A high-ratio wind power fast frequency modulation power distribution method

Abstract

The present application relates to the technical field of power distribution, in particular to a kind of high proportion wind power's quick frequency modulation power distribution method, comprising the following steps, the square difference of physical rotor speed and lower limit is calculated and combined with frequency wave to generate total power scheduling record, compare charge parameter and generate deviation record, determine the output dynamic weight of charge and discharge combined penalty factor according to record, weight and distribution parameter are fused to obtain cost data and extract micro-increment sequence, sequence is assigned to trial parameter and iteration is generated to generate balance array, screening standard feature adjusts communication format and generates distribution record.In the present application, the inertia gap of power grid is accurately completed by extracting available kinetic energy and frequency fluctuation characteristics, a matrix is constructed by combining charge deviation and penalty factor, and a micro-increment sequence is calculated, data is issued based on error convergence degree to balance multi-node loss and avoid over-limit, to suppress excessive attenuation of energy storage from the bottom, to ensure the collaborative efficiency and stability of wide-area frequency modulation resources.

Description

Technical Field

[0001] This invention relates to the field of power distribution technology, and in particular to a fast frequency regulation power distribution method for high-proportion wind power. Background Technology

[0002] The field of power allocation technology mainly involves the dispatching and control of power among various power generation equipment and energy storage systems in power systems. Core aspects of this field include grid load demand monitoring, active and reactive power dispatching, multi-power node coordination, and the mechanism for issuing unit combination commands. This field systematically studies how to dynamically balance power supply and consumption. Specifically, it involves collecting operating boundary parameters of each participating power source, grid node capacity, and real-time frequency deviation values. A distributed control model is used to disperse overall dispatching needs, and based on the collected values, the specific active power output setpoints for each synchronous generator, wind turbine, or energy storage converter are calculated. Among them, the traditional high-proportion wind power fast frequency regulation power allocation method refers to the technical issue that grid-connected wind farms cannot spontaneously respond to frequency changes due to the decoupling of wind turbine rotor speed and grid frequency. It usually adopts variable pitch control technology to directly change the mechanical pitch angle of wind turbine blades to reserve reserve active power capacity, or uses rotor kinetic energy control technology to adjust the electromagnetic torque on the generator stator side to release the rotational kinetic energy stored in the wind turbine body for a short time. In some cases, it is also combined with a centralized energy storage system. The total active power demand value is directly distributed to each lithium battery converter node inside the station according to a fixed equal distribution ratio through the main control layer of the station, and constant power charging and discharging actions are performed.

[0003] Traditional grid-connected wind farms often employ variable pitch control technology to alter the mechanical pitch angle and reserve capacity, or adjust electromagnetic torque to release rotational kinetic energy for short periods. Combined with a centralized energy storage system, the total active power demand is distributed to the converter nodes for constant power charging and discharging according to a fixed distribution ratio. This fixed-ratio distribution mode does not consider the real-time state of charge and operating loss costs of each node, causing some nodes at the low capacity boundary to operate excessively, exacerbating the overall fatigue and degradation of the equipment. The single kinetic energy release mechanism lacks accurate algebraic compensation for synchronous frequency fluctuations, resulting in limited local frequency regulation support or power overshoot when facing complex grid conditions. This significantly reduces the stability of dynamic frequency regulation across the entire field and the economic benefits of overall resource coordination and allocation. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a method for rapid frequency regulation power allocation of high-proportion wind power, comprising the following steps:

[0005] S1: Calculate the square values ​​of the actual speed of the physical rotor and the lower limit of safety respectively, and calculate the difference to obtain the physical square difference. Multiply it by the moment of inertia to generate the available kinetic energy storage. Combine it with the synchronous frequency fluctuation rate to calculate the virtual inertia increment, and sum it with the basic frequency regulation parameters to generate the total power scheduling record.

[0006] S2: Compare the real-time and reference charge parameters to obtain the capacity deviation ratio, calculate the span data through absolute polarity conversion to generate the state deviation parameter, and combine the real-time charge parameters to generate the energy storage node state deviation record.

[0007] S3: Based on the reverse characteristics of the total power scheduling record, determine the charging and discharging stage, call the energy storage node state deviation record to extract the state deviation parameter and real-time charge parameter, compare the benchmark parameter to determine the charge over-limit state and reconstruct the matrix, combine the penalty factor to calculate the weighted fusion ratio, generate the penalty weight parameter and output the dynamic penalty weight record.

[0008] S4: The penalty weight parameter recorded by the dynamic penalty weight is fused with the structural distribution parameter to generate total cost data. The slope change rate is calculated based on the gradient evolution trend to generate a corrected micro-increment sequence. The state micro-increment allocation record is generated by combining the format arrangement.

[0009] S5: Assign power trial parameters to the modified incremental sequence of the state incremental allocation record, calculate the convergence degree of the approximation iteration error to generate a gradient balance array, select array features that meet the distribution difference criteria to generate single-machine power configuration data, and adjust the communication format to generate a fast frequency modulation power allocation record.

[0010] As a further embodiment of the present invention, the total power scheduling record specifically includes the basic active power quota, the virtual inertia base, and the frequency regulation cycle timestamp; the energy storage node state deviation record includes the converter physical number, the absolute deviation difference, and the capacity polarity flag; the dynamic penalty weight record specifically includes the charging and discharging direction label, the capacity over-limit judgment symbol, and the weight amplification coefficient; the state micro-increment allocation record specifically includes the loss fusion index, the objective function gradient scalar, and the micro-increment evolution step size; and the fast frequency regulation power allocation record includes the single-machine execution quota, the communication message header, and the underlying frequency regulation control word.

[0011] As a further aspect of the present invention, the step of obtaining the total power scheduling record specifically includes:

[0012] S101: Monitor the actual speed and safety lower limit of the physical rotor under grid-connected conditions, perform exponentiation algebra calculations on the actual speed and safety lower limit respectively, obtain the square values ​​of the two, perform difference calculation on the square values ​​to generate physical square difference, obtain the inherent rotational inertia of the equipment, calculate the product fusion characteristics of physical square difference and rotational inertia, establish boundary limit base, and generate available kinetic energy storage.

[0013] S102: Collect the synchronous frequency fluctuation rate of the power grid nodes, perform micro-evolution analysis of the synchronous frequency fluctuation rate in the time dimension, extract the wave frequency derivative characteristics, perform algebraic compensation calculation based on the wave frequency derivative characteristics and the available kinetic energy stock, obtain transient support distribution data in the kinetic energy conversion process, establish an inertial compensation scalar, perform mapping analysis on the power compensation coefficient based on the inertial compensation scalar, and generate virtual inertial increment;

[0014] S103: Obtain the basic frequency regulation parameters of the scheduling node, perform feature identification on the numerical distribution and polarity attributes of the virtual inertia increment, extract the inertial support polarity identifier, combine the basic frequency regulation parameters and the virtual inertia increment to perform summation calculation, obtain the comprehensive adjustment quota, establish the power superposition base, perform data field format encapsulation processing on the power superposition base, and generate the total power scheduling record.

[0015] As a further aspect of the present invention, the step of obtaining the energy storage node state deviation record specifically includes:

[0016] S201: During the evaluation phase, collect the real-time charge parameters and reference charge parameters of the energy storage converter, perform relative algebraic comparison calculations on the real-time charge parameters and reference charge parameters, obtain the deviation of the energy storage battery's underlying charge, establish a difference mapping relationship, and calculate the relative difference ratio of the capacity dimension based on the difference mapping relationship to generate the capacity deviation ratio.

[0017] S202: The positive and negative polarity attributes of the capacity deviation ratio are numerically calibrated, and the absolute feature transformation mapping calculation is performed on the capacity deviation ratio in combination with the polarity attributes. The physical direction features of the value are stripped off, and the span data of the transformed capacity deviation ratio in the corresponding capacity range is extracted to generate the state deviation parameter.

[0018] S203: Call the state deviation parameter and the real-time charge parameter, perform multi-dimensional data cross-splitting operation on the state deviation parameter and the real-time charge parameter, combine the correlation features of the underlying power state and the deviation magnitude, establish a multi-dimensional structure of the node energy storage state, encapsulate the multi-dimensional structure in a standard communication format, and generate an energy storage node state deviation record.

[0019] As a further aspect of the present invention, the step of obtaining the dynamic penalty weight record specifically includes:

[0020] S301: Call the total power scheduling record under the scheduling condition, perform parsing operation on the power reversal feature of the total power scheduling record, extract the charging and discharging stage of the energy flow dimension, perform decomposition and extraction operation on the energy storage node state deviation record, peel off the underlying state deviation parameter and real-time charge parameter, and generate the energy storage basic parameter set.

[0021] S302: Based on the energy storage basic parameter set, obtain the preset reference charge parameters, perform a relative capacity level comparison operation on the real-time charge parameters and the reference charge parameters, obtain the deviation status identifier of the capacity boundary, establish the charge over-limit state, call the charging and discharging stage, and perform multi-dimensional cross-recombination operation on the charge over-limit state and the charging and discharging stage to generate a reconstruction mapping matrix.

[0022] S303: Based on the energy storage basic parameter set, call the state deviation parameter, obtain the preset penalty factor, combine the correlation features indicated by the reconstruction mapping matrix, perform a weighted fusion operation on the state deviation parameter and the penalty factor in terms of numerical dimensions, obtain the weight ratio base related to the adjustment cost, establish a penalty weight parameter, perform data domain communication format encapsulation processing on the penalty weight parameter, and generate a dynamic penalty weight record.

[0023] As a further aspect of the present invention, the process of obtaining the preset reference charge parameters specifically includes:

[0024] Read the initial upper limit threshold and initial lower limit threshold configured at the factory of the energy storage device, collect the cumulative charge and discharge cycle number data of the energy storage device, and perform interval reduction calculation on the initial upper limit threshold and initial lower limit threshold based on the cumulative charge and discharge cycle number data and the device capacity decay mapping table to generate dynamic scaling interval parameters, calculate the midpoint of the value of the dynamic scaling interval parameters, and generate the reference charge parameter.

[0025] The process of obtaining the preset penalty factor is as follows:

[0026] The frequency regulation urgency index data and the aging rate parameter of the energy storage device are collected and distributed. A numerical product operation is performed on the frequency regulation urgency index data and the aging rate parameter to obtain the initial deviation weight value. The upper limit penalty boundary value and the lower limit penalty boundary value configured in the device operation constraint specification are read. An interval limiting operation is performed on the initial deviation weight value based on the upper limit penalty boundary value and the lower limit penalty boundary value to obtain the limiting value and generate the penalty factor.

[0027] As a further aspect of the present invention, the step of obtaining the state incremental allocation record specifically includes:

[0028] S401: Obtain the structural distribution parameters of equipment loss characteristics during the configuration phase; extract the penalty weight parameters for the dynamic penalty weight record; perform feature combination operation on the penalty weight parameters and structural distribution parameters to obtain the fusion base of the two; establish the cost basis of node collaboration; and generate total cost data.

[0029] S402: Call the total cost data to obtain the corresponding power specification, extract the micro-evolution features of the mapping interval between the total cost data and the power specification, obtain the gradient evolution trend of the power interval, perform differential difference calculation based on the gradient evolution trend, extract the slope change rate of the change curve, establish the adjustment step size sequence of the power allocation node, and generate the correction micro-increment sequence.

[0030] S403: Reconstruct and transform the underlying data type for the modified incremental sequence, identify the sequence arrangement characteristics of each element, adjust the internal data arrangement format based on the sequence arrangement characteristics, establish a structure-aligned fine-tuning allocation array, perform communication encapsulation processing on the data fields for the fine-tuning allocation array, and generate a state incremental allocation record.

[0031] As a further aspect of the present invention, the process of obtaining the structural distribution parameters of the equipment loss characteristics during the configuration phase specifically includes:

[0032] During the configuration phase, the charging and discharging current timing sequence and the operating voltage timing sequence of the energy storage device were collected.

[0033] Time-domain product-integral operations are performed on the charging and discharging current timing sequence and the operating terminal voltage timing sequence to extract the total charging input energy data and the total discharging output energy data;

[0034] Perform algebraic difference calculation on the total charging input energy data and the total discharging output energy data to obtain charge-discharge conversion loss data;

[0035] Monitor the port static voltage drop and self-discharge current of the energy storage device in standby mode, and calculate the static standby loss parameters;

[0036] Extract multiple operating power reference nodes corresponding to the configuration phase;

[0037] The charge-discharge conversion loss data and the static standby loss parameters are mapped to the corresponding multiple operating power reference nodes, and a quadratic polynomial fitting operation is performed to obtain the combination of loss curve fitting coefficients.

[0038] A multidimensional loss feature matrix is ​​constructed based on the combination of the loss curve fitting coefficients, and the structural distribution parameters are generated.

[0039] As a further aspect of the present invention, the step of obtaining the fast frequency modulation power allocation record specifically includes:

[0040] S501: Obtain the running data of the issued working condition, call the state micro-increment allocation record, extract the internal correction micro-increment sequence, allocate the corresponding power test parameter based on the correction micro-increment sequence, perform approximation iterative operation on the power test parameter, calculate the error convergence during the iterative calculation process, establish the feature distribution mapping relationship based on the error convergence, and generate a gradient balance array.

[0041] S502: Perform a horizontal comparison calculation on the internal element features of the gradient balancing array to obtain the distribution differences between multiple element features, read the constraints used for boundary filtering, perform a threshold comparison judgment on the distribution differences and constraints, filter the array features whose distribution differences meet the constraints, calculate the output quota value of the corresponding device in combination with the array features, and generate single-machine power configuration data.

[0042] S503: Perform a low-level message parsing and reading operation on the single-unit power configuration data, extract the internal communication format, adjust the communication format according to the content layout requirements of the standard communication protocol, modify the message header and control field of the single-unit power configuration data, establish a format-aligned node scheduling instruction sequence, and generate a fast frequency modulation power allocation record.

[0043] As a further aspect of the present invention, the process of calculating the error convergence during the iterative calculation is specifically as follows:

[0044] Extract the total demand power baseline data contained in the issued working condition, perform numerical summation operation on all allocated power test parameters to obtain the global test power accumulation value, calculate the absolute difference between the global test power accumulation value and the total demand power baseline data, extract the dynamic power deviation span and define it as the error convergence degree;

[0045] The process of reading the constraints used for boundary filtering is as follows:

[0046] Read the rated maximum output power extreme value, rated minimum output power extreme value, inherent physical ramp rate parameter and action response time constant of the corresponding device, perform a product operation on the inherent physical ramp rate parameter and the action response time constant to obtain the single maximum power jump amplitude, integrate the rated maximum output power extreme value, the rated minimum output power extreme value and the single maximum power jump amplitude to construct the operating boundary set, read the iteration tolerance lower limit of the algorithm configuration file, and define the operating boundary set and the iteration tolerance lower limit together as the constraint condition.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0048] In this invention, the physical square difference is obtained by calculating the square of the actual speed of the physical rotor and the lower limit of safety, and the virtual inertia increment is calculated by combining the moment of inertia and frequency fluctuation rate to generate quotas. This accurately fills the inertia gap in the early stage of grid connection. The energy storage charge parameters are extracted and compared with the benchmark capacity to generate the deviation ratio. A reconstruction matrix and penalty weight parameters are established by combining the penalty factor. The total cost data of equipment structure distribution is calculated to refine and correct the micro-increment sequence. The configuration data is issued based on the error convergence degree to balance the charging and discharging costs of multiple nodes and avoid over-limit operation. The excessive loss of a single node is suppressed from the underlying architecture, and the efficiency of wide-area resource collaborative calling and the accuracy of dynamic frequency regulation are guaranteed in all aspects. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a schematic diagram of the steps of the present invention;

[0051] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0052] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0053] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0054] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0055] Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0056] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0057] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0058] Please see Figure 1 This invention provides a method for rapid frequency regulation power allocation of high-proportion wind power, comprising the following steps:

[0059] S1: Calculate the square values ​​of the actual speed of the physical rotor and the lower limit of safety respectively, and calculate the difference to obtain the physical square difference. Multiply it by the moment of inertia to generate the available kinetic energy storage. Combine it with the synchronous frequency fluctuation rate to calculate the virtual inertia increment, and sum it with the basic frequency regulation parameters to generate the total power scheduling record.

[0060] S2: Compare the real-time and reference charge parameters to obtain the capacity deviation ratio, calculate the span data through absolute polarity conversion to generate the state deviation parameter, and combine the real-time charge parameters to generate the energy storage node state deviation record.

[0061] S3: Determine the charging and discharging stage based on the reverse characteristics of the total power scheduling record, call the energy storage node state deviation record to extract the state deviation parameter and real-time charge parameter, compare the benchmark parameter to determine the charge over-limit state and reconstruct the matrix, combine the penalty factor to calculate the weighted fusion ratio, generate the penalty weight parameter and output the dynamic penalty weight record.

[0062] S4: The penalty weight parameter recorded by the dynamic penalty weight is fused with the structural distribution parameter to generate total cost data. The slope change rate is calculated based on the gradient evolution trend to generate a corrected micro-increment sequence. Combined with the format arrangement, a state micro-increment allocation record is generated.

[0063] S5: Assign power trial parameters to the corrected incremental sequence of the state incremental assignment record, calculate the convergence degree of the approximation iteration error to generate a gradient balance array, select array features that meet the distribution difference criteria to generate single-machine power configuration data, and adjust the communication format to generate a fast frequency modulation power allocation record.

[0064] The total power scheduling record specifically includes the basic active power quota, virtual inertia base, and frequency regulation cycle timestamp. The energy storage node state deviation record includes the converter physical number, absolute deviation difference, and capacity polarity flag. The dynamic penalty weight record specifically includes the charging and discharging direction label, capacity over-limit judgment symbol, and weight amplification coefficient. The state micro-increment allocation record specifically includes the loss fusion index, objective function gradient scalar, and micro-increment evolution step size. The fast frequency regulation power allocation record includes the single-unit execution quota, communication message header, and underlying frequency regulation control word.

[0065] Please see Figure 2 The specific steps for obtaining the total power scheduling record are as follows:

[0066] S101: Monitor the actual speed and safety lower limit of the physical rotor under grid-connected conditions, perform exponentiation algebra calculations on the actual speed and safety lower limit respectively, obtain the square values ​​of the two, perform difference calculation on the square values ​​to generate physical square difference, obtain the inherent rotational inertia of the equipment, calculate the product fusion characteristics of physical square difference and rotational inertia, establish boundary limit base, and generate available kinetic energy storage.

[0067] The rotational angular velocity of the physical rotor is acquired in real time using an optical encoder installed on the main shaft of the wind turbine, with a sampling period of 10 milliseconds. A pre-set safe lower limit value for the rotor speed, set to 70% of the rated speed, is read from the local main control unit memory address of the wind turbine. The acquired actual speed value and the safe lower limit value are then multiplied by themselves (the actual speed multiplied by itself to obtain the square of the actual speed, and the safe lower limit multiplied by itself to obtain the square of the lower limit). The square of the actual speed is then subtracted from the square of the lower limit to obtain the algebraic difference, which is defined as the physical square difference. The total moment of inertia parameter of the wind turbine and generator rotor, referred to the low-speed shaft, is obtained by reading the data table on the wind turbine's nameplate; this moment of inertia parameter is, for example, 5,000,000 kg / m². The calculated physical squared difference is algebraically multiplied with the aforementioned moment of inertia parameter. In this multiplication, the result is then multiplied by a constant of 0.5 to obtain the actual rotational kinetic energy that the physical rotor can currently release to the grid. This value, along with its upper and lower limits, forms the boundary constraint base, ultimately generating the available kinetic energy reserve. The calculation results of this available kinetic energy reserve demonstrate the upper limit of energy reserves that the current wind turbine can provide for the grid frequency response without triggering the underlying low-speed shutdown protection logic, providing accurate energy boundary data for subsequent inertia support.

[0068] S102: Collect the synchronous frequency fluctuation rate of the power grid nodes, perform micro-evolution analysis of the synchronous frequency fluctuation rate in the time dimension, extract the wave frequency derivative characteristics, perform algebraic compensation calculation based on the wave frequency derivative characteristics and available kinetic energy stock, obtain transient support distribution data in the kinetic energy conversion process, establish an inertial compensation scalar, perform mapping analysis on the power compensation coefficient based on the inertial compensation scalar, and generate virtual inertial increment.

[0069] The phase-locked loop (PLL) device at the grid connection point acquires three-phase voltage signals in real time at 5-millisecond intervals and extracts the synchronous frequency fluctuation data sequence of the grid node. A time-dimensional micro-evolution analysis based on a 100-millisecond sliding window is performed on this synchronous frequency fluctuation data sequence. Specifically, the instantaneous rate of change of frequency is calculated by subtracting the frequency fluctuation values ​​of two adjacent sampling points and dividing by the sampling interval of 5 milliseconds, thereby extracting the wave frequency derivative feature. The available kinetic energy stock value generated in the previous steps is used, and the extracted wave frequency derivative feature is directly algebraically multiplied with the available kinetic energy stock value to obtain transient support distribution data during the kinetic energy conversion process. This establishes an inertial compensation scalar containing information on support amplitude and energy release direction. After establishing the inertial compensation scalar, a preset base power compensation coefficient is read from the control logic core register, with a value set to 1.5. The calculated inertial compensation scalar is multiplied by this base power compensation coefficient for mapping analysis, yielding the specific value of the additional active power that the wind turbine needs to release or absorb under this specific frequency change rate, ultimately generating the virtual inertia increment. The advantage of this virtual inertia increment calculation process is that by dynamically coupling the available physical kinetic energy of the rotor with the severity of the grid frequency change, the adaptive adjustment of the virtual inertia support strength of the grid-connected wind turbine is achieved.

[0070] S103: Obtain the basic frequency regulation parameters of the scheduling node, identify the features of the numerical distribution and polarity attributes of the virtual inertia increment, extract the inertia support polarity identifier, combine the basic frequency regulation parameters with the virtual inertia increment to perform summation calculation, obtain the comprehensive regulation quota, establish the power superposition base, perform data field format encapsulation processing on the power superposition base, and generate the total power scheduling record.

[0071] The basic frequency regulation parameters issued by the current dispatch node are obtained from the wind farm's automatic generation control host via the internal control network communication bus. These parameters represent the static active power output adjustment command calculated based on the conventional frequency deviation, and their value is, for example, 2 MW. The virtual inertia increment data generated in the previous steps is retrieved, and feature identification and judgment are performed on the numerical distribution and positive / negative polarity attributes of the virtual inertia increment. Specifically, if the virtual inertia increment value is greater than 0, a positive inertia support polarity identifier representing discharge or increased generation is extracted; if the virtual inertia increment value is less than 0, a negative inertia support polarity identifier representing charging or reduced generation is extracted. Combining the extracted inertia support polarity identifiers, the obtained basic frequency regulation parameters and the virtual inertia increment are directly algebraically added. For example, when the basic frequency regulation parameter is 2 MW, the virtual inertia increment is 0.5 MW, and the identifier is positive, the two are added together to obtain a total comprehensive regulation quota of 2.5 MW, and this value is used to establish the power superposition base. Subsequently, in accordance with the message data structure requirements that the underlying wind turbine converter controller can recognize, the power superposition base is encapsulated in the standard format of the data field, and a standard data frame header, addressing code bit and cyclic redundancy check code are added. Finally, a total power scheduling record that can directly guide the underlying hardware to perform electromagnetic torque adjustment actions is generated.

[0072] Please see Figure 3 The specific steps for obtaining the energy storage node state deviation record are as follows:

[0073] S201: During the evaluation phase, collect the real-time charge parameters and reference charge parameters of the energy storage converter, perform relative algebraic comparison calculations on the real-time charge parameters and reference charge parameters, obtain the deviation of the energy storage battery's underlying charge, establish a difference mapping relationship, and calculate the relative difference ratio of the capacity dimension based on the difference mapping relationship to generate the capacity deviation ratio.

[0074] Using a battery management device deployed on the DC side of the energy storage converter, real-time charge parameters of each battery cluster connected to the energy storage converter are collected at a refresh rate of 20 milliseconds during the evaluation phase. Simultaneously, a pre-set reference charge parameter for the current aging state of the energy storage converter is read from the non-volatile memory of the control unit; this reference charge parameter is set to 50%. A relative algebraic comparison is performed by subtracting the reference charge parameter value from the real-time charge parameter value to obtain the actual capacity deviation of the underlying energy storage battery. For example, when the real-time charge parameter is 65%, the difference with the reference charge parameter yields a capacity deviation of +15%. Based on this deviation, a difference mapping relationship is established between it and the entire available capacity range. This involves dividing the calculated capacity deviation by the total span of the rated available charge range of the specific energy storage converter (e.g., set to 80%), and performing a division calculation based on this difference mapping relationship to obtain the relative difference percentage in the capacity dimension. In the example above, dividing 15% by 80% yields a calculated value of 0.1875. This result is then used to generate the capacity deviation ratio. Table 1 lists the specific capacity deviation ratios obtained through division comparison under different real-time charging parameters.

[0075] Table 1 Calculation Table of Energy Storage Node Capacity Deviation Ratio

[0076]

[0077] As shown in Table 1, the capacity deviation ratio is accurately calculated through quantitative division algebraic operations, providing a reliable mathematical quantitative basis for subsequent judgment of the charge and discharge margin of each distributed energy storage node.

[0078] S202: Numerical calibration of positive and negative polarity attributes for capacity deviation ratio, and absolute feature transformation mapping calculation of capacity deviation ratio in combination with polarity attribute, stripping physical direction features of numerical value, extracting span data of capacity deviation ratio in corresponding capacity range after transformation, and generating state deviation parameter.

[0079] The capacity deviation ratio calculated in the preceding stages is extracted, and its positive or negative polarity is determined logically. When the capacity deviation ratio is greater than 0, its polarity is explicitly marked as positive, indicating that the current energy storage converter node is in a state of relative power surplus; when the value is less than 0, its polarity is marked as negative, indicating that the current node is in a state of power deficit. Based on the determined polarity status attribute, an absolute feature transformation mapping calculation is performed on the capacity deviation ratio. Specifically, the absolute value of the algebraic value of the capacity deviation ratio is directly taken, thereby removing the positive or negative sign representing the physical charging and discharging direction of the original value. Through this absolute value extraction calculation, the pure span data of the transformed capacity deviation ratio within the corresponding total available capacity range is obtained. For example, when the input capacity deviation ratio is -0.125, the extracted undirected span data after transformation is 0.125. Finally, this span data is assigned a new variable name to generate a state deviation parameter. The advantage of generating this state deviation parameter is that by eliminating the mathematical interference of positive and negative energy direction characteristics, it enables the subsequent penalty weight allocation algorithm used for cost optimization to uniformly adopt distance span as the metric, thus simplifying the code complexity of multi-node collaborative control logic.

[0080] S203: Call the state deviation parameter and the real-time charge parameter, perform multi-dimensional data cross-splitting operation on the state deviation parameter and the real-time charge parameter, combine the correlation features of the underlying power state and the deviation magnitude, establish a multi-dimensional structure of the node energy storage state, encapsulate the multi-dimensional structure in the standard communication format, and generate the energy storage node state deviation record.

[0081] The system retrieves the state deviation parameter data obtained in the previous step from the cache and simultaneously reads the corresponding real-time charge parameter raw data from the register. A multi-dimensional data cross-concatenation operation is performed on the state deviation parameter and the real-time charge parameter. Specifically, a 16-byte contiguous data block is allocated in memory. The real-time charge parameter value, representing the underlying physical electrical state, is converted to a 32-bit single-precision floating-point number and stored in the first 4 bytes. The state deviation parameter, representing the deviation distance, is converted to a 32-bit single-precision floating-point number and immediately stored in the following 4 bytes. This tightly combines these two key correlated features at the physical memory level, establishing a data carrier reflecting the multi-dimensional structure of the node's energy storage operation status. This multi-dimensional structure is then encapsulated using a standard communication format adapted for industrial LANs. A 2-byte converter node physical addressing identifier is added to the beginning of the data block, and a 2-byte data checksum field is appended to the end of the data block to ensure data transmission integrity on a bus with strong electromagnetic interference. After rigorous memory byte alignment and packet header and trailer encapsulation, a record of energy storage node state deviation that can be broadcast on the communication network is finally generated.

[0082] Please see Figure 4 The specific steps for obtaining the dynamic penalty weight record are as follows:

[0083] S301: Call the total power scheduling record under the scheduling condition, perform parsing operation on the power reversal feature of the total power scheduling record, extract the charging and discharging stage of the energy flow dimension, perform decomposition and extraction operation on the energy storage node state deviation record, peel off the underlying state deviation parameter and real-time charge parameter, and generate the energy storage basic parameter set.

[0084] The algorithm invokes the total power scheduling record under the current scheduling condition through the underlying task scheduling mechanism. It then parses the algebraic positive and negative signs of the total power scheduling command values ​​carried in this record to extract the actual power reversal characteristics. If the parsed total power scheduling command value is positive, the algorithm determines that the current task requirement is to inject active power into the external power grid, thus extracting the energy flow dimension as the discharge execution phase. If the parsed value is negative, the algorithm determines that the task requirement is to absorb active power from the power grid, thus extracting the energy flow dimension as the charging execution phase. Subsequently, the algorithm performs byte-level decomposition and extraction operations on the aforementioned energy storage node state deviation records issued by the network. According to the pre-set byte offset settings of the communication protocol, the communication packet header and footer are stripped, and the floating-point values ​​of the state deviation parameter and the real-time charge parameter hidden in the underlying message data load segment are read. The energy charging / discharging stage tag bits, the specific values ​​of the state deviation parameter, and the specific values ​​of the real-time charge parameter extracted above are then subjected to unified memory address mapping and clustering to generate a basic energy storage parameter set containing the current core operating parameters of the specific energy storage device.

[0085] S302: Based on the energy storage basic parameter set, obtain the preset reference charge parameters, perform a relative capacity level comparison operation on the real-time charge parameters and the reference charge parameters, obtain the deviation status identifier of the capacity boundary, establish the charge over-limit state, call the charging and discharging stage, and perform multi-dimensional cross-recombination operation on the charge over-limit state and the charging and discharging stage to generate a reconstruction mapping matrix.

[0086] Based on the energy storage fundamental parameter set constructed in the preceding steps, a preset benchmark charge parameter value is obtained from the core operating threshold configuration table of the control unit. This benchmark value clearly defines the mathematical center point of the optimal healthy operating range of the energy storage battery pack. The real-time charge parameters from the parameter set are then called, and their real-time collected values ​​are compared with the benchmark charge parameters at the capacity level. In the comparison logic, if the real-time charge parameter value exceeds the preset safe capacity upper limit (e.g., 90%) or falls below the preset safe capacity lower limit (e.g., 10%), the deviation status identifier of the capacity boundary is output as a value of 1, thus establishing a charge over-limit state with a risk of equipment lifespan reduction; if it is within the safe range of 10% to 90%, the identifier output is 0. Next, the previously successfully extracted charge / discharge stage identifier is called, and a multi-dimensional cross-recombination operation is performed between the charge over-limit state identifier and the charge / discharge stage identifier. Specifically, a two-dimensional reconstruction mapping matrix is ​​allocated and assigned in memory, where the row indices correspond to the charge / discharge stages, and the column indices correspond to the charge over-limit state. Table 2 shows the numerical values ​​of the associated risk characteristics output by the reconstructed mapping matrix under different combinations.

[0087] Table 2. Output of Risk Characteristics of Reconstructed Mapping Matrix

[0088]

[0089] As shown in Table 2, by constructing and querying this two-dimensional reconstruction mapping matrix, the control architecture can output the characteristic value of the operational constraint risk level faced by the current node under a specific power action command, providing an accurate state coordinate mapping system for the subsequent mathematical calculation of penalty weights.

[0090] S303: Based on the basic parameter set of energy storage, call the state deviation parameter, obtain the preset penalty factor, combine the correlation features indicated by the reconstruction mapping matrix, perform weighted fusion calculation of the numerical dimension on the state deviation parameter and the penalty factor, obtain the weight ratio base related to the adjustment cost, establish the penalty weight parameter, encapsulate the communication format of the data domain on the penalty weight parameter, and generate dynamic penalty weight record.

[0091] Based on the memory pointer guidance provided by the basic energy storage parameter set, the state deviation parameter value corresponding to the energy storage node is invoked. Simultaneously, a preset basic penalty factor floating-point value is obtained by reading the underlying frequency regulation algorithm configuration file. This value is based on the control architecture's maximum tolerance for uneven battery charge levels across the entire battery pack, for example, a fixed value of 2.5. Combined with the risk characteristic indication value queried from the reconstructed mapping matrix in the previous step, if the matrix indicates that the current charge / discharge movement will worsen the risk of exceeding limits (characteristic value 3.0), the penalty factor is multiplied by this characteristic value of 3.0 to amplify it; if it helps alleviate the risk of exceeding limits (characteristic value 0.5), the penalty factor is multiplied by 0.5 to reduce it. For the state deviation parameter and the final penalty factor after adjustment and amplification or reduction, a weighted fusion operation is performed on the numerical dimensions to obtain the weight ratio base directly related to the adjustment optimization cost. This weight ratio base is established as a dedicated penalty weight parameter in the program logic. Subsequently, the penalty weight parameter is encapsulated in the standard communication format of the data domain, and the timestamp of the current microprocessor clock and the physical addressing encoding sequence of the target converter node are added to it. Finally, the dynamic penalty weight record is output and generated.

[0092] Please see Figure 5 The specific steps for obtaining the state increment allocation record are as follows:

[0093] S401: Obtain the structural distribution parameters of equipment loss characteristics during the configuration phase, extract the penalty weight parameters for the dynamic penalty weight records, perform feature combination operations on the penalty weight parameters and structural distribution parameters, obtain the fusion base of the two, establish the cost basis of node collaboration, and generate total cost data.

[0094] The system retrieves a structural distribution parameter array representing various operational loss characteristics of the equipment during the configuration phase from the site-level equipment asset health management database. It then performs message decomposition and parsing on the dynamic penalty weight records transmitted from the upper-level network socket, extracting single-precision floating-point penalty weight parameter values. These extracted penalty weight parameter values ​​are used as mathematical operators, performing feature combination operations with each specific loss element in the aforementioned structural distribution parameter array. The specific mathematical combination operation logic is as follows: the penalty weight parameter is treated as an independent multiplication weight coefficient, algebraically multiplied with the converter basic conversion loss coefficient value recorded in the structural distribution parameter array, and then the product is directly algebraically added to the aging depreciation cost base of the equipment battery pack. Through this composite algebraic operation process of multiplication followed by addition, the final fused base of the two physical costs is obtained. Within the algorithm framework, this fused base is explicitly established as a unified cost base scalar for multi-node collaborative power allocation, and a total cost data sequence representing the comprehensive operating cost of a single machine is generated based on this value. This cross-dimensional fusion and combination operation effectively breaks down the data barrier between physical loss hardware parameters and algorithmic virtual penalty weights.

[0095] S402: Call the total cost data, obtain the corresponding power specifications, extract the micro-evolution features of the mapping interval between the total cost data and the power specifications, obtain the gradient evolution trend of the power interval, perform differential difference calculation based on the gradient evolution trend, extract the slope change rate of the change curve, establish the adjustment step size sequence of the power allocation node, and generate the correction micro-increment sequence.

[0096] The calculated total cost data sequence is loaded into the arithmetic register array, and the corresponding rated power limit values ​​of each converter device are simultaneously obtained from the equipment ledger. A function mapping interval is constructed in the algorithm memory, with the total cost data as the dependent variable and the adjustable power specification as the independent variable. Micro-evolutionary features are extracted for this mapping interval, specifically using a two-point discrete forward difference algorithm to obtain the cost gradient evolution trend at discrete sampling nodes throughout the entire adjustable power range. Next, based on the extracted cost gradient evolution trend, further differential difference algebraic calculations are performed, i.e., the cost gradient difference between two adjacent discrete power nodes is extracted and divided by the actual physical increment magnitude of these two corresponding power nodes, thereby calculating and extracting the discrete second derivative of the cost function change curve, which is defined as the slope change rate value. This slope change rate value is multiplied by the preset global convergence step size scaling constant in the algorithm configuration file to establish the adjustment step size sequence for each power allocation participating node in the subsequent iterative algorithm approaching the optimal solution. The control logic finally establishes this array containing step size values ​​and generates a corrected micro-increment sequence.

[0097] S403: Reconstruct and transform the underlying data type for the modified incremental sequence, identify the sequence arrangement characteristics of each element, adjust the internal data arrangement format based on the sequence arrangement characteristics, establish a structure-aligned fine-tuning allocation array, perform communication encapsulation processing on the data fields for the fine-tuning allocation array, and generate state incremental allocation records.

[0098] For the incremental correction sequence remaining in memory from the previous calculation stage, a reconstruction and conversion operation of the underlying data type is performed. The 64-bit double-precision floating-point array originally used for high-precision calculations in the main control unit is uniformly truncated according to rules and converted into a 32-bit single-precision floating-point data format to adapt to the limited communication bus bandwidth of the underlying slave controllers within the site. The physical arrangement order and layout characteristics of each fine-tuning step element in the sequence are identified and extracted through a traversal algorithm. Combined with the identified memory location sequence layout characteristics, the final arrangement format of the internal data is adjusted, and each single-precision incremental floating-point data is tightly structured and bound to its corresponding converter device underlying logical index number, establishing an internally structured aligned fine-tuning allocation array. Communication layer encapsulation processing is performed on the data fields of this fine-tuning allocation array, wrapping it with a message header conforming to the control protocol rules, read / write function control codes, and target hardware register start address identifier bits. Finally, it is serialized to generate a state incremental allocation record that can be transmitted via an Ethernet switch.

[0099] Please see Figure 6 The specific steps for obtaining the fast frequency modulation power allocation record are as follows:

[0100] S501: Obtain the running data of the issued working conditions, call the status micro-increment allocation record, extract the internal correction micro-increment sequence, allocate the corresponding power test parameters based on the correction micro-increment sequence, perform approximation iterative operation on the power test parameters, calculate the error convergence during the iterative calculation process, establish the feature distribution mapping relationship based on the error convergence, and generate the gradient balance array.

[0101] The main control unit obtains real-time operating data from the upper-level power grid dispatch automation platform via its remote Ethernet physical interface, extracting the total demand power baseline data, which includes active power baseline values. It uses memory pointers to access previously resident state increment allocation records in the memory stack, removes packet headers using a protocol parsing function, and extracts the internally packaged array of pure correction increment sequences. Based on the independent step-size weight values ​​assigned to each converter device in this correction increment sequence, it assigns initial power probe parameter values ​​for the first iteration to each standby device within the substation. It performs approximation iterative algebraic operations on these distributed power probe parameters. Specifically, within the processor's logic control loop, it sums the current power probe parameters of each device and performs a direct algebraic subtraction operation with the extracted total demand power baseline data. The absolute value of the mathematical residual between the two in the current iteration is calculated, and this residual value is defined in the algorithm program as the error convergence degree of the current computing node. Based on the discrete error convergence generated by several consecutive iterations and the corresponding power allocation parameter trial combination scheme, a feature distribution mapping relationship is established, and then the feature group of the scheme that meets the convergence standard is used to generate a gradient balance array.

[0102] S502: Perform a horizontal comparison calculation on the internal element features of the gradient balancing array to obtain the distribution differences between multiple element features, read the constraints used for boundary filtering, perform a threshold comparison judgment on the distribution differences and constraints, filter the array features whose distribution differences meet the constraints, combine the array features to calculate the output quota value of the corresponding device, and generate single-unit power configuration data.

[0103] A lateral scan and comparison calculation is performed on all internal element features of the gradient balancing array temporarily established in the previous calculation stage. By extracting the percentage value of the trial power allocated to different energy storage nodes relative to their inherent rated capacity in the same iterative batch of data, a multi-node percentage range calculation is performed to obtain the output distribution span between the element features of multiple converter equipment nodes. Simultaneously, hard constraint parameters for boundary exceedance filtering are read from the configuration file of the underlying protection execution program. A threshold comparison judgment operation is performed on the calculated output distribution range value and the hard constraint condition: if the distribution range corresponding to the array feature exceeds the set maximum allowable limit, the execution logic removes the feature array from the available solution candidate pool; after a loop filtering mechanism, the array feature group whose distribution difference meets the constraint condition is finally selected and retained. Combining the specific power adjustment base percentage contained in the selected set of optimal array features, it is multiplied by the corresponding hardware rated active capacity limit value of the converter equipment to calculate the kilowatt-level output quota value that the equipment needs to execute. The control logic generates single-unit power configuration data from this set of values ​​containing the target of all units.

[0104] S503: Performs low-level message parsing and reading operations on single-unit power configuration data, extracts the internal communication format, adjusts the communication format according to the content layout requirements of the standard communication protocol, modifies the message header and control field of single-unit power configuration data, establishes a format-aligned node scheduling instruction sequence, and generates a fast frequency modulation power allocation record.

[0105] For the single-unit power configuration data set finally determined by the above optimization calculation, the underlying network message architecture is parsed and bit-level read operations are performed to extract the custom structure communication format temporarily used for transmission between algorithm components. Based on the control word content layout requirements of the communication protocol followed by the grid-connected inverter equipment, the final communication format appearance of the configuration data is readjusted. Specific modifications include modifying the physical message header identifier of the single-unit power configuration data to match it to the factory MAC hardware address of each target inverter; simultaneously, the calculated floating-point kilowatt-level output rating is written into the specific reserved control field register mapping data area within the message structure. Through the above protocol message conversion and header addressing bit rewriting operations, a node concurrent scheduling instruction sequence aligned with the underlying format is established in the communication memory buffer. This serves as the final step to generate a fast frequency modulation power allocation record that can be directly sent to the network physical layer for level signal conversion. This generated message record will trigger the pulse width modulation duty cycle adjustment of the underlying insulated-gate bipolar transistor, completing all hardware closed-loop control steps.

[0106] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for rapid frequency regulation power allocation of high-proportion wind power, characterized in that, Includes the following steps: S1: Calculate the square values ​​of the actual speed of the physical rotor and the lower limit of safety respectively, and calculate the difference to obtain the physical square difference. Multiply it by the moment of inertia to generate the available kinetic energy storage. Combine it with the synchronous frequency fluctuation rate to calculate the virtual inertia increment, and sum it with the basic frequency regulation parameters to generate the total power scheduling record. S2: Compare the real-time and reference charge parameters to obtain the capacity deviation ratio, calculate the span data through absolute polarity conversion to generate the state deviation parameter, and combine the real-time charge parameters to generate the energy storage node state deviation record. S3: Based on the reverse characteristics of the total power scheduling record, determine the charging and discharging stage, call the energy storage node state deviation record to extract the state deviation parameter and real-time charge parameter, compare the benchmark parameter to determine the charge over-limit state and reconstruct the matrix, combine the penalty factor to calculate the weighted fusion ratio, generate the penalty weight parameter and output the dynamic penalty weight record. S4: The penalty weight parameter recorded by the dynamic penalty weight is fused with the structural distribution parameter to generate total cost data. The slope change rate is calculated based on the gradient evolution trend to generate a corrected micro-increment sequence. The state micro-increment allocation record is generated by combining the format arrangement. S5: Assign power trial parameters to the modified incremental sequence of the state incremental allocation record, calculate the convergence degree of the approximation iteration error to generate a gradient balance array, select array features that meet the distribution difference criteria to generate single-machine power configuration data, and adjust the communication format to generate a fast frequency modulation power allocation record.

2. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 1, characterized in that, The total power scheduling record specifically includes the basic active power quota, virtual inertia base, and frequency regulation cycle timestamp. The energy storage node state deviation record includes the converter physical number, absolute deviation difference, and capacity polarity flag. The dynamic penalty weight record specifically includes the charging and discharging direction label, capacity over-limit judgment symbol, and weight amplification coefficient. The state micro-increment allocation record specifically includes the loss fusion index, objective function gradient scalar, and micro-increment evolution step size. The fast frequency regulation power allocation record includes the single-machine execution quota, communication message header, and underlying frequency regulation control word.

3. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 1, characterized in that, The specific steps for obtaining the total power scheduling record are as follows: S101: Monitor the actual speed and safety lower limit of the physical rotor under grid-connected conditions, perform exponentiation algebra calculations on the actual speed and safety lower limit respectively, obtain the square values ​​of the two, perform difference calculation on the square values ​​to generate physical square difference, obtain the inherent rotational inertia of the equipment, calculate the product fusion characteristics of physical square difference and rotational inertia, establish boundary limit base, and generate available kinetic energy storage. S102: Collect the synchronous frequency fluctuation rate of the power grid nodes, perform micro-evolution analysis of the synchronous frequency fluctuation rate in the time dimension, extract the wave frequency derivative characteristics, perform algebraic compensation calculation based on the wave frequency derivative characteristics and the available kinetic energy stock, obtain transient support distribution data in the kinetic energy conversion process, establish an inertial compensation scalar, perform mapping analysis on the power compensation coefficient based on the inertial compensation scalar, and generate virtual inertial increment; S103: Obtain the basic frequency regulation parameters of the scheduling node, perform feature identification on the numerical distribution and polarity attributes of the virtual inertia increment, extract the inertial support polarity identifier, combine the basic frequency regulation parameters and the virtual inertia increment to perform summation calculation, obtain the comprehensive adjustment quota, establish the power superposition base, perform data field format encapsulation processing on the power superposition base, and generate the total power scheduling record.

4. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 3, characterized in that, The specific steps for obtaining the energy storage node state deviation record are as follows: S201: During the evaluation phase, collect the real-time charge parameters and reference charge parameters of the energy storage converter, perform relative algebraic comparison calculations on the real-time charge parameters and reference charge parameters, obtain the deviation of the energy storage battery's underlying charge, establish a difference mapping relationship, and calculate the relative difference ratio of the capacity dimension based on the difference mapping relationship to generate the capacity deviation ratio. S202: The positive and negative polarity attributes of the capacity deviation ratio are numerically calibrated, and the absolute feature transformation mapping calculation is performed on the capacity deviation ratio in combination with the polarity attributes. The physical direction features of the value are stripped off, and the span data of the transformed capacity deviation ratio in the corresponding capacity range is extracted to generate the state deviation parameter. S203: Call the state deviation parameter and the real-time charge parameter, perform multi-dimensional data cross-splitting operation on the state deviation parameter and the real-time charge parameter, combine the correlation features of the underlying power state and the deviation magnitude, establish a multi-dimensional structure of the node energy storage state, encapsulate the multi-dimensional structure in a standard communication format, and generate an energy storage node state deviation record.

5. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 4, characterized in that, The specific steps for obtaining the dynamic penalty weight record are as follows: S301: Call the total power scheduling record under the scheduling condition, perform parsing operation on the power reversal feature of the total power scheduling record, extract the charging and discharging stage of the energy flow dimension, perform decomposition and extraction operation on the energy storage node state deviation record, peel off the underlying state deviation parameter and real-time charge parameter, and generate the energy storage basic parameter set. S302: Based on the energy storage basic parameter set, obtain the preset reference charge parameters, perform a relative capacity level comparison operation on the real-time charge parameters and the reference charge parameters, obtain the deviation status identifier of the capacity boundary, establish the charge over-limit state, call the charging and discharging stage, and perform multi-dimensional cross-recombination operation on the charge over-limit state and the charging and discharging stage to generate a reconstruction mapping matrix. S303: Based on the energy storage basic parameter set, call the state deviation parameter, obtain the preset penalty factor, combine the correlation features indicated by the reconstruction mapping matrix, perform a weighted fusion operation on the state deviation parameter and the penalty factor in terms of numerical dimensions, obtain the weight ratio base related to the adjustment cost, establish a penalty weight parameter, perform data domain communication format encapsulation processing on the penalty weight parameter, and generate a dynamic penalty weight record.

6. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 5, characterized in that, The process of obtaining the preset reference charge parameters is as follows: Read the initial upper limit threshold and initial lower limit threshold configured at the factory of the energy storage device, collect the cumulative charge and discharge cycle number data of the energy storage device, and perform interval reduction calculation on the initial upper limit threshold and initial lower limit threshold based on the cumulative charge and discharge cycle number data and the device capacity decay mapping table to generate dynamic scaling interval parameters, calculate the midpoint of the value of the dynamic scaling interval parameters, and generate the reference charge parameter. The process of obtaining the preset penalty factor is as follows: The frequency regulation urgency index data and the aging rate parameter of the energy storage device are collected and distributed. A numerical product operation is performed on the frequency regulation urgency index data and the aging rate parameter to obtain the initial deviation weight value. The upper limit penalty boundary value and the lower limit penalty boundary value configured in the device operation constraint specification are read. An interval limiting operation is performed on the initial deviation weight value based on the upper limit penalty boundary value and the lower limit penalty boundary value to obtain the limiting value and generate the penalty factor.

7. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 5, characterized in that, The specific steps for obtaining the state incremental allocation record are as follows: S401: Obtain the structural distribution parameters of equipment loss characteristics during the configuration phase; extract the penalty weight parameters for the dynamic penalty weight record; perform feature combination operation on the penalty weight parameters and structural distribution parameters to obtain the fusion base of the two; establish the cost basis of node collaboration; and generate total cost data. S402: Call the total cost data to obtain the corresponding power specification, extract the micro-evolution features of the mapping interval between the total cost data and the power specification, obtain the gradient evolution trend of the power interval, perform differential difference calculation based on the gradient evolution trend, extract the slope change rate of the change curve, establish the adjustment step size sequence of the power allocation node, and generate the correction micro-increment sequence. S403: Reconstruct and transform the underlying data type for the modified incremental sequence, identify the sequence arrangement characteristics of each element, adjust the internal data arrangement format based on the sequence arrangement characteristics, establish a structure-aligned fine-tuning allocation array, perform communication encapsulation processing on the data fields for the fine-tuning allocation array, and generate a state incremental allocation record.

8. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 7, characterized in that, The process of obtaining the structural distribution parameters of equipment loss characteristics during the configuration phase is specifically as follows: During the configuration phase, the charging and discharging current timing sequence and the operating voltage timing sequence of the energy storage device were collected. Time-domain product-integral operations are performed on the charging and discharging current timing sequence and the operating terminal voltage timing sequence to extract the total charging input energy data and the total discharging output energy data; Perform algebraic difference calculation on the total charging input energy data and the total discharging output energy data to obtain charge-discharge conversion loss data; Monitor the port static voltage drop and self-discharge current of the energy storage device in standby mode, and calculate the static standby loss parameters; Extract multiple operating power reference nodes corresponding to the configuration phase; The charge-discharge conversion loss data and the static standby loss parameters are mapped to the corresponding multiple operating power reference nodes, and a quadratic polynomial fitting operation is performed to obtain the combination of loss curve fitting coefficients. A multidimensional loss feature matrix is ​​constructed based on the combination of the loss curve fitting coefficients, and the structural distribution parameters are generated.

9. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 7, characterized in that, The specific steps for obtaining the fast frequency modulation power allocation record are as follows: S501: Obtain the running data of the issued working condition, call the state micro-increment allocation record, extract the internal correction micro-increment sequence, allocate the corresponding power test parameter based on the correction micro-increment sequence, perform approximation iterative operation on the power test parameter, calculate the error convergence during the iterative calculation process, establish the feature distribution mapping relationship based on the error convergence, and generate a gradient balance array. S502: Perform a horizontal comparison calculation on the internal element features of the gradient balancing array to obtain the distribution differences between multiple element features, read the constraints used for boundary filtering, perform a threshold comparison judgment on the distribution differences and constraints, filter the array features whose distribution differences meet the constraints, calculate the output quota value of the corresponding device in combination with the array features, and generate single-machine power configuration data. S503: Perform a low-level message parsing and reading operation on the single-unit power configuration data, extract the internal communication format, adjust the communication format according to the content layout requirements of the standard communication protocol, modify the message header and control field of the single-unit power configuration data, establish a format-aligned node scheduling instruction sequence, and generate a fast frequency modulation power allocation record.

10. The method for rapid frequency regulation power allocation of high-proportion wind power according to claim 9, characterized in that, The process of calculating the error convergence during the iterative calculation is specifically as follows: Extract the total demand power baseline data contained in the issued working condition, perform numerical summation operation on all allocated power test parameters to obtain the global test power accumulation value, calculate the absolute difference between the global test power accumulation value and the total demand power baseline data, extract the dynamic power deviation span and define it as the error convergence degree; The process of reading the constraints used for boundary filtering is as follows: Read the rated maximum output power extreme value, rated minimum output power extreme value, inherent physical ramp rate parameter and action response time constant of the corresponding device, perform a product operation on the inherent physical ramp rate parameter and the action response time constant to obtain the single maximum power jump amplitude, integrate the rated maximum output power extreme value, the rated minimum output power extreme value and the single maximum power jump amplitude to construct the operating boundary set, read the iteration tolerance lower limit of the algorithm configuration file, and define the operating boundary set and the iteration tolerance lower limit together as the constraint condition.

Images