, the real-time state deviation data of this air conditioner remains consistently at 30.0 kW. Substituting 30.0 kW, 2 seconds, and 300 samples into the element calculation formula of the demand deficit vector, the single scalar product is 30.0 kW multiplied by 2 seconds, yielding a single-step energy deficit value of 60.0 kW / s. Performing 300 iterative summation operations yields a total sum of 18000.0 kW / s. Dividing 18000.0 by the constant 3600, the total demand deficit for this air conditioner is calculated to be 5.0 kWh. This means that during peak shaving, the air conditioner consumes 5 kWh less electricity than the target control state. The integral summation calculation unit concatenates the calculated value 5.0 and the calculation results of other devices according to their addresses to generate a demand deficit vector, thus completing a precise digital mapping of the energy shortage state of heterogeneous loads.

[0039] Optionally, the method further includes: Extract the individual power deficit value of each flexible load in the demand deficit vector, extract the allowable delay seconds of each flexible load in the recovery allowable delay duration sequence, calculate the physical quotient of the individual power deficit value and the allowable delay seconds, and generate the cold start transient impact power equivalent.

[0040] The impact prediction calculation unit extracts the demand deficit vector and the recovery permit delay duration sequence, respectively. From the demand deficit vector, it extracts the individual deficit energy value of the target flexible load, representing the total energy the equipment desires to replenish. From the recovery permit delay duration sequence, it extracts the permit delay seconds corresponding to the target flexible load, representing the mandatory power restoration buffer time. The impact prediction calculation unit performs an algebraic division operation to calculate the physical quotient of the individual deficit energy value and the permit delay seconds. The physical meaning of this division operation is to convert static "energy" into dynamic "power," that is, to predict the concentrated power release generated by the equipment during the short window period allowed for power restoration in order to quickly replenish the deficit energy. To ensure dimensional consistency, a time conversion coefficient needs to be introduced before performing the division. The calculation relationship of the cold start transient impact power equivalent satisfies the following formula: , Among them, letters Represents the equivalent cold-start transient impact power of the target equipment; letters Represents the amount of electricity deficit per unit; (digital) A constant representing the number of seconds contained in one hour; letters This represents the permitted delay in seconds. The unit of the single-unit power deficit is in kilowatt-hours (kWh), which, after multiplying by 3600, is converted to kilowatt-seconds. After dividing by the permitted delay in seconds, the cold-start transient impact power equivalent on the left side of the equation strictly reverts to kilowatts. This power equivalent is a predictive scalar quantity specifically used to measure the physical impact intensity at the moment the warning equipment is restored to power.

[0041] For example, the impact prediction calculation unit continues to process commercial air conditioning unit number 01. From the previous steps, the unit's individual power deficit is extracted to be 5.0 kWh, and the permissible delay seconds allocated to it by the system are extracted to be 120 seconds. 5.0 kWh, the constant 3600, and 120 seconds are substituted into the cold start transient impact power equivalent calculation formula. First, 5.0 kWh is multiplied by 3600, converting to 18000.0 kWh / second. Then, a physical quotient calculation is performed, dividing 18000.0 kWh / second by 120 seconds. After calculation, the time dimension of seconds is canceled out, resulting in a value of 150.0. Therefore, the mathematical calculation of the cold start transient impact power equivalent of this air conditioning unit is 150.0 kW. This value clearly warns the system that if power is restored according to the 120-second delay schedule, the air conditioner will generate a transient impact power of up to 150 kW at startup, providing a direct physical basis for over-limit assessment and topology switch action.

[0042] Extract the demand deficit vector, calculate the percentile characteristic value of each flexible load in the demand deficit vector, calculate the product parameter of the percentile characteristic value and the current grid net load ramp rate, and generate a recovery permit delay duration sequence. Optionally, the generation of the recovery license delay duration sequence includes: Perform an ascending sort operation on the values ​​of the demand deficit vector to construct an ordered sequence, extract the position index of each flexible load in the ordered sequence, calculate the algebraic ratio of the position index to the total number of loads, and generate percentile feature values. The sequence sorting calculation unit obtains the demand deficit vector. The values ​​within this vector represent the accumulated power shortage of each heterogeneous flexible load during the reduction period. The sequence sorting calculation unit uses a quicksort algorithm to rearrange all individual power shortage values ​​within the demand deficit vector according to ascending order, generating an ordered sequence. During the sorting process, the sequence sorting calculation unit simultaneously retains the device identification code corresponding to the original value, thus maintaining the mapping relationship between data and physical devices. Subsequently, the sequence sorting calculation unit extracts the specific rank number of each flexible load in the ordered sequence and establishes it as a position index. The position index is an integer increasing from a constant to the total number of loads. The total number of loads is the total number of heterogeneous flexible load devices currently participating in scheduling. The sequence sorting calculation unit performs a division operation between the position index and the total number of loads to calculate the algebraic ratio between the two, defining this algebraic ratio as the percentile eigenvalue. The calculation relationship of the percentile eigenvalue satisfies the following formula: , Among them, letters Represents percentile eigenvalues; letters Represents the position index of the target flexible load in an ordered sequence; letter This represents the total number of loads. Since both the numerator (position index) and the denominator (total load number) are dimensionless integers, the percentile characteristic value generated by dividing them is a dimensionless real number greater than zero and less than or equal to one. The absolute capacities of different flexible loads vary greatly, and directly using electricity values ​​cannot achieve fair scheduling. By converting to percentile characteristic values, the deficit levels of all equipment can be forcibly normalized to the same relative ranking scale. The more severely the equipment is deficited, the closer its characteristic value is to a constant of one.

[0043] For example, the sequence sorting calculation unit obtains a demand deficit vector containing three heterogeneous flexible loads. Based on the equipment identification coding order, the deficit amounts for the three devices are 8.0 kWh, 2.0 kWh, and 5.0 kWh, respectively. The sequence sorting calculation unit performs an ascending sort operation on these three values, constructing an ordered sequence with values ​​of 2.0, 5.0, and 8.0. The position indices of this ordered sequence are extracted: the device with a deficit of 2.0 kWh is in the first position, with a position index of 1; the device with a deficit of 5.0 kWh is in the second position, with a position index of 2; and the device with a deficit of 8.0 kWh is in the third position, with a position index of 3. The current total load quantity is a constant of 3. The position indices and the total load quantity are substituted into the percentile eigenvalue calculation formula. For equipment with a deficit of 2.0 kWh, calculating 1 divided by 3 yields a dimensionless value of 0.333; for equipment with a deficit of 5.0 kWh, calculating 2 divided by 3 yields a dimensionless value of 0.667; and for equipment with a deficit of 8.0 kWh, calculating 3 divided by 3 yields a dimensionless value of 1.0. Through this arithmetic process, the absolute energy deficit of the equipment is perfectly transformed into a percentile eigenvalue representing the ranking weight.

[0044] The product parameter is generated by calculating the scalar product of the percentile characteristic value and the current net load ramp rate of the power grid. The basic recovery time span value of the power grid area is collected. The physical product result of the product parameter and the basic recovery time span value is calculated. The value is recombined according to the equipment identification code to generate a recovery permit delay time sequence.

[0045] The delay duration allocation unit reads the current net load ramp rate of the power grid from the distribution management system. The unit performs a scalar multiplication operation between the previously calculated percentile characteristic value and the current net load ramp rate to generate a product parameter. The physical meaning of this product parameter is to equivalently modulate the overall ramp pressure of the power grid using the load's own deficit urgency. Simultaneously, the delay duration allocation unit reads the basic recovery time span value from the distribution transformer overload tolerance parameter table. This basic recovery time span value refers to the reference time interval during which the transformer is allowed to connect loads in batches under maximum safety redundancy; its value originates from the heat dissipation constant of the equipment's factory setting. To eliminate dimensional conflicts and convert the modulation parameter into a specific time physical quantity, the delay duration allocation unit calculates the physical product of the product parameter and the basic recovery time span value. The calculation relationship for the permissible recovery delay duration satisfies the following formula: , Among them, letters Indicates the duration of the recovery license delay; letters The product parameter is equal to the percentile eigenvalue multiplied by the current net load ramp rate of the power grid; the letter represents the product parameter. Represents the basic recovery time span value; letters This represents the rate normalization constant. The product parameter inherits the dimension of the ramp rate, i.e., kilowatts per minute; the dimension of the base recovery time span is seconds. Direct multiplication would result in kilowatt-seconds per minute (kW / s) of skewed dimensions. Therefore, the formula introduces a rate normalization constant stored at the system's underlying layer to directly offset the rate dimension of the product parameter. After simplification by division, the dimension of the recovery permission delay duration on the left side of the equation strictly converges to seconds. Finally, the delay duration allocation unit, based on the initially recorded device identification code, restores the shuffled recovery permission delay duration to its original physical sequence position on the device, assembling it to generate a recovery permission delay duration sequence for issuing control. Figure 2 As shown, the more severe the energy deficit of the load and the greater the grid ramp-up pressure, the more strictly the mathematically linearly the power restoration delay buffer time is allocated. This data-driven mechanism solves the physical conditions for massive heterogeneous loads to trigger resonance rebound at the same time.

[0046] Based on the recovery permit delay duration sequence, calculate the local power rebound prediction value of each feeder node, extract the target feeder node whose local power rebound prediction value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnect the upstream grid connection switch of the target feeder node and close the local energy storage connection switch.

[0047] Optionally, disconnecting the upstream grid connection switch of the target feeder node and closing the local energy storage connection switch includes: Obtain the physical access address of each flexible load, divide the cold start transient impact power equivalent and the recovery permission delay time sequence to feeder nodes according to the physical access address, calculate the cold start transient impact power equivalent under the same feeder node and perform scalar accumulation operation to generate local power rebound prediction value; The node topology aggregation unit reads the physical access addresses of each flexible load from the static topology database of the power distribution management system. The physical access address is the digital code of the specific feeder branch electrical node to which the electrical equipment belongs within the power distribution network area. This code is automatically generated and fixed by a topology identification algorithm during equipment installation and grid connection. Based on this physical access address, the node topology aggregation unit classifies and categorizes the cold-start transient impact power equivalent and the recovery allowable delay time sequence. It extracts all cold-start transient impact power equivalents belonging to the same feeder node and performs scalar accumulation. The calculation relationship of the local power rebound prediction value satisfies the following formula: , Among them, letters Represents the predicted value of local power rebound; letter Represents the total number of flexible loads belonging to the same feeder node; letters This represents the equivalent cold-start transient impact power of the m-th flexible load under this feeder node; symbol This represents the scalar summation of the power equivalent of all devices under this node. Since the physical dimension of the cold start transient impact power equivalent is kilowatt, after simple arithmetic summation of the same dimension, the physical dimension of the local power rebound prediction value strictly remains kilowatt. Its physical meaning is to accurately characterize the comprehensive power impact peak value that may be triggered by the release of control and concentrated load start-up of this single feeder branch.

[0048] For example, the node topology aggregation unit reads the physical access addresses of three flexible loads from the database. The first and second devices belong to the physical access address coded as feeder A, and the third device belongs to the physical access address coded as feeder B. By extracting the results of previous calculations, the cold start transient impact power equivalent of the first device is 150.0 kW, the second device is 50.0 kW, and the third device is 80.0 kW. The node topology aggregation unit performs a scalar accumulation operation on feeder A, substituting 150.0 kW and 50.0 kW into the accumulation formula and adding them together, thus calculating the predicted local power rebound value of feeder A to be 200.0 kW. For feeder B, since only one device is connected, 80.0 kW is directly extracted, and the predicted local power rebound value of feeder B is determined to be 80.0 kW. This process successfully aggregates the discrete individual impact parameters into a feeder-level physical prediction boundary through physical topology constraints.

[0049] Calculate the algebraic subtraction difference between the predicted local power rebound value and the equivalent power threshold of the current grid net load ramp rate, extract the target feeder node whose algebraic subtraction difference is greater than zero, generate an electrical isolation trigger pulse signal, and use the electrical isolation trigger pulse signal to disconnect the upstream grid-connected switch of the target feeder node and close the local energy storage connection switch.

[0050] The isolation control execution unit obtains the current grid net load ramp rate. To allow the rate dimension to be compared with the predicted local power rebound value representing absolute power, an evaluation time window parameter must be introduced to perform a physical multiplication mapping on the ramp rate. The evaluation time window parameter is a time constant used by the grid's underlying relay protection device to measure transient power safety; here, it is read as a constant of 1.0 minute based on the factory default configuration of the distribution area protection controller. The calculation relationship of the equivalent power threshold satisfies the following formula: , Among them, letters Represents the equivalent power threshold; letters Represents the current net load ramp-up rate of the power grid; letters This represents the evaluation time window parameter. Because... The dimension is kilowatts per minute. The units of measurement are minutes. After performing a scalar multiplication, the time dimension is eliminated, and the resulting equivalent power threshold unit strictly converges to kilowatts. Subsequently, the isolation control execution unit calculates the algebraic subtraction difference between the predicted local power bounce value and the equivalent power threshold. The calculation relationship of this algebraic subtraction difference satisfies the following formula: , Among them, letters Represents the difference in algebraic subtraction; letters Represents the predicted value of local power rebound; letter This represents the equivalent power threshold. When the calculated algebraic subtraction difference is greater than zero, it indicates that the transient power rebound that is about to occur on the feeder will exceed the grid's safe tolerance boundary. The isolation control execution unit extracts the feeder branch that exceeds the limit, locks it as the target feeder node, and the underlying digital signal processor of the control motherboard generates an electrical isolation trigger pulse signal with a pulse width of 100 milliseconds and an amplitude of 24 volts DC. The isolation control execution unit uses this electrical isolation trigger pulse signal to directly drive the circuit breaker trip coil of the target feeder node, instantly mechanically disconnecting the upstream grid-connected switch on the feeder; and after the auxiliary contacts of the grid-connected switch confirm the disconnection, it drives the AC contactor to close the local energy storage connection switch. Through this electrical isolation action, the high-risk rebound impact energy is contained within the distribution substation island without interrupting the power supply to the user side, preventing grid collapse. Figure 3 As shown, the process of accurately extracting over-limit feeder nodes by algebraic subtraction difference is revealed, demonstrating the underlying topology blocking mechanism of the present invention that protects the main network security through electrical and physical isolation under extreme rebound conditions.

[0051] For example, the isolation control execution unit obtains the current net grid load ramp rate as 120.0 kW per minute through the measurement gateway and reads the evaluation time window parameter as 1.0 minute. Substituting 120.0 kW per minute and 1.0 minute into the equivalent power threshold calculation formula, the equivalent power threshold is calculated to be 120.0 kW. Simultaneously, the generated local power rebound prediction value for feeder A is obtained as 200.0 kW, and the local power rebound prediction value for feeder B is 80.0 kW. For feeder A, substituting 200.0 kW and the threshold of 120.0 kW into the algebraic subtraction formula, subtracting 120.0 kW from 200.0 kW, the calculated algebraic subtraction difference is positive 80.0 kW. Since positive 80.0 kW is greater than zero, it is determined to face a severe impact risk, and feeder A is extracted as the target feeder node. The control board immediately sends a 24-volt electrical isolation trigger pulse signal to the switchgear where feeder A is located. The trip unit disconnects the grid-connected switch of feeder A, and then closes the energy storage access switch, allowing the local battery to maintain the 200.0 kW transient rebound demand. For feeder B, substituting 80.0 kW and 120.0 kW into the formula, the algebraic subtraction difference is calculated to be -40.0 kW. Since -40.0 kW is less than zero, it is determined that the rebound amplitude of feeder B is absolutely within the safety boundary. The isolation control execution unit does not send an electrical isolation trigger pulse signal, and feeder B continues to maintain its original grid-connected operation. This completes the final safety closed loop from algorithm prediction to hardware isolation.

[0052] Based on the same inventive concept, the present invention also provides a heterogeneous flexible load group response rebound effect suppression system, such as... Figure 4 As shown, the system includes: The phase state sequence generation module is used to collect real-time operating parameters of heterogeneous flexible loads, extract temperature deviation characteristics and state of charge deviation characteristics, calculate the normalized operating phase value of each flexible load in the working cycle, and splice the normalized operating phase value to generate a load group phase state sequence. The phase dispersion critical coefficient generation module is used to calculate the global phase synchronization coefficient based on the load group phase state sequence, collect the current power grid net load ramp rate, perform the algebraic mapping from the current power grid net load ramp rate to the constraint boundary, and generate the phase dispersion critical coefficient. The power reduction depth instruction set generation module is used to divide the heterogeneous flexible load into phase sub-intervals based on the phase state sequence of the load group, calculate the power reduction depth allocation weight of the phase sub-intervals according to the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, and generate a power reduction depth instruction set. The deficit vector generation module is used to receive the reduction depth instruction set, collect real-time state deviation data of each flexible load during the power reduction period, perform time integration operation on the real-time state deviation data step by step, and generate the demand deficit vector of each load. The delay duration sequence generation module is used to extract the demand deficit vector, calculate the percentile feature value of each flexible load in the demand deficit vector, calculate the product parameter of the percentile feature value and the current grid net load ramp rate, and generate the recovery permit delay duration sequence. The rebound prediction and topology isolation module is used to calculate the local power rebound prediction value of each feeder node based on the recovery permission delay duration sequence, extract the target feeder node whose local power rebound prediction value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnect the upstream grid connection switch of the target feeder node and close the local energy storage connection switch.

[0053] It should be noted that the functional division and information interaction between the various modules described above are logical, but in terms of physical implementation, they can be integrated on the same software platform or deployed in a distributed manner. The connections between them represent data flow and control flow, aiming to collaboratively achieve the objectives of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of protection of this invention.

Claims

1. A method for suppressing a response rebound effect of a heterogeneous flexible load group, characterized by, The method includes: Real-time operating parameters of heterogeneous flexible loads are collected, temperature deviation characteristics and state of charge deviation characteristics are extracted, normalized operating phase values ​​of each flexible load in the working cycle are calculated, and the normalized operating phase values ​​are spliced ​​together to generate a load group phase state sequence. Based on the load group phase state sequence, calculate the global phase synchronization coefficient, collect the current power grid net load ramp rate, perform the algebraic mapping from the current power grid net load ramp rate to the constraint boundary, and generate the phase dispersion critical coefficient. Based on the load group phase state sequence, the heterogeneous flexible load is divided into phase sub-intervals. According to the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, the power reduction depth allocation weight of the phase sub-interval is calculated, and a reduction depth instruction set is generated. Receive the reduction depth instruction set, collect real-time state deviation data of each flexible load during the power reduction period, perform time-integration operation on the real-time state deviation data step by step, and generate demand deficit vector of each load; Extract the demand deficit vector, calculate the percentile characteristic value of each flexible load in the demand deficit vector, calculate the product parameter of the percentile characteristic value and the current grid net load ramp rate, and generate a recovery permit delay duration sequence. Based on the recovery permit delay duration sequence, calculate the local power rebound prediction value of each feeder node, extract the target feeder node whose local power rebound prediction value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnect the upstream grid connection switch of the target feeder node and close the local energy storage connection switch.

2. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The process of splicing the normalized operating phase values ​​to generate the load group phase state sequence includes: The real-time operating parameters are analyzed to extract the temperature deviation characteristics between the ambient temperature of the temperature control device and the set control dead zone, and the current remaining power of the energy storage device and the rated charge / discharge capacity are extracted to generate the state of charge deviation characteristics. Based on the temperature deviation characteristics and the state of charge deviation characteristics, the algebraic ratio is calculated, and the algebraic ratio is linearly mapped to the standard circumferential phase angle interval to generate a normalized operating phase value. A one-dimensional vector is then concatenated according to the equipment physical address sequence to generate a load group phase state sequence.

3. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The critical coefficient for generating phase dispersion includes: Extract the normalized operating phase value of the load group phase state sequence, calculate the Euler complex exponent of the normalized operating phase value, perform arithmetic mean summation absolute modulus operation on the Euler complex exponent, and generate the global phase synchronization coefficient. Calculate the mathematical ratio parameter between the current net load ramp rate of the power grid and the maximum allowable current conversion threshold of the distribution area, and use the mathematical ratio parameter to perform scalar mapping calculation to generate the phase dispersion critical coefficient.

4. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The set of instructions for generating reduction depth includes: Extract the maximum normalized operating phase value and the minimum normalized operating phase value of the load group phase state sequence, calculate the numerical range, perform equal interval division on the numerical range to construct the phase sub-interval boundary, and map the heterogeneous flexible load to the phase sub-interval; Calculate the algebraic difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, use the algebraic difference to divide the basic allocation weights, count the device number density of the phase sub-intervals, use the device number density to perform scalar scaling calculation on the basic allocation weights, generate power reduction depth allocation weights and convert them into reduction depth instruction sets.

5. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The generation of demand deficit vectors for each load includes: The target reference baseline parameter is extracted by parsing the reduction depth instruction set, the current physical feedback parameter of each flexible load is collected during the power reduction period, the algebraic subtraction difference between the target reference baseline parameter and the current physical feedback parameter is calculated, and real-time state deviation data is constructed. Extract the discrete time step parameter of the control loop, calculate the scalar product of the real-time state deviation data and the discrete time step parameter, perform a traversal accumulation and summation operation on the scalar product within the power reduction cycle, and concatenate the accumulation and summation results according to the load node index to generate the demand deficit vector of each load.

6. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The method further includes: Extract the individual power deficit value of each flexible load in the demand deficit vector, extract the allowable delay seconds of each flexible load in the recovery allowable delay duration sequence, calculate the physical quotient of the individual power deficit value and the allowable delay seconds, and generate the cold start transient impact power equivalent.

7. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 1, characterized in that, The sequence of time for generating recovery permission delay includes: Perform an ascending sort operation on the values ​​of the demand deficit vector to construct an ordered sequence, extract the position index of each flexible load in the ordered sequence, calculate the algebraic ratio of the position index to the total number of loads, and generate percentile feature values. The product parameter is generated by calculating the scalar product of the percentile characteristic value and the current net load ramp rate of the power grid. The basic recovery time span value of the power grid area is collected. The physical product result of the product parameter and the basic recovery time span value is calculated. The value is recombined according to the equipment identification code to generate a recovery permit delay time sequence.

8. The method for suppressing the rebound effect of heterogeneous flexible load group response according to claim 6, characterized in that, The step of disconnecting the upstream grid connection switch of the target feeder node and closing the local energy storage connection switch includes: Obtain the physical access address of each flexible load, divide the cold start transient impact power equivalent and the recovery permission delay time sequence to feeder nodes according to the physical access address, calculate the cold start transient impact power equivalent under the same feeder node and perform scalar accumulation operation to generate local power rebound prediction value; Calculate the algebraic subtraction difference between the predicted local power rebound value and the equivalent power threshold of the current grid net load ramp rate, extract the target feeder node whose algebraic subtraction difference is greater than zero, generate an electrical isolation trigger pulse signal, and use the electrical isolation trigger pulse signal to disconnect the upstream grid-connected switch of the target feeder node and close the local energy storage connection switch.

9. A heterogeneous flexible load group response rebound effect suppression system, applied to the heterogeneous flexible load group response rebound effect suppression method as described in any one of claims 1-8, characterized in that, The system includes: The phase state sequence generation module is used to collect real-time operating parameters of heterogeneous flexible loads, extract temperature deviation characteristics and state of charge deviation characteristics, calculate the normalized operating phase value of each flexible load in the working cycle, and splice the normalized operating phase value to generate a load group phase state sequence. The phase dispersion critical coefficient generation module is used to calculate the global phase synchronization coefficient based on the load group phase state sequence, collect the current power grid net load ramp rate, perform the algebraic mapping from the current power grid net load ramp rate to the constraint boundary, and generate the phase dispersion critical coefficient. The power reduction depth instruction set generation module is used to divide the heterogeneous flexible load into phase sub-intervals based on the phase state sequence of the load group, calculate the power reduction depth allocation weight of the phase sub-intervals according to the difference between the global phase synchronization coefficient and the phase dispersion critical coefficient, and generate a power reduction depth instruction set. The deficit vector generation module is used to receive the reduction depth instruction set, collect real-time state deviation data of each flexible load during the power reduction period, perform time integration operation on the real-time state deviation data step by step, and generate the demand deficit vector of each load. The delay duration sequence generation module is used to extract the demand deficit vector, calculate the percentile feature value of each flexible load in the demand deficit vector, calculate the product parameter of the percentile feature value and the current grid net load ramp rate, and generate the recovery permit delay duration sequence. The rebound prediction and topology isolation module is used to calculate the local power rebound prediction value of each feeder node based on the recovery permission delay duration sequence, extract the target feeder node whose local power rebound prediction value is greater than the equivalent power threshold of the current grid net load ramp rate, disconnect the upstream grid connection switch of the target feeder node and close the local energy storage connection switch.

Images