A flexible load optimization adjustment method for a local power system

Abstract

The application relates to the technical field of power supply, in particular to a flexible load optimization adjustment method of a local power system, which comprises the following steps: constructing an improved ZIP model of various types of continuously adjustable flexible loads in each load node in the local power system; obtaining a quantitative evaluation result of output instability of new energy power of each load node; combining the improved ZIP model, the quantitative evaluation result of output instability of new energy power, and performing distribution adjustment on power distribution amounts of the various types of continuously adjustable flexible loads through a Shapley value method. The application aims to solve the defects that a traditional method ignores coordination of the various types of continuously adjustable flexible loads and instability analysis of new energy output, and significantly improves auxiliary support effects of flexible load adjustment on local power grid stability and frequency response.

Description

Technical Field

[0001] This application relates to the field of power supply technology, specifically to a method for flexible load optimization and adjustment of a local power system. Background Technology

[0002] With the accelerated global energy transition and the continuous advancement of the construction of new power systems dominated by new energy sources, the flexibility and stability of power systems are facing unprecedented challenges. Flexible loads, as a core adjustable resource on the demand side, can effectively ensure the safe and stable operation of high-proportion new energy power systems through refined and intelligent optimization and adjustment. This maximizes the utilization rate of green electricity, activates demand-side resources, and transforms traditional passive "consumers" of electricity into active "producers and consumers" who can participate in grid interaction, energy production, and consumption, thus flexibly constructing a new type of power system.

[0003] There are various types of adjustable flexible loads in power systems. Compared with non-discontinuously adjustable flexible loads, continuously adjustable flexible loads can better respond to small fluctuations in the output of new energy sources within a local power system. By continuously adjusting active and reactive power, they assist in the stability and frequency response of the local power grid. However, traditional methods for optimizing and adjusting flexible loads in local power systems usually only focus on fixed-frequency non-discontinuously adjustable flexible loads and ignore the coordination and cooperation among various types of continuously adjustable flexible loads. This leads to insufficient complementarity among continuously adjustable flexible loads and reduces the utilization rate of power system regulation resources. At the same time, the lack of analysis on the instability of new energy power output may weaken the auxiliary support effect of flexible loads on grid stability and frequency response. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a flexible load optimization and adjustment method for a local power system, employing the following technical solution: One embodiment of this application provides a method for flexible load optimization and adjustment of a local power system, the method comprising the following steps: An improved ZIP model is constructed for each type of continuously adjustable flexible load in each load node of a local power system. The improved ZIP model independently models each type of continuously adjustable flexible load and introduces a frequency-active power coupling correction term based on the voltage-power relationship. The improved ZIP model is used to output the power adjustability margin of each type of continuously adjustable flexible load in each load node at any time within the observation window. To obtain quantitative assessment results of the output instability of new energy power at each load node in the local power system; Based on the power adjustability margin output by the improved ZIP model and the quantitative evaluation results of the output instability, the power distribution of each type of continuously adjustable flexible load within each load node of the local power system under the observation window is allocated using the Shapley value method.

[0005] Preferably, the frequency-active power coupling correction term is constructed based on the deviation between the rated frequency and the actual frequency of each type of continuously adjustable flexible load and the corresponding frequency deviation sensitivity coefficient; the frequency deviation sensitivity coefficient represents the rate of change of active power of the continuously adjustable flexible load when the frequency deviates from the rated value by a unit percentage.

[0006] Preferably, the method further includes a dynamic adjustment coefficient for obtaining the frequency deviation sensitivity coefficient, the dynamic adjustment coefficient being used to adjust the frequency deviation sensitivity coefficient; the dynamic adjustment coefficient is calculated based on the response rate of each type of continuously adjustable flexible load in each load node within the observation window and the line transmission path distance between the load node and the main frequency modulation unit within the observation window.

[0007] Preferably, when obtaining the response rate, all abrupt changes in the sequence of power values ​​of each type of continuously adjustable flexible load within the observation window are first identified. After each power adjustment command is issued within the observation window, the abrupt change point with the shortest time interval among various types of continuously adjustable flexible loads is selected as each adjustment abrupt change point. Curve fitting is performed on the sequence of power values ​​between each adjustment abrupt point and the next adjustment abrupt point in each type of continuously adjustable flexible load within the observation window. The mean of the maximum slope of all fitted curves in each type of continuously adjustable flexible load within the observation window is calculated as the response rate of each type of continuously adjustable flexible load within the observation window.

[0008] Preferably, when obtaining the quantitative assessment results of the power output instability, the high-frequency mode components and low-frequency mode components of the total active power data of new energy power of each load node within the observation window are extracted; Calculate the product of the coefficient of variation of the total power generation of new energy sources at each load node within the observation window and the coefficient of variation of the absolute value of the differenced grid frequency data at the corresponding grid connection point. Based on the ratio of the amplitude of the high-frequency mode component to the amplitude of the low-frequency mode component and the result of the product, the power output fluctuation coefficient of each load node within the observation window is constructed as the quantitative assessment result of the power output instability of new energy power.

[0009] Preferably, before calculating the power output fluctuation coefficient of the new energy source, the data on the total active power of new energy sources, the total power generation of new energy sources, and the grid frequency data of the corresponding grid connection point of each load node within the observation window are normalized.

[0010] Preferably, before allocating the power distribution of each type of continuously adjustable flexible load within the observation window, the method further includes determining the total adjustment demand of each load node within the observation window based on the new energy output fluctuation coefficient, the sum of the rated active power of the new energy power generation units connected to the same bus at the load node under the standard test condition STC within the observation window, and the sign of the average difference between the actual output and the predicted output of the new energy in the observation window.

[0011] As a preferred method, when allocating the power distribution of various types of continuously adjustable flexible loads within each load node under the observation window using the Shapley value method: First, the Shapley quality factor of each type of continuously adjustable flexible load within the observation window is constructed using the Shapley value method. The Shapley quality factor is obtained by weighted calculation based on the normalized response speed, normalized adjustment accuracy, and normalized continuity of the minimum and maximum ramp rates of the corresponding type of continuously adjustable flexible load. Then, based on the power adjustability margin at all times within the observation window output by the improved ZIP model, determine the minimum adjustable power support capacity of each type of continuously adjustable flexible load within the observation window. Using the Shapley quality factor, the new energy output fluctuation coefficient, and the minimum adjustable power support capacity, the technical contribution of each type of continuously adjustable flexible load within each load node in the observation window is calculated; the technical contribution is the normalized sum of the technical contributions of all continuously adjustable flexible loads within all load nodes in the same observation window. Using the technical contribution of each type of continuously adjustable flexible load in the previous observation window of the observation window where each load node in the local power system is located at the current time as the allocation weight, the total power allocation obtained by the local power system at each time of the observation window where each load node in the local power system is located at the current time is linearly proportionally distributed, so as to allocate the power allocation of each type of continuously adjustable flexible load in each load node in the local power system at each time of the observation window where the current time is located.

[0012] Preferably, the minimum power support adjustable capacity is determined by the minimum power adjustable margin of each type of continuously adjustable flexible load within each load node at all times in the observation window.

[0013] Preferably, the improved ZIP model also includes constraints to construct a power margin assessment model for various types of continuously adjustable flexible loads in the load nodes. The types of constraints include: power system AC power flow constraints, power system generator ramp rate constraints, power system line current constraints, load node voltage and phase angle constraints, and power system generator output constraints, which serve as constraint boundaries for the coordinated allocation of power distribution to the various types of continuously adjustable flexible loads.

[0014] This application has at least the following beneficial effects: 1. Compared with traditional local power system flexible load optimization and adjustment, which usually only focuses on fixed-frequency non-continuously adjustable flexible loads, ignores the coordination between various types of continuously adjustable flexible loads, and lacks analysis of the instability of new energy power output, which may result in insufficient auxiliary support effect of flexible loads for grid stability and frequency response, this application analyzes the technical problems. It improves the ZIP model of various types of continuously adjustable flexible loads at load nodes, and conducts targeted modeling for existing user energy storage charging loads and temperature control loads, providing an accurate data foundation for subsequent power allocation of continuously adjustable flexible loads. 2. The output fluctuation coefficient of new energy power was obtained quantitatively, and the high-frequency disturbance degree and unstable fluctuation of active power at the grid connection of new energy power were accurately analyzed. This provides an important basis for the analysis of the instability of new energy power output in subsequent power allocation, and avoids the risk of the instability of new energy power aggravating the load response fluctuation of the power system. 3. By adjusting the power allocation of various types of continuously adjustable flexible loads using the Shapley value method, the power system can further improve the auxiliary support effect of flexible load adjustment on power system stability and frequency response, and ensure the safe operation of the power system, based on differentiating different types of continuously adjustable flexible loads for the absorption of new energy power in the power system. Attached Figure Description

[0015] Figure 1 A flowchart of a flexible load optimization and adjustment method for a local power system provided in this application. Detailed Implementation

[0016] This application provides an embodiment of a flexible load optimization and adjustment method for a local power system, which can be found in the following details. Figure 1 The method includes the following steps: Step 100: Construct an improved ZIP model for each type of continuously adjustable flexible load within each load node of the local power system.

[0017] At any given moment, for each load node in the local power system, the original user load ZIP model is as follows: In the formula, , These are the active power and reactive power of the i-th load node in the local power system, respectively. , These are the ground-state active power and ground-state reactive power of the i-th load node, respectively. , These are the rated voltage and actual voltage of the i-th load node, respectively. Let be the reactive power compensation capacity of the capacitor bank at the i-th load node; , , These are the active power parameters of the ZIP model (representing the proportions of constant impedance, constant current, and constant power components in the total active power, respectively). , , These are the reactive power parameters of the ZIP model (representing the proportions of constant impedance, constant current, and constant power components in the total reactive power, respectively).

[0018] In the construction of the aforementioned user load ZIP model, the voltage-active power relationship of the load is mainly considered, while the frequency-active power relationship is lacking. Furthermore, various types of continuously adjustable flexible loads are considered comprehensively without independent analysis of each type of continuously adjustable flexible load. When facing the task of optimizing and adjusting the flexible load of a local power system, the influence of the change in active power of continuously adjustable flexible load nodes with frequency fluctuations (i.e., load frequency characteristics) will form a natural frequency regulation effect, significantly affecting the dynamic process of system frequency. Therefore, it is necessary to analyze the correlation between frequency and active power.

[0019] For any given time point, the active power expression in the adjusted user load ZIP model is calculated as follows: In the above formula, the other indicators, compared to the original user load ZIP model, only refer to specific types of continuously adjustable flexible loads; the specific data types remain unchanged. That is: Let be the active power of the k-th continuously adjustable flexible load at the i-th load node in the local power system; Let be the ground-state active power of the k-th type of continuously adjustable flexible load at the i-th load node; , These represent the rated voltage and actual voltage of the k-th type of continuously adjustable flexible load at the i-th load node, respectively. Let be the frequency deviation sensitivity coefficient of the k-th continuously adjustable flexible load at the i-th load node in a local power system, representing the rate of change of active power of the k-th continuously adjustable flexible load at the i-th load node when the frequency deviates from the rated value by 1% (in percentage). ; The actual frequency of the k-th type of continuously adjustable flexible load at the i-th load node; The rated frequency of the k-th type of continuously adjustable flexible load at the i-th load node (the rated frequency of all load nodes in this scheme is constant at 50Hz).

[0020] Among them, This is denoted as the frequency-active power coupling correction term. The frequency-active power coupling correction term is constructed based on the deviation between the rated frequency and the actual frequency of each type of continuously adjustable flexible load and the corresponding frequency deviation sensitivity coefficient. It is used to introduce the frequency-active power coupling correction term on the basis of the voltage-power relationship to construct an improved ZIP model for each type of continuously adjustable flexible load at each load node in the local power system. The improved ZIP model independently models each type of continuously adjustable flexible load.

[0021] The revised ZIP model has two improvements over the original user load ZIP model: firstly, it distinguishes between different types of continuously adjustable loads; secondly, it adds a frequency-active power coupling correction term. Its physical significance lies in: when That is, when the frequency is normal, the frequency-active power coupling correction term is 1, and the adjusted ZIP model is still the original voltage-active power ZIP model; when Less than That is, when the generated electricity is insufficient, if If the value is greater than 0, the product correction term will be less than 1, the load will decrease automatically, and a positive frequency response will be provided to the power system, which is beneficial to the system frequency recovery. when Less than That is, when the generated electricity is insufficient, if If the product correction term is less than 0, it may be greater than 1. This could indicate an error in the setting of load adjustment control parameters at the load node of the local power system or a false triggering of the local voltage compensation logic. Conversely, when Greater than That is, when there is a surplus of electricity generation, if If the value is less than 0, the product correction term will be less than 1. In this case, the reduction in load will worsen the power surplus state and will not be conducive to absorbing the excess power. when Greater than That is, when there is a surplus of electricity generation, if If the value is greater than 0, the product correction term will be greater than 1. At this time, the load actively increases its power consumption, which can effectively absorb the excess power in the system and help suppress the frequency from rising further.

[0022] In the process of optimizing and adjusting flexible loads in local power systems, the proportion of active power change caused by unit frequency deviation alone cannot effectively characterize the active power-frequency sensitivity at each load node. Furthermore, it lacks analysis of the dynamic response characteristics of different types of continuously adjustable loads, and relies on static scalars. This could lead to a situation where one continuously adjustable flexible load has already responded at full capacity, while another continuously adjustable flexible load has not yet activated. At the same time, even in a local power system, there is still significant spatial heterogeneity among the load nodes, which can cause differences in the voltage changes of each node under the same frequency deviation, thus leading to inaccurate assessment of the active power of the load nodes in the ZIP model.

[0023] Accordingly, this application also includes a dynamic adjustment coefficient for obtaining the frequency deviation sensitivity coefficient, the dynamic adjustment coefficient being used to adjust the frequency deviation sensitivity coefficient; the dynamic adjustment coefficient is calculated based on the response rate of each type of continuously adjustable flexible load in each load node within the observation window and the line transmission path distance between the load node and the dominant frequency modulation unit.

[0024] In this scheme, the continuously adjustable flexible loads mainly include two categories: user energy storage charging loads and temperature control loads. In this embodiment, each 30-minute interval is used as an observation window. The data collection objects are the power values ​​of each type of continuously adjustable flexible load in each load node of the local power system within the observation window, the issuance time of each power adjustment command, and the line transmission path distance between each load node and the main frequency regulation unit. The minimum-maximum scaling method is used to normalize the power values ​​of all types of continuously adjustable flexible loads and the line transmission path distance between all load nodes and the main frequency regulation unit within each observation window. The minimum-maximum scaling method is well known to those skilled in the art, and the specific process will not be described in detail.

[0025] The data acquisition methods and equipment configurations for power values, power regulation command issuance time, and line transmission path distance between load nodes and the main frequency regulation unit are as follows: Power value data is obtained through smart meters, which are installed at the power supply circuit inlet of each type of continuously adjustable flexible load at each load node, using a series connection method, and the acquisition frequency is set to 1Hz; the power regulation command issuance time is obtained through the terminal log of the AGC automatic generation control system, and the acquisition frequency is set to 1Hz; through the distribution automation terminal FTU / DTU, the switch status is collected in real time along the distribution line and uploaded to the distribution automation master station SCADA for topology analysis, automatically tracking the connection path length between load nodes and power supply points, and the acquisition frequency is set to 1Hz.

[0026] Furthermore, this application first identifies all abrupt changes in the sequence of power values ​​of each type of continuously adjustable flexible load within the observation window (this embodiment identifies abrupt changes in the sequence using Bayesian abrupt change detection, which is well known to those skilled in the art, and the specific process will not be described in detail). After each power adjustment command is issued within the observation window, the abrupt change point with the shortest time interval among each type of continuously adjustable flexible load is selected as each adjustment abrupt change point. The sequence of all power values ​​between each adjustment abrupt change point and the next adjustment abrupt change point in each type of continuously adjustable flexible load within the observation window is used as input. A fitting curve is obtained using the least squares method. The average maximum slope of the fitting curve corresponding to all adjustment abrupt changes in each type of continuously adjustable flexible load within the observation window is calculated and recorded as the response rate of the corresponding type of continuously adjustable flexible load. A higher response rate indicates a higher rate of power change for the corresponding type of continuously adjustable flexible load within the observation window, and a faster response to the power system frequency. It should be noted that the sequences are constructed in chronological order.

[0027] It is worth noting that when the number of commands issued within the observation window is less than two, resulting in the inability to form a pair of adjustment mutation points, the current observation window directly inherits the response rate of the previous observation window.

[0028] Furthermore, this embodiment constructs the frequency deviation sensitivity coefficient of the i-th load node in the local power system for the k-th type of continuously adjustable flexible load within the j-th observation window. Dynamic adjustment coefficient The expression is as follows: In the above formula, The response rate of the i-th load node within the j-th observation window for the k-th type of continuously adjustable flexible load; The line transmission path distance between the i-th load node and the dominant frequency regulation unit within the j-th observation window; For normalization function, The normalization method adopts the min-max scaling method, and the normalization range is the total time range of the first U historically adjacent observation windows, with the j-th observation window as the final observation window. To accommodate both long-term and short-term time series trend analysis requirements, the value range is 10~20; in this scheme, 15 is used.

[0029] In the field of frequency-active power sensitivity analysis, there is a positive correlation between the response rate and adjustable capacity of continuously adjustable flexible loads. The higher the response rate, the more abundant the adjustable capacity of the continuously adjustable flexible load at the load node, and the faster and greater the actual power change caused by a unit frequency deviation. This indicates a higher coupling sensitivity between frequency and active power at the load node. Meanwhile, in the local power system topology, the farther the electrical distance between load nodes, the more significant the impact of power disturbances caused by frequency deviations on the load node voltage. Voltage changes can indirectly amplify the power response through the voltage terms (constant impedance, constant current, and constant power components) of the ZIP model, making the coupling correlation between frequency and active power at the load node stronger.

[0030] The frequency deviation sensitivity coefficient of the i-th load node in the local power system, within the j-th observation window, for the k-th type of continuously adjustable flexible load, is calculated as above. Dynamic adjustment coefficient of frequency deviation sensitivity coefficient The product of the sum of the constant d and the frequency deviation sensitivity coefficient is used as the new frequency deviation sensitivity coefficient. Let be the frequency-active power coupling correction term. Substitute it back into the adjusted user load ZIP model. In this scheme, the constant d is constant at 1.

[0031] To reduce the difficulty of solving subsequent problems, linearization is performed at any time within the j-th observation window at the local power system's ground-state operating point (the operating point under normal steady-state operating conditions of the power system, i.e., the rated / initial operating state after convergence in the power flow calculation), as follows: In the above formula, , , These are the changes in active power, frequency, and voltage of the k-th type of continuously adjustable flexible load at the i-th load node in the local power system (all three changes are changes at any time within the j-th observation window, and the frequency and voltage changes are the data differences between any time within the j-th observation window and the previous time). Let be the reactive power change of the k-th type of continuously adjustable flexible load at the i-th load node in the local power system. Let be the ground-state reactive power of the k-th type of continuously adjustable flexible load at the i-th load node; This represents the reactive power compensation capacity of the capacitor bank for the k-th type of continuously adjustable flexible load at the i-th load node; the remaining parameters remain consistent with the above formula.

[0032] Furthermore, constraints are set for the improved ZIP model to define the boundary conditions for the coordinated allocation of power distribution to various types of continuously adjustable flexible loads: (1) Given the active power, reactive power, phase angle, voltage change and frequency change of each type of continuously adjustable flexible load at the load node, construct the AC power flow constraint of the power system; given the active / reactive power of each type of continuously adjustable flexible load at the load node at each moment and the active / reactive power of the generator at the load node, obtain the active / reactive power change of each type of continuously adjustable flexible load at each load node at each moment. (2) Use the upper and lower limits of the ramp rate of generator sets in the local power system as the ramp rate constraint of generator sets in the power system; (3) The difference between the square of the ground state current flowing through the line at each moment and the square of the maximum allowable current flowing through the line and the square of the change in current flowing through the line at each moment shall be used as the line current constraint of the power system. (4) The minimum / maximum allowable voltage / phase angle of the load node are respectively used as the lower / upper limits of the voltage / phase angle base state value and the sum of voltage / phase angle changes at each moment of the load node, and are denoted as the load node voltage and phase angle constraints; (5) The minimum / maximum active / reactive output of the generators at the load node is used as the lower / upper limit constraint of the base state value of the active / reactive output of the generators at the load node at each moment and the sum of the changes in the active / reactive output of the generators at the load node, which is denoted as the generator output constraint of the power system. By improving the ZIP model with linearized user load and using all the constraints set above, a power margin assessment model for each type of continuously adjustable flexible load in the user load node is constructed. This yields the power adjustability margin of each type of continuously adjustable flexible load in each load node at any time under each observation window (i.e., the change in active power of each type of continuously adjustable flexible load in each load node at any time under each observation window). Since the power margin assessment model is a well-known technology, the specific acquisition process will not be elaborated further.

[0033] Step 200: Obtain the quantitative assessment results of the power output instability of new energy power at each load node.

[0034] In the process of optimizing and adjusting the power allocation of various types of continuously adjustable flexible loads in a local power system, although continuously adjustable loads have a rapid response capability, their adjustment domain is constrained by state variables such as SOC / temperature. When the prediction error of renewable energy output is large, a mismatch may occur between the load adjustment command and the actual output, leading to malfunction or premature depletion of adjustment resources. At the same time, a sudden drop in renewable energy output will exacerbate the frequency fluctuation of the power system. Continuously adjustable flexible loads need to make timely predictions based on output stability to avoid delayed response.

[0035] Based on the above analysis, this scheme constructs a new energy power output fluctuation coefficient in the following manner to reflect the instability and unpredictability of new energy power output in the local power system, and is used to construct a quantitative assessment result of the power output instability of new energy.

[0036] The data collection objects for analyzing the fluctuation coefficient of new energy power output are: total new energy power generation data, total new energy power active power data, and grid frequency data at the new energy power grid connection point. The data collection method and equipment configuration are as follows: the total new energy power generation data (the sum of wind power generation and photovoltaic power generation) and the total new energy power active power data are obtained in real time through smart meters deployed at the new energy power grid connection point or inverter outlet. The grid frequency data at the grid connection point is obtained through the PMU synchronous phasor measurement unit deployed at the new energy power plant grid connection point. The data collection frequency is set to 1KHz. The entropy weight method is used to normalize all the total new energy power generation data, all the total new energy power active power data, and all the grid frequency data at the grid connection point obtained in each observation window. The specific method of entropy weight method normalization is known to those skilled in the art.

[0037] When obtaining the quantitative assessment results of the power output instability, the high-frequency and low-frequency mode components of the total active power data of new energy power at each load node within the observation window are first extracted. In this embodiment, the specific method is as follows: The target time series is sampled using an equal-interval sampling method. The sampled active power is then arranged in the order of acquisition time to form a reduced-order sequence of the total active power of new energy power in the observation window. Using this reduced-order sequence as input, the EMD (empirical mode decomposition) algorithm is used to obtain all mode components in the reduced-order sequence of the total active power of new energy power. Since the mode components after EMD are arranged from high to low frequency, the first and last mode components are selected as the high / low-frequency mode components of the total active power of new energy power at each load node within the observation window. The equal-interval sampling method and the EMD algorithm are well-known to those skilled in the art, and their specific processes will not be described in detail.

[0038] Furthermore, the product of the coefficient of variation of the total renewable energy power generation data of each load node within the observation window and the coefficient of variation of the absolute value of the difference of the grid frequency data at the corresponding grid connection point is calculated; based on the ratio of the amplitude of the high-frequency mode component to the amplitude of the low-frequency mode component and the result of the product, the renewable energy output fluctuation coefficient of each load node within the observation window is constructed as the quantitative assessment result of the renewable energy power output instability.

[0039] Specifically, in this embodiment, the new energy power output fluctuation coefficient is constructed in the following way: In the above formula, Let be the power output fluctuation coefficient of the new energy source at the i-th load node in the j-th observation window in the local power system; It is the ratio of the mean amplitude of all high-frequency mode components to the mean amplitude of all low-frequency mode components of the total active power of new energy power of the i-th load node in the j-th observation window; It is the product of the coefficient of variation of the total power generation of new energy sources at the i-th load node in the j-th observation window, arranged in chronological order, and the coefficient of variation of the absolute value of the first difference of the corresponding grid frequency at the grid connection point, arranged in chronological order. For normalization function, The normalization method adopts the minimum-maximum scaling method, and the normalization range is the total time range of the first U historically adjacent observation windows, with the j-th observation window as the final observation window.

[0040] In the field of active power analysis technology for power systems, there is a positive correlation between the proportion of high-frequency components of active power in the frequency domain and the rapid changes in active power disturbances. When the active power of load nodes connected to renewable energy exhibits faster, shorter-period, and steeper characteristics of power disturbances, the proportion of high-frequency components in the active power data is higher, and the ratio between the mean amplitudes of all amplitudes in the high-frequency mode components and the low-frequency mode components is larger. Meanwhile, in the field of data analysis, a high coefficient of variation indicates a higher degree of dispersion in the data sequence. Therefore, this scheme reflects the fluctuations in power generation and grid frequency at the renewable energy grid connection points of each load node by calculating the product of the coefficients of variation of the absolute values ​​of the first-order difference of the absolute values ​​of the renewable energy generation and the grid frequency at the grid connection point within the observation window.

[0041] Step 300: Combining the improved ZIP model and the quantitative assessment results of the output instability of new energy power, the power allocation of various types of continuously adjustable flexible loads is adjusted using the Shapley value method.

[0042] A Shapley participant set is constructed for all continuously adjustable flexible loads at all load nodes in the local power system (in this scheme, the user energy storage charging load and temperature control load are optimized and adjusted).

[0043] First, before allocating the power distribution of each type of continuously adjustable flexible load within the observation window at each load node, the total adjustment demand of each load node within the observation window is determined based on the renewable energy output fluctuation coefficient obtained in step 200, the sum of the rated active power of renewable energy power generation units connected to the same bus at the load node under standard test conditions (STC) within the observation window, and the sign of the average difference between the actual and predicted renewable energy output within the observation window. In the above formula, The total regulation demand of the i-th load node in the local power system during the j-th observation window; Let be the power output fluctuation coefficient of the new energy source at the i-th load node in the j-th observation window in the local power system; It is the sum of the rated active power of the new energy power generation units connected to the same bus at the i-th load node under the standard test conditions (STC); The prediction deviation of the i-th load node in the local power system during the j-th observation window is given (this can be calculated by averaging the differences between the actual and predicted output of renewable energy at all times within the j-th observation window; the predicted output of renewable energy is achieved using the EMA exponential moving average algorithm, and since EMA is a well-known technology, the specific acquisition process will not be elaborated further); sgn is the sign function, when... When greater than 0, The value is 1 (the load needs to be increased to absorb excess renewable energy). When less than 0, The value is -1 (the load needs to be reduced to fill the gap); when When the value equals 0, it indicates that the prediction is accurate and there is no need for adjustment.

[0044] The power margin assessment model constructed using the improved ZIP model and the set constraints in step 100 can obtain the power adjustability margin of each type of continuously adjustable flexible load within each load node of the local power system at any time within each observation window. Since adjusting the total power allocation at the time-level involves a large computational burden and is prone to causing abnormal power fluctuations, this application, for each observation window, uses the minimum power adjustability margin of each type of continuously adjustable flexible load within each load node of the local power system at all times within each observation window as the minimum adjustable power capacity supported by each type of continuously adjustable flexible load for each load node in the local power system within each observation window.

[0045] When allocating the power distribution of various types of continuously adjustable flexible loads within each load node under the observation window, this application first constructs the Shapley quality factor of each type of continuously adjustable flexible load within each load node under the observation window using the Shapley value method. This factor characterizes the output contribution value of each type of continuously adjustable flexible load within each load node in each observation window. The Shapley quality factor is obtained by weighting the normalized response speed, normalized adjustment accuracy, and normalized continuity of the minimum and maximum ramp rates of the corresponding type of continuously adjustable flexible load. In this embodiment, the specific calculation method of the Shapley quality factor is as follows: In the above formula, Shapley quality factor for the k-th type of continuously adjustable flexible load within the i-th load node of a local power system during the j-th observation window; , , These are the normalized response speed, normalized adjustment accuracy, and normalized continuity of the minimum and maximum ramp rates of the k-th type of continuously adjustable flexible load within the i-th load node of the local power system in the j-th observation window (wherein, the response speed, adjustment accuracy, and continuity of the minimum and maximum ramp rates can be obtained in real time using known techniques). For the normalization operation of each of the three indicators, this application normalizes all continuously adjustable flexible load types within the same observation window for the same load node. In this scheme, continuously adjustable flexible loads are only for user energy storage charging loads and temperature control loads. This scheme uses the min-max scaling method to normalize the above indicators to eliminate physical dimensions, and uniformly maps the response speed, adjustment accuracy, and continuity to the same value space, thereby preserving the initial performance advantages and disadvantages of each type of flexible load under the same dimension. , , These are the quality weights, and the sum of the three is 1. The initial values ​​are set to 0.4, 0.3, and 0.3 by default (implementers can set them according to their actual situation). Since the calculation of Shapley values ​​is a technique known to those skilled in the art, the specific method of obtaining them will not be elaborated further.

[0046] Furthermore, using the Shapley quality factor, the renewable energy output fluctuation coefficient, and the minimum adjustable capacity of power support, the technical contribution of each type of continuously adjustable flexible load within the observation window is calculated. The specific calculation method is as follows: In the above formula, The technical contribution of the i-th load node in the local power system to the k-th type of continuously adjustable flexible load in the j-th observation window; It is a minimum value function; For normalization function, The normalization method adopts the sum-of-the-parts normalization method, and the normalization range is the technical contribution of all continuously adjustable flexible loads in all load nodes under the j-th observation window; The minimum adjustable capacity of the power support for the k-th continuously adjustable flexible load at the i-th load node in the j-th observation window of the local power system; This contributes to the effective absorption of the k-th type of continuously adjustable flexible load by the i-th load node in the j-th observation window of the local power system.

[0047] Furthermore, this application constructs a technology contribution score, organically integrating effective absorption contribution, comprehensive regulation quality (response speed, regulation accuracy, and continuity) based on the Shapley quality factor, and the renewable energy output fluctuation coefficient. This enables a refined and quantitative assessment of the regulation value of various types of continuously adjustable flexible loads within each load node. This technology contribution score effectively identifies high-quality flexible load resources with fast response speeds, high regulation accuracy, and strong continuous regulation capabilities, and assigns them higher priority in power allocation, avoiding the problem of inefficient use of high-quality regulation resources or delayed response of inferior regulation resources under traditional average allocation methods. Simultaneously, by introducing the renewable energy output fluctuation coefficient for weighting, flexible loads with rapid response capabilities in high-volatility scenarios receive greater allocation priority, enhancing the dynamic adaptability of the local power system to the uncertainty of renewable energy output.

[0048] Based on this technical contribution, this application further utilizes the technical contribution of each type of continuously adjustable flexible load in the previous observation window of the current observation window of each load node in the local power system as the allocation weight. The total power allocation obtained by the local power system at each moment in the current observation window of each load node is linearly distributed proportionally. This is to allocate the power allocation of each type of continuously adjustable flexible load in the local power system at each moment in the current observation window, so as to realize a precise allocation mechanism for on-demand allocation among flexible loads.

[0049] Specifically, this scheme first obtains the technical contribution of each type of continuously adjustable flexible load in the previous observation window of the observation window where each load node in the local power system is located at the current time. Based on the obtained technical contribution, the technical contribution of each type of continuously adjustable flexible load in the previous observation window of the observation window where each load node in the local power system is located at the current time is used as the allocation weight. The allocation weight is multiplied by the total power allocation obtained by the local power system at each time in the observation window where the current time is located to obtain the power allocation of each type of continuously adjustable load in each load node at each time in the observation window where the current time is located.

[0050] The total power allocation (total thermal power generation and renewable energy generation) can be obtained through the Automatic Generation Control (AGC) system in the local power system. The acquisition process is well known to those skilled in the art, and the specific process will not be described in detail.

[0051] Thus, a flexible load optimization and adjustment method for a local power system can be realized through the above approach.

[0052] It is worth noting that, in the calculations using the formula in this application, if the denominator is 0, a positive number needs to be added to the denominator. (like (The implementer can also choose a value according to the dimensions) to prevent the denominator from being zero.

Claims

1. A method for flexible load optimization and adjustment in a local power system, characterized in that, The method includes the following steps: An improved ZIP model is constructed for each type of continuously adjustable flexible load in each load node of a local power system. The improved ZIP model independently models each type of continuously adjustable flexible load and introduces a frequency-active power coupling correction term based on the voltage-power relationship. The improved ZIP model is used to output the power adjustability margin of each type of continuously adjustable flexible load in each load node at any time within the observation window. To obtain quantitative assessment results of the output instability of new energy power at each load node in the local power system; Based on the power adjustability margin output by the improved ZIP model and the quantitative evaluation results of the output instability, the power distribution of each type of continuously adjustable flexible load within each load node of the local power system under the observation window is allocated using the Shapley value method.

2. The flexible load optimization and adjustment method for a local power system as described in claim 1, characterized in that, The frequency-active power coupling correction term is constructed based on the deviation between the rated frequency and the actual frequency of each type of continuously adjustable flexible load and the corresponding frequency deviation sensitivity coefficient. The frequency deviation sensitivity coefficient represents the rate of change of active power of this type of continuously adjustable flexible load when the frequency deviates from the rated value by a unit percentage.

3. The flexible load optimization and adjustment method for a local power system as described in claim 2, characterized in that, It also includes a dynamic adjustment coefficient for obtaining the frequency deviation sensitivity coefficient, the dynamic adjustment coefficient being used to adjust the frequency deviation sensitivity coefficient; the dynamic adjustment coefficient is calculated based on the response rate of each type of continuously adjustable flexible load in each load node within the observation window and the line transmission path distance between the load node and the main frequency modulation unit within the observation window.

4. The flexible load optimization and adjustment method for a local power system as described in claim 3, characterized in that, When obtaining the response rate, first identify all abrupt changes in the sequence of power values ​​of each type of continuously adjustable flexible load within the observation window; After each power adjustment command is issued within the observation window, the abrupt change point with the shortest time interval among various types of continuously adjustable flexible loads is selected as each adjustment abrupt change point. Curve fitting is performed on the sequence of power values ​​between each adjustment abrupt point and the next adjustment abrupt point in each type of continuously adjustable flexible load within the observation window. The mean of the maximum slope of all fitted curves in each type of continuously adjustable flexible load within the observation window is calculated as the response rate of each type of continuously adjustable flexible load within the observation window.

5. The flexible load optimization and adjustment method for a local power system as described in claim 1, characterized in that, When obtaining the quantitative assessment results of the power output instability, the high-frequency mode components and low-frequency mode components of the total active power data of new energy power of each load node within the observation window are extracted; Calculate the product of the coefficient of variation of the total power generation of new energy sources at each load node within the observation window and the coefficient of variation of the absolute value of the differenced grid frequency data at the corresponding grid connection point. Based on the ratio of the amplitude of the high-frequency mode component to the amplitude of the low-frequency mode component and the result of the product, the power output fluctuation coefficient of each load node within the observation window is constructed as the quantitative assessment result of the power output instability of new energy power.

6. The flexible load optimization and adjustment method for a local power system as described in claim 5, characterized in that, Before calculating the new energy output fluctuation coefficient, the method further comprises normalizing the sum of active power data of new energy power, total power generation data of new energy power and corresponding grid-connected point grid frequency data of each load node within the observation window.

7. The flexible load optimization and adjustment method for a local power system as described in claim 5, characterized in that, Before distributing the power distribution amount of each type of continuously adjustable flexible load in the observation window of each load node, the method further comprises determining the total adjustment demand of each load node within the observation window according to the new energy output fluctuation coefficient, the sum of rated active power of new energy power generation units connected to the same bus of the load node under standard test conditions (STC) within the observation window, and the sign of the average difference between the actual output and the predicted output of the new energy within the observation window.

8. The flexible load optimization and adjustment method for a local power system as described in claim 7, characterized in that, When distributing the power distribution amount of each type of continuously adjustable flexible load in the observation window of each load node by the Shapley value method: First, the Shapley value method is used to construct the Shapley quality factor of each type of continuously adjustable flexible load in the observation window of each load node, which is calculated by weighting the normalized response speed, normalized adjustment accuracy and normalized continuity of the minimum ramp rate and maximum ramp rate of the corresponding type of continuously adjustable flexible load; Then, the power minimum support adjustable capacity of each type of continuously adjustable flexible load in the observation window is determined according to the power adjustable margin output by the improved ZIP model at all times within the observation window; The technical contribution degree of each type of continuously adjustable flexible load in the observation window of each load node is calculated using the Shapley quality factor, the new energy output fluctuation coefficient and the power minimum support adjustable capacity; the technical contribution degree is the normalized result of the sum of the technical contribution degrees of all continuously adjustable flexible loads in all load nodes within the same observation window; The total power distribution amount obtained by the local power system at each time within the observation window of the current time is linearly proportionally distributed by using the technical contribution degree of each type of continuously adjustable flexible load in the previous observation window of the current time as the distribution weight, so as to distribute the power distribution amount of each type of continuously adjustable flexible load in each time within the observation window of the current time in the local power system.

9. The method of claim 8, wherein, The power minimum support adjustable capacity is determined by the minimum value of the power adjustable margin of each type of continuously adjustable flexible load in all times within the observation window of each load node.

10. The method of claim 9, wherein, The improved ZIP model is also provided with a constraint condition for constructing a power margin evaluation model of each type of continuously adjustable flexible load in the load node; the types of the constraint condition include power system alternating current flow constraint, power system generator set ramp rate constraint, power system line current constraint, load node voltage and phase angle constraint, and power system generator output constraint, which are used as the constraint boundary for coordinating and distributing the power distribution amount of each type of continuously adjustable flexible load.

Images