Power distribution network distributed resource participation multi-element power auxiliary service scheduling method, system, equipment and medium

Abstract

The invention relates to the technical field of distributed resource scheduling, and discloses a power distribution network distributed resource participation multi-element power auxiliary service scheduling method, system, device and medium, and the method comprises the steps: obtaining distributed resource real-time operation data, carrying out the preprocessing, removing missing values and abnormal values, enabling the data to be accurate and comparable through the normalization processing, and obtaining the real-time operation data of the distributed resource; and a foundation is laid for subsequent classification and evaluation. During granulation classification, factors such as response time, precision and sustainable time are comprehensively considered, resources are scientifically classified, and resource characteristics are accurately grasped. The resources are evaluated based on the exclusive indexes of peak regulation and frequency modulation, and are reasonably divided into peak regulation, frequency modulation and standby classes, so that the resource allocation better meets the actual demand. A multi-objective optimization model is constructed, the minimum total scheduling cost, the maximum total performance weight and the scheduling amount weighted sum are taken into consideration, and efficient resource scheduling is achieved under the constraint that peak and frequency regulation requirements, resource capacity, scheduling amount precision and the like are met.

Description

Technical Field

[0001] This invention relates to the field of distributed resource scheduling technology, and in particular to a method, system, device and medium for distributed resources in power distribution networks to participate in the scheduling of multiple power auxiliary services. Background Technology

[0002] In the journey of building a new power system, a large number of diverse and heterogeneous resources, such as distributed power sources (e.g., photovoltaic power generation, wind power generation), energy storage devices, and controllable loads, have entered the distribution network, completely overturning the traditional operation mode and control logic of the distribution network.

[0003] Taking distributed power sources as an example, photovoltaic power generation is highly susceptible to interference from natural factors such as sunlight intensity and duration, exhibiting significant intermittency and volatility. Wind power output, on the other hand, is closely dependent on changes in wind speed and direction, exhibiting high randomness and unpredictability. While energy storage devices possess rapid response capabilities, they are limited by their own capacity and lifespan constraints based on the number of charge-discharge cycles. Furthermore, the adjustment potential of controllable loads is closely linked to users' electricity consumption habits and production processes, resulting in significant variations in their adjustment flexibility and response speed. These diverse resources, with their vast differences in output characteristics, response speed, and capacity, make the operation of the distribution network extremely complex, posing an unprecedented and severe challenge to its peak-shaving and frequency-regulating capabilities.

[0004] Meanwhile, the gradual opening and deepening reform of the electricity market have placed unprecedented demands on the dispatch of distribution network resources. With the emergence of new trading mechanisms such as the electricity spot market and ancillary services market, distribution network resources must actively participate in ancillary services such as peak shaving, frequency regulation, and reserve to cope with grid power fluctuations caused by the large-scale integration of renewable energy in my country, ensuring real-time balance between power supply and demand, and stabilizing power quality. However, it is undeniable that the current development of distribution network resource dispatch technology has clearly lagged behind actual needs, revealing numerous shortcomings in key areas such as resource allocation, indicator setting, and market transactions. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides a method, system, equipment, and medium for the dispatch of distributed resources in a distribution network to participate in multi-functional power auxiliary services. This invention can solve the problems existing in the current distribution network resource dispatch technology in key aspects such as resource allocation, indicator setting, and market transactions, and achieve more efficient and reasonable dispatch of distributed resources in the distribution network.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a method for dispatching distributed resources in a power distribution network to participate in multi-faceted power ancillary services, comprising:

[0009] Acquire real-time operational data of several types of distributed resource data in the target power grid, and perform a first preprocessing on the real-time operational data;

[0010] The preprocessed real-time running data is granularized and classified to obtain a set of distributed resource granules with different characteristics;

[0011] Preset peak shaving and frequency regulation specific indicators, evaluate the distributed resources in the distributed resource granular set based on the peak shaving and frequency regulation specific indicators, and divide the distributed resources into peak shaving resources, frequency regulation resources and reserve resources;

[0012] Obtain the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching and scheduling principle, and determine the priority of peak-shaving and frequency regulation resources;

[0013] A multi-objective optimization model is constructed to determine the scheduling volume of various types of distributed resource data, and to schedule distributed resources in the power ancillary services market.

[0014] As a preferred embodiment of the method for dispatching distributed resources in a power distribution network with diversified power ancillary services according to the present invention, the first preprocessing of the real-time operating data includes:

[0015] The real-time running data is processed for missing values ​​and outliers are removed to obtain the first cleaned data;

[0016] The first cleaned data is normalized using the max-min normalization method to obtain the first normalized dataset;

[0017] Missing values ​​are filled using linear interpolation or by filling in the missing values ​​based on the mean of historical data.

[0018] As a preferred embodiment of the method for dispatching distributed resources in a power distribution network participating in multi-functional power auxiliary services as described in this invention, the granular classification of the first preprocessed real-time operating data includes:

[0019] The first feature vector is constructed based on the response time, response accuracy, and response duration of each distributed resource;

[0020] The first feature vector is mapped to the hypersphere feature space, and Mahalanobis distance is used to measure the similarity between samples;

[0021] The idea of ​​density peak clustering is introduced to optimize the selection of initial cluster centers, and iterative clustering is performed by combining the improved K-means+ algorithm to obtain multiple sets of first resource particles.

[0022] This preferred scheme enables a more scientific and accurate classification of distributed resources. By constructing a first feature vector, key factors such as response time, response accuracy, and response duration of distributed resources are fully considered. These factors directly affect the performance and role of resources in power ancillary services, allowing the classification to closely align with the actual characteristics of the resources. Mapping the first feature vector to a hypersphere feature space and using Mahalanobis distance to measure the similarity between samples effectively handles the correlation and distribution characteristics of the data, avoiding the limitations of traditional distance measurement methods and improving the accuracy and reliability of the classification.

[0023] As a preferred embodiment of the method for dispatching distributed resources in a power distribution network to participate in multi-power auxiliary services as described in this invention, the response time includes instruction delay time, startup time, rise time, and settling time.

[0024] The response accuracy is calculated using the error rate between the actual response power and the command power.

[0025] The duration of the response is dynamically assessed based on available energy, regulated power, and charge / discharge efficiency factor.

[0026] As a preferred embodiment of the method for dispatching distributed resources in a distribution network to participate in multiple power auxiliary services as described in this invention, wherein: the peak-shaving-specific indicators include peak-shaving capacity and peak-shaving rate, and the frequency regulation-specific indicators include frequency response coefficient and frequency regulation capacity margin;

[0027] Among them, peak shaving capacity represents the total amount of active power that can be adjusted within the scheduling period, and peak shaving rate represents the change in active power per unit time.

[0028] The frequency response coefficient is the ratio of the change in active power to the system frequency deviation, and the frequency regulation capacity margin is the available frequency regulation power reserve after the resources meet the current operating requirements.

[0029] As a preferred embodiment of the method for dispatching distributed resources in a distribution network to participate in multiple power ancillary services as described in this invention, the evaluation of distributed resources based on peak-shaving and frequency-regulation specific indicators includes:

[0030] Construct a first evaluation matrix, where rows correspond to each distributed resource and columns correspond to each evaluation indicator;

[0031] The first weight vector of each evaluation indicator was determined using the analytic hierarchy process.

[0032] Calculate the first comprehensive score of each distributed resource, and divide the resources based on a preset first threshold and a second threshold, wherein the second threshold is greater than the first threshold;

[0033] Specifically, when the first comprehensive score is greater than or equal to the second threshold, it is classified as a frequency modulation resource; when it is less than the second threshold but greater than or equal to the first threshold, it is classified as a peak shaving resource; otherwise, it is classified as a reserve resource.

[0034] As a preferred embodiment of the method for dispatching distributed resources in a power distribution network with multiple power ancillary services as described in this invention, the construction of the multi-objective optimization model includes:

[0035] The first optimization objective is to minimize the total scheduling cost, and the second optimization objective is to maximize the weighted sum of the total performance weight and the scheduling amount.

[0036] A first set of constraints is set, which includes: peak shaving and frequency regulation demand satisfaction constraints, resource capacity constraints, and scheduling quantity accuracy constraints.

[0037] Among them, the minimum scheduling unit for peak shaving service is defined as the first unit value, and the minimum scheduling unit for frequency regulation service is defined as the second unit value in the scheduling quantity precision constraint.

[0038] Secondly, the present invention provides a system for dispatching distributed resources in a power distribution network to participate in multi-functional power auxiliary services, comprising:

[0039] The data acquisition and processing module is used to acquire real-time operational data of several types of distributed resource data in the target power grid, and to perform a first preprocessing on the real-time operational data;

[0040] The classification set acquisition module is used to granularize the first preprocessed real-time running data to obtain a distributed resource granular set with different characteristics;

[0041] The partitioning module is used to preset peak shaving and frequency regulation specific indicators, evaluate the distributed resources in the distributed resource granular set based on the peak shaving and frequency regulation specific indicators, and partition the distributed resources into peak shaving resources, frequency regulation resources and reserve resources.

[0042] The priority acquisition module is used to acquire the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching scheduling principle, and determine the priority of peak-shaving and frequency regulation resources.

[0043] The scheduling module is used to build a multi-objective optimization model, determine the scheduling volume of various types of distributed resource data, and perform scheduling of distributed resources in the power ancillary services market.

[0044] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0045] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0046] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a method for dispatching distributed resources in a power distribution network using multi-functional power auxiliary services. By acquiring and preprocessing real-time operational data of distributed resources, missing and outlier values ​​are removed, making the data more accurate and reliable. Normalization also makes different types of data comparable, laying a solid foundation for subsequent classification and evaluation. In the granular classification process, key factors such as response time, response accuracy, and response duration of distributed resources are comprehensively considered. A scientific method is used to classify resources, enabling a more accurate grasp of the characteristics of various resource types. Distributed resources are evaluated based on peak-shaving and frequency-regulation specific indicators, and resources are reasonably divided into peak-shaving, frequency-regulation, and reserve categories, making resource allocation more in line with actual needs. A multi-objective optimization model is constructed, simultaneously considering minimizing the total dispatch cost and maximizing the weighted sum of total performance weights and dispatch quantities. Under the constraints of peak-shaving and frequency-regulation requirements, resource capacity, and dispatch quantity accuracy, efficient resource dispatch is achieved. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart illustrating a method for dispatching distributed resources in a power distribution network to participate in multi-functional power auxiliary services, as provided in one embodiment of the present invention.

[0049] Figure 2 This is an internal structure diagram of an electronic device for a method of scheduling distributed resources in a power distribution network for participating in multi-power auxiliary services, as provided in an embodiment of the present invention. Detailed Implementation

[0050] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0051] Example 1, referring to Figure 1This is the first embodiment of the present invention, which provides a method for dispatching distributed resources in a power distribution network to participate in multi-faceted power ancillary services, including:

[0052] Existing technologies have several drawbacks. For example, the lack of a unified and effective standard for the scheduling of distributed resources in power distribution networks makes it difficult for different types of resources to achieve efficient coordination when participating in diverse power ancillary services. The control strategies for each distributed resource are often formulated independently, without fully considering compatibility with other resources and the entire power distribution network system. This makes resource waste or scheduling conflicts prone to occur during actual scheduling.

[0053] Furthermore, there is insufficient real-time monitoring and evaluation capability for distributed resources. In diversified power ancillary services, it is necessary to accurately grasp the real-time status and performance of distributed resources, but existing monitoring technologies and methods have certain limitations, making it impossible to obtain key information about resources in a timely and accurate manner, thus affecting the scientific nature and timeliness of dispatch decisions.

[0054] Moreover, existing technologies have significant shortcomings in addressing the complex and ever-changing electricity market environment and user demands. As the electricity market continues to develop and users' requirements for electricity service quality increase, distributed resources in the distribution network need to participate in diversified electricity ancillary services more flexibly and intelligently. However, current dispatching methods are unable to adapt quickly to these changes and cannot adequately meet the diverse needs of the market and users.

[0055] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to realize the method of dispatching distributed resources of the power distribution network to participate in multiple power auxiliary services in conjunction with several embodiments.

[0056] Figure 1 A flowchart illustrating a method for dispatching distributed resources in a power distribution network using multiple power ancillary services is shown, including:

[0057] S101, acquire real-time operational data of several types of distributed resource data in the target power grid, and perform a first preprocessing on the real-time operational data, wherein:

[0058] It should be noted that, in order to enable distributed resources in the distribution network to participate in the dispatch of diversified power ancillary services, real-time operational data of distributed resources can be collected from the target power grid. This distributed resource data includes, but is not limited to, real-time operational data of distributed power sources (such as solar photovoltaic and wind power), energy storage devices (such as battery energy storage), and controllable loads, because these resources play different roles in the distribution network, and the accurate acquisition of their operational data is crucial for subsequent dispatch.

[0059] In one optional implementation, after acquiring real-time operational data of several types of distributed resources in the target power grid, preprocessing operations can be performed on the real-time operational data. Preprocessing operations may include cleaning and normalizing the real-time operational data. During the cleaning process, missing values ​​in the data are filled using linear interpolation or by imputing the mean of historical data to ensure data integrity. Outliers are removed by setting a reasonable threshold range to avoid interference from abnormal data in subsequent analysis.

[0060] In one optional implementation, the real-time operational data undergoes initial preprocessing, starting with missing value handling and outlier removal. During actual data acquisition, data gaps may occur due to sensor malfunctions, communication interruptions, or other reasons. Using linear interpolation or imputation based on the mean of historical data to fill in missing values ​​makes the data more complete. Simultaneously, some outliers may exist in the data, possibly due to external interference or measurement errors. Removing these outliers prevents them from adversely affecting subsequent analysis and scheduling. Then, the cleaned data is normalized using a max-min normalization method, unifying different types of data into a comparable range. This eliminates the influence of data units and orders of magnitude, providing a more accurate data foundation for subsequent granular classification and evaluation.

[0061] In this embodiment of the invention, the first preprocessing of real-time running data includes:

[0062] Missing values ​​and outliers are processed and removed from the real-time running data to obtain the first cleaned data;

[0063] The first cleaned data is normalized using the max-min normalization method to obtain the first normalized dataset;

[0064] Missing values ​​are filled using linear interpolation or by filling in the missing values ​​based on the mean of historical data.

[0065] It should be noted that acquiring real-time operational data of several types of distributed resources in the target power grid and performing initial preprocessing on this data can improve data quality and availability, providing a solid foundation for subsequent resource performance assessment, classification confirmation, and scheduling decisions. Preprocessed data is more complete and accurate, more realistically reflecting the operational status of distributed resources, thus leading to more precise resource performance assessment. Accurate assessment results help to rationally classify different types of distributed resources and clarify the characteristics and advantages of each type.

[0066] S102, the real-time running data after the first preprocessing is granularized and classified to obtain a set of distributed resource granules with different characteristics, wherein:

[0067] It should be noted that after obtaining the real-time operational data after the first preprocessing, this data needs to be granularized and classified to obtain a set of distributed resource granules with different characteristics. This step plays a crucial role in the entire method for dispatching distributed resources in a multi-faceted power ancillary services within the distribution network. Based on the accurate and comparable data obtained after the first preprocessing, it further explores the intrinsic characteristics of distributed resources.

[0068] In one alternative implementation, granular classification can be achieved through different algorithms and techniques. Feature vectors can be constructed to quantify and integrate various key indicators of distributed resources. In addition to factors such as response time, response accuracy, and response duration mentioned earlier, factors such as the power adjustment range and the stability of the adjustment rate can also be considered. A comprehensive consideration of these factors can more fully reflect the characteristics of distributed resources, resulting in more detailed and accurate classification results.

[0069] In this embodiment of the invention, granular classification of the first preprocessed real-time running data includes:

[0070] The first feature vector is constructed based on the response time, response accuracy, and response duration of each distributed resource;

[0071] The first feature vector is mapped to the hypersphere feature space, and Mahalanobis distance is used to measure the similarity between samples;

[0072] The idea of ​​density peak clustering is introduced to optimize the selection of initial cluster centers, and iterative clustering is performed by combining the improved K-means+ algorithm to obtain multiple sets of first resource particles.

[0073] In this embodiment of the invention, the response time includes instruction delay time, startup time, rise time, and settling time;

[0074] Response accuracy is calculated using the error rate between the actual response power and the commanded power.

[0075] The duration of the response is dynamically assessed based on available energy, regulated power, and charge / discharge efficiency factor.

[0076] It should be noted that the first preprocessed real-time running data refers to the distributed resource running data after preliminary cleaning, normalization and formatting. This data can be directly used for subsequent analysis and modeling.

[0077] For example, in one optional implementation, the real-time power data uploaded by the distributed energy storage device is filtered and denoised before being converted into a unified time series format for subsequent feature extraction and classification.

[0078] It should be noted that response time refers to the time period required for the system to complete the specified action from receiving the instruction. This time period comprehensively reflects the dynamic characteristics of the device.

[0079] For example, in one optional implementation, the response time of an energy storage device includes the startup time from receiving a dispatch command to the completion of device startup, and the rise time and settling time from the completion of startup to reaching a stable output power.

[0080] It should be noted that response accuracy refers to the degree of deviation between the actual output power of a distributed resource and the target command power. The smaller the error rate, the stronger the response capability of the resource.

[0081] For example, in one optional implementation, a photovoltaic inverter receives a 10kW power regulation command and actually outputs 9.8kW, with a response accuracy error rate of 2%.

[0082] It should be noted that response duration refers to the length of time that distributed resources can operate continuously under specific power regulation requirements. This indicator is affected by the energy reserves and efficiency of the equipment.

[0083] For example, in one optional implementation, if a battery energy storage system has an available energy of 50 kWh, a regulating power of 5 kW, and a charge / discharge efficiency factor of 0.9, then its response duration is approximately 9 hours.

[0084] It should be noted that the first feature vector refers to a multi-dimensional feature representation composed of response time, response accuracy, and response duration, which is used to characterize the comprehensive performance of distributed resources;

[0085] For example, in one optional implementation, the first feature vector of a distributed resource consists of parameters such as instruction latency of 0.2 seconds, response accuracy error rate of 3%, and response duration of 10 hours.

[0086] It should be noted that the hypersphere feature space refers to a high-dimensional space to which the original features are mapped through mathematical transformations, in which the similarity between samples can be measured more effectively;

[0087] For example, in an alternative implementation, the first feature vectors of multiple distributed resources are mapped to a three-dimensional hypersphere space, and the similarity between them is calculated using Mahalanobis distance.

[0088] It should be noted that density peak clustering is a method for determining cluster centers based on the local density and relative distance of data points, used to optimize the selection of initial cluster centers;

[0089] For example, in an optional implementation, the sample with the highest density and the largest distance from other high-density points is selected as the initial cluster center by analyzing the local density and distance distribution of the distributed resource feature vectors.

[0090] It should be noted that the improved K-means+ algorithm refers to an iterative method that introduces optimization strategies on the basis of the traditional K-means algorithm to improve the clustering effect and convergence speed.

[0091] For example, in an optional implementation, the cluster centers of the K-means+ algorithm are initialized by combining the density peak clustering results, and the final set of multiple first resource particles is obtained through multiple iterations.

[0092] Specifically, this invention collects real-time operational data of various resources in the power distribution network (such as solar energy, wind energy, energy storage devices, and adjustable loads) through various data acquisition devices such as smart meters, distributed energy monitoring systems, and load monitoring devices. Specific data items include charging and discharging power, remaining capacity, power consumption of adjustable loads, adjustable range, and characteristic parameters such as response time, response accuracy, and response duration. The data sampling frequency can be set according to actual needs, such as sampling once every 15 minutes, to ensure the timeliness and accuracy of the data.

[0093] Furthermore, the preprocessing of the collected data mainly includes data cleaning and normalization. During data cleaning, outliers and missing values ​​are removed. For missing values, linear interpolation or imputation based on the mean of historical data is used. Let the original data sequence be x = [x1, x2, ..., x...]. n If the position of a missing value is i, then linear interpolation is used to calculate the missing value x. i The formula is:

[0094]

[0095] Furthermore, the normalization process employs a max-min normalization method, mapping the data to the interval [0, 1], as shown in the formula:

[0096]

[0097] Where, x j For the original data, x′ j This is the normalized data.

[0098] Furthermore, the preprocessed data is analyzed based on the resource's response time τ, response accuracy f, and response duration γ.

[0099] The response time τ can be divided into four stages: instruction delay time τ1: the time from instruction issuance to device reception; startup time τ2: the time from device reception to the start of response; rise time τ3: the time from initial response to reaching 90% of the target value; and settling time τ4: the time from reaching 90% of the target value to entering a steady state.

[0100] τ=τ1+τ2+τ3+τ4

[0101] Furthermore, the response accuracy f is measured by the error rate between the actual response power and the command power, as shown in the formula:

[0102]

[0103] Among them, P actual (t) represents the actual response power, P command (t) represents the command power.

[0104] Furthermore, the response duration γ is the length of time that resources can continuously meet scheduling requirements, which is dynamically evaluated in conjunction with the device's SOC status as follows:

[0105]

[0106] Among them, E avail P is the available energy. reg To adjust the power, η is the charge / discharge efficiency factor.

[0107] Furthermore, granular computing technology is employed, using resource response time τ, response accuracy f, and response duration γ as classification features to construct a classification model. An improved K-means+ algorithm is used for resource granulation, incorporating density peak clustering to optimize initial center selection. Adding density calculation to the granular computing step effectively overcomes the limitations of traditional K-means clustering on data distribution. The distribution of power distribution network resource data, characterized by response time, response accuracy, and response duration, may not be a regular sphere, exhibiting density differences and irregular shapes. Density calculation identifies the density of the data distribution, enabling the algorithm to adapt to complex data patterns and accurately segment resource granular sets with different characteristics. Simultaneously, density calculation effectively identifies and removes noise and outliers, preventing them from interfering with the clustering results, enhancing the robustness of the clustering results, and ensuring the accuracy and reliability of resource classification. This provides a more reasonable and practical foundation for subsequent steps such as setting specific indicators, resource classification, and market quotation.

[0108] Furthermore, let the resource data sample be X = [x1, x2, ..., x...]. n ], each sample x i It includes three feature dimensions: response time τ, response accuracy f, and response duration γ, i.e., xi =[τ i ,f i ,γ i The three-dimensional features (τ,f,γ) are mapped onto the hypersphere space.

[0109] Mahalanobis distance is used to measure the similarity between samples:

[0110]

[0111] Where S is the sample covariance matrix

[0112] Furthermore, density calculations are performed:

[0113] ρ i =∑ j χ(d ij -d c )

[0114] Where χ(x) is the indicator function, d c For cutoff distance

[0115] Furthermore, we perform minimum distance calculation:

[0116]

[0117] Furthermore, cluster centers are selected by choosing points with larger ρ and δ values ​​as initial cluster centers.

[0118] Furthermore, iterative optimization can be carried out:

[0119] The objective function is set as follows:

[0120]

[0121] The first term represents the traditional K-means+ objective, and the second term represents the density constraint. c represents the clustering result, K is the preset number of clusters, and μ... k Let k be the kth cluster center.

[0122] Furthermore, the formula is iteratively updated:

[0123]

[0124] By iteratively updating the cluster centers to minimize the objective function J(c), the resources are ultimately divided into sets of resource particles with different characteristics, G = {G1, G2, ..., G...}. n}

[0125] It should be noted that granular classification of the real-time operational data after the first preprocessing stage, resulting in distributed resource granular sets with different characteristics, provides a clear and targeted data foundation for subsequent peak shaving and frequency regulation ancillary service information collection. These distributed resource granular sets, each with different characteristics, encompass the unique performance and features of the resources. When collecting peak shaving and frequency regulation ancillary service information, relevant data can be collected accurately based on these characteristics, improving the efficiency and accuracy of information collection.

[0126] S103, presets peak-shaving and frequency-regulation specific indicators, evaluates distributed resources in the distributed resource granular set based on these indicators, and classifies distributed resources into peak-shaving resources, frequency-regulation resources, and reserve resources, wherein:

[0127] It should be noted that once a set of distributed resource granules with different characteristics is obtained, it is necessary to pre-define peak shaving and frequency regulation specific indicators to evaluate and classify these distributed resources. Peak shaving and frequency regulation are very important ancillary services in power system operation, and different types of distributed resources have different performances and roles in these two services.

[0128] In this embodiment of the invention, peak-shaving specific indicators include peak-shaving capacity and peak-shaving rate, and frequency modulation specific indicators include frequency response coefficient and frequency modulation capacity margin.

[0129] Among them, peak shaving capacity represents the total amount of active power that can be adjusted within the scheduling period, and peak shaving rate represents the change in active power per unit time.

[0130] The frequency response coefficient is the ratio of the change in active power to the system frequency deviation, and the frequency regulation capacity margin is the available frequency regulation power reserve after the resources meet the current operating requirements.

[0131] In this embodiment of the invention, the evaluation of distributed resources based on peak shaving and frequency modulation specific indicators includes:

[0132] Construct a first evaluation matrix, where rows correspond to each distributed resource and columns correspond to each evaluation indicator;

[0133] The first weight vector of each evaluation indicator was determined using the analytic hierarchy process.

[0134] Calculate the first comprehensive score of each distributed resource, and divide the resources based on a preset first threshold and a second threshold, wherein the second threshold is greater than the first threshold;

[0135] Specifically, when the first comprehensive score is greater than or equal to the second threshold, it is classified as a frequency modulation resource; when it is less than the second threshold but greater than or equal to the first threshold, it is classified as a peak shaving resource; otherwise, it is classified as a reserve resource.

[0136] It should be noted that peak-shaving capacity refers to the total amount of active power that resources can adjust during the scheduling period. It reflects the resource's ability to adapt to power system load fluctuations within a specific time period.

[0137] For example, in one alternative implementation, an energy storage device has a peak-shaving capacity of 50MW per day, which means that the device can adjust up to 50MW of active power through charging and discharging operations on the same day to meet the peak-shaving needs of the system.

[0138] It should be noted that the peak shaving rate refers to the change in active power per unit time, which is used to measure the speed and flexibility of the resource response system to peak shaving demands.

[0139] For example, in one optional implementation, the peak shaving rate of a distributed power source is 10MW / min, which means that the power source can increase or decrease its active power output by 10MW per minute, thereby quickly adapting to the dynamic adjustment requirements of the power system.

[0140] It should be noted that the frequency response coefficient is the ratio of the change in active power to the system frequency deviation. This indicator reflects the sensitivity and adjustment capability of resources to system frequency fluctuations.

[0141] For example, in one alternative implementation, a frequency response coefficient of a certain frequency modulation resource is 20MW / Hz, indicating that when the system frequency deviates by 0.1Hz, the resource can provide 2MW of active power adjustment to help restore frequency stability.

[0142] It should be noted that frequency regulation capacity margin refers to the available frequency regulation power reserve of resources after meeting current operating requirements. It is an important basis for measuring the potential frequency regulation capability of resources.

[0143] For example, in one alternative implementation, a generator set is currently operating at 80MW and its maximum output is 120MW, so its frequency regulation capacity margin is 40MW, which can be used to cope with sudden frequency regulation tasks.

[0144] It should be noted that the first comprehensive score refers to the quantitative result calculated based on the first evaluation matrix. This score comprehensively reflects the performance of each distributed resource in peak shaving and frequency regulation specific indicators.

[0145] For example, in one optional implementation, a distributed resource has a first comprehensive score of 0.85, which is higher than a preset second threshold of 0.8, and is therefore classified as a frequency modulation resource for priority participation in the system's frequency regulation task.

[0146] Specifically, for peak-shaving resources, the peak-shaving capacity C is set. peak Peak shaving rate V peakSpecific indicators, such as peak-shaving capacity C. peak The formula represents the total active power that resources can increase or decrease within a certain period of time.

[0147] C peak =max(P max -P current ,P current -P min )

[0148] Among them, P max P represents the maximum active power of the resource. min P is the minimum active power of the resource. current This refers to the active power of resources that do not participate in peak shaving and frequency regulation.

[0149] Peak shaving rate V peak The formula for expressing the change in active power of resources per unit time is:

[0150]

[0151] Among them, P t Let P be the active power at time t. t-Δt Let be the active power at time t-Δt, where Δt is the time interval.

[0152] Furthermore, for frequency modulation resources, the frequency response coefficient α and the frequency modulation capacity margin M are set. freq Specific indicators, etc. The frequency response coefficient α represents the ratio of the change in active power of resources to the change in system frequency, and the formula is:

[0153]

[0154] Where ΔP is the change in active power of resources, and Δf is the change in system frequency.

[0155] Furthermore, the frequency modulation capacity margin M freq This represents the active power margin available for frequency regulation after resources have met current operational needs. The formula is:

[0156] M freq =P max -P current -P reserve

[0157] Among them, P reserve Reserved backup power.

[0158] Furthermore, based on the established peak-shaving and frequency-regulation specific indicators, and combined with the response time τ, response accuracy f, and response duration γ of the resources in the first part, the resources in each resource granular set are evaluated. An evaluation matrix A is established, where element aij This represents the evaluation value of the i-th resource under the j-th indicator (including peak shaving, frequency modulation-specific indicators, and three basic characteristic indicators).

[0159] Furthermore, the Analytic Hierarchy Process (AHP) is used to determine the weights w of each indicator. j By calculating the comprehensive score S of resources score The formula for classification is:

[0160]

[0161] Among these, shorter response time, higher response accuracy, and longer response duration result in higher scores for the corresponding indicators in the evaluation. Thresholds S are set for peak shaving and frequency modulation. peak_threshold and S freq_threshold Because frequency modulation (FM) has higher requirements for resource response performance, therefore, S freq_threshold >S peak_threshold The thresholds for peak modulation and frequency modulation here are the first threshold and the second threshold mentioned earlier.

[0162] When S peak_threshold score freq_threshold Resources are categorized into those participating in peak shaving; when S score >S freq_threshold Resources are categorized into frequency modulation classes, when S score peak_threshold The resources are designated as reserve resources.

[0163] It should be noted that pre-setting specific indicators for peak shaving and frequency regulation, and evaluating distributed resources within the distributed resource granular set based on these indicators, categorizes distributed resources into peak shaving resources, frequency regulation resources, and reserve resources. This provides a clear resource classification basis for subsequent determination of scheduling priorities and resource allocation. Different types of resources play different roles in the power system. Peak shaving resources are mainly used to cope with load fluctuations in the power system, balancing power supply and demand by adjusting active power. Frequency regulation resources focus on maintaining the frequency stability of the power system, responding quickly to changes in system frequency. Reserve resources serve as an emergency reserve, providing additional power support when the system experiences unexpected situations.

[0164] S104: Obtain the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching scheduling principle, and determine the priority of peak-shaving and frequency regulation resources, wherein:

[0165] ​​​It's important to note that after classifying distributed resources, the next step is to determine resource priorities based on the target power grid's peak-shaving and frequency regulation needs, as well as market information. The target power grid's peak-shaving needs primarily address significant fluctuations in electricity load across different time periods to ensure real-time balance between power supply and demand. Frequency regulation needs aim to maintain grid frequency stability and ensure the safe and reliable operation of the power system. Market information includes pricing data for various resources and trading rules; these factors all influence resource priority determination.

[0166] S105, Construct a multi-objective optimization model to determine the scheduling volume of various distributed resource data, and perform scheduling of distributed resources in the power ancillary services market, wherein:

[0167] It should be noted that once the priority of peak-shaving and frequency regulation resources is determined, constructing a multi-objective optimization model to determine the scheduling volume of various distributed resource data becomes particularly crucial. In the power ancillary services market, multiple objectives need to be comprehensively considered to achieve optimal resource allocation and utilization.

[0168] In one alternative implementation, the objective function of a multi-objective optimization model typically includes several aspects. One aspect is considering the economics of the power system, reducing the costs of power generation and dispatch. For example, different types of distributed resources differ in operating and maintenance costs, and the model needs to select resources with lower costs for dispatch while meeting peak-shaving and frequency regulation requirements.

[0169] In one alternative implementation, on the other hand, it is necessary to ensure the reliability and stability of the power system. This means that the model must ensure that during the scheduling process, unreasonable allocation of resources will not lead to problems such as frequency fluctuations or voltage instability in the system.

[0170] In this embodiment of the invention, constructing a multi-objective optimization model includes:

[0171] The first optimization objective is to minimize the total scheduling cost, and the second optimization objective is to maximize the weighted sum of the total performance weight and the scheduling amount.

[0172] Set a first set of constraints, which includes: peak shaving and frequency regulation demand satisfaction constraints, resource capacity constraints, and scheduling quantity accuracy constraints.

[0173] Among them, the minimum scheduling unit for peak shaving service is defined as the first unit value, and the minimum scheduling unit for frequency regulation service is defined as the second unit value in the scheduling quantity precision constraint.

[0174] Specifically, based on the resource classification results, the peak-shaving and frequency-regulating capabilities of the resources participating in peak shaving and frequency regulation are further quantitatively assessed. For peak-shaving resources, a comprehensive peak-shaving capability value C is calculated by combining peak-shaving capacity, peak-shaving rate, response time, response accuracy, response duration, and historical peak-shaving performance. peak_com The formula is

[0175] C peak_com =w1C peak +w2V peak +w3τ+w4f+w5γ+w6H peak

[0176] Where w1, w2, w3, w4, w5, w6 are weighting coefficients, and H peak Score the historical peak-shaving performance.

[0177] Furthermore, for FM resources, the comprehensive FM capability value C is calculated by combining the frequency response coefficient, FM capacity margin, response time, response accuracy, response duration, and historical FM performance. fre_com

[0178] C freq_com =w7α+w8M freq +w9τ+w 10 f+w 11 γ+w 12 H freq

[0179] Among them, w7, w8, w9, w 10 ,w 11 ,w 12 H is the weighting coefficient. freq Score the historical peak-shaving performance.

[0180] Furthermore, traditional scheduling principles prioritize economic efficiency and do not consider resource performance. The scheduling principle proposed in this invention balances economic priority with optimal performance matching, specifically manifested as follows:

[0181] (1) Economic priority principle: Resources with lower costs are prioritized for peak shaving and frequency regulation ancillary services. Cost assessment covers resource allocation costs, maintenance costs, etc., and the priority scheduling order is determined by comprehensively comparing the economics of different resources.

[0182] (2) Optimal Performance Matching Principle: Based on the different performance requirements of peak shaving and frequency regulation, performance indicators are matched with resource characteristics. Peak shaving prioritizes resources with large peak shaving capacity, high rate, and short response time, i.e., the comprehensive resource score S. score Resources that meet the requirements for peak shaving are prioritized; for frequency modulation, resources with high frequency response coefficients, large frequency modulation capacity margins, high response accuracy, and long response duration are given priority, i.e., the comprehensive score S of the resources.score Resources that meet the requirements for frequency modulation.

[0183] Furthermore, for peak-shaving resources, based on peak-shaving capacity (C) peak Peak shaving rate (V) peak The priority ranking of peak-shaving resources is determined by comprehensively considering indicators such as response time (τ), response accuracy (f), and response duration (γ).

[0184] For frequency modulation resources, based on the frequency response coefficient (α) and frequency modulation capacity margin (M) freq The ranking is based on key indicators such as response time (τ), response accuracy (f), and response duration (γ).

[0185] Furthermore, it involves collecting data on grid load forecasts, generation plans, electricity price signals, and the demand for peak shaving and frequency regulation in the market clearing process.

[0186] Analyze the total demand for power grid peak shaving and frequency regulation, as well as the specific demand distribution in different regions and time periods.

[0187] Furthermore, frequency modulation resource priority is determined based on the overall resource score S. score Among the resources that meet the frequency modulation (FM) criteria, they are sorted from high to low according to their frequency response coefficient (α) to form a FM resource priority sequence; this is then combined with the FM capacity margin (M). freq The priority can be fine-tuned by factors such as response accuracy (f) and response duration (γ).

[0188] Determining the priority of peak-shaving resources: based on the comprehensive score S of the resources. score Among resources that meet the peak-shaving requirements, based on peak-shaving capacity (C) peak Sort the resources from largest to smallest to form a priority sequence for peak-shaving resources; combine this with the peak-shaving rate (V) peak Optimize priority based on factors such as response time (τ).

[0189] Furthermore, the determination of frequency regulation resource allocation is based on the grid frequency regulation demand (D). freq Taking into account the performance of frequency modulation resources, the scheduling amount F of frequency modulation resource i participating in frequency modulation ancillary services is determined based on the principle of optimal economy. i .

[0190] Determining the peak-shaving resource allocation: Based on the power grid's peak-shaving demand (D) peak Taking into account the performance of peak-shaving resources, the scheduling amount P of peak-shaving resource j participating in peak-shaving ancillary services is determined based on the principle of optimal economic efficiency. j .

[0191] Furthermore, the scheduling amount F of frequency modulation resource i participating in frequency modulation ancillary services i And the scheduling amount P of peak-shaving resource j participating in peak-shaving ancillary servicesj The calculation requires consideration of both economic efficiency and optimal performance matching principles. This is also the focus of the scheduling principle proposed in this invention.

[0192] Furthermore, the price for distributed resources to participate in ancillary services is determined based on the comprehensive value of their peak-shaving and frequency-regulating capabilities and market supply and demand. For peak-shaving resources, the price is relatively high when market demand for peak-shaving is high and their own peak-shaving capabilities are strong; conversely, the price is relatively low when market demand for peak-shaving is low or their own peak-shaving capabilities are weak. The specific pricing formula is as follows:

[0193]

[0194] Among them, R peak For peak shaving pricing, R peak_base This is the base price for peak shaving, where β1 and β2 are adjustment coefficients.

[0195] D represents the comprehensive value of the market's average peak-shaving capacity. peak To meet market peak-shaving demand, D peak_total To regulate the total demand in the market.

[0196] Furthermore, for FM resources, similarly, a pricing formula is formulated based on FM capability and market FM demand:

[0197]

[0198] Among them, R freq For peak shaving pricing, R freq_base This is the base price for FM tuning, where β3 and β4 are adjustment coefficients.

[0199] C freq D represents the overall average FM capability of the market. freq To meet market demand for frequency tuning, D freq_total This is to regulate the total demand in the market.

[0200] Furthermore, the Multidimensional Dynamic Marginal Cost-Performance Co-optimization Algorithm (MDBPCPO) is used to calculate the scheduling amount of each resource.

[0201] Based on the existing invention, this section aims to propose a novel and advanced method for determining dispatch quantity. It takes the principle of optimal economic efficiency as its core, while also considering the performance matching degree of resources, to ensure that when distributed resources participate in the power ancillary services market, the allocation of dispatch quantity not only meets the grid's peak-shaving and frequency regulation needs, but also minimizes the total cost.

[0202] The multidimensional dynamic marginal cost-performance co-optimization algorithm innovatively integrates a dynamic marginal cost adjustment mechanism with a multidimensional performance weight co-optimization mechanism. It constructs a multi-objective optimization model and introduces an intelligent optimization algorithm to determine the optimal scheduling quantity. Its core lies in not only considering the static cost and performance of resources but also dynamically adjusting marginal costs in real time to reflect cost changes under different scheduling quantities. Simultaneously, it measures resource performance from multiple dimensions to accurately match grid demand.

[0203] Furthermore, considering the real-time operating status of peak-shaving resources and market supply and demand dynamics, calculate their dynamic marginal cost:

[0204]

[0205] in, This indicates that peak-shaving resource j is in the scheduling quantity P. j Dynamic marginal cost at time; C op,j ΔC represents the fixed unit cost of resource j. maint,j (P j ) is related to the scheduling quantity P j The increase in maintenance costs reflects the differences in maintenance costs at different peak-shaving depths; λ peak This is the peak-shaving performance adjustment coefficient, which measures the weight of the overall peak-shaving capacity on cost. This represents the rate of change of the overall peak-shaving capacity with the amount of scheduling, reflecting the marginal performance contribution of resources at different peak-shaving depths.

[0206] Furthermore, similarly, calculate the dynamic marginal cost of frequency modulation resources:

[0207]

[0208] in, This indicates that peak-shaving resource i is in the scheduling quantity F i Dynamic marginal cost at time; C op,i The fixed unit cost of resource i; ΔC maint,j (F i ) is dependent on the scheduling quantity F i The incremental maintenance cost reflects the difference in maintenance costs under different frequency modulation depths; λ freq This is the frequency modulation performance adjustment coefficient, which measures the weight of the overall frequency modulation capability on cost. This represents the rate of change of the overall frequency regulation capability with the amount of scheduling, reflecting the marginal performance contribution of resources at different frequency regulation depths.

[0209] Furthermore, considering multiple performance indicators of peak-shaving resources, their multidimensional performance weights are calculated:

[0210]

[0211] in, For the multidimensional performance weights of peak-shaving resource j; C peak_com,j It is the comprehensive value of peak-shaving capability; V peak,j The peak shaving rate reflects the resource's ability to quickly respond to peak shaving demands; This is the reciprocal of the peak-shaving response time, reflecting the timeliness of resource response. This multi-dimensional collaborative approach more accurately measures the overall performance advantage of resources in peak-shaving tasks.

[0212] Furthermore, for frequency modulation resources, the multidimensional performance weighting calculation formula is as follows:

[0213]

[0214] in, For the multidimensional performance weights of frequency modulation resource i; C freq_com,i It is the overall value of frequency modulation capability; α i γ is the frequency response coefficient, representing the resource's sensitivity and adjustment capability to frequency changes; i The duration of response reflects the sustained availability of resources during frequency regulation.

[0215] In summary, this invention proposes a method for dispatching distributed resources in a power distribution network using multi-functional power ancillary services. By acquiring and preprocessing real-time operational data of distributed resources, missing and outlier values ​​are removed, making the data more accurate and reliable. Normalization also ensures comparability between different types of data, laying a solid foundation for subsequent classification and evaluation. In the granular classification process, key factors such as response time, response accuracy, and response duration of distributed resources are comprehensively considered. A scientific method is used to classify resources, enabling a more accurate understanding of the characteristics of each type of resource. Distributed resources are evaluated based on peak-shaving and frequency-regulation specific indicators, and resources are rationally divided into peak-shaving, frequency-regulation, and reserve categories, making resource allocation more aligned with actual needs. A multi-objective optimization model is constructed, simultaneously considering minimizing the total dispatch cost and maximizing the weighted sum of total performance weights and dispatch quantities. Under constraints such as peak-shaving and frequency-regulation requirements, resource capacity, and dispatch quantity accuracy, efficient resource dispatch is achieved.

[0216] Example 2, in a preferred embodiment, involves constructing a multi-objective optimization model to determine the scheduling volume of various types of distributed resource data. The scheduling of distributed resources in the power ancillary services market further includes:

[0217] Construct a multi-objective optimization model that minimizes total cost and achieves optimal performance matching:

[0218]

[0219] The first objective is to minimize the total scheduling cost, and the second objective is to maximize the weighted sum of the total performance weight and the scheduling amount, reflecting the principle of optimal performance matching.

[0220] Furthermore, the peak shaving and frequency regulation requirements meet the following constraints:

[0221]

[0222] Resource capacity constraints:

[0223] 0≤P j ≤P max,j ,0≤F i ≤F max,i

[0224] Among them, P max,j F represents the maximum adjustable amount of peak-shaving resource j. max,i This represents the maximum adjustable value of frequency modulation resource i.

[0225] Scheduling quantity precision constraints (fine-grained control to ensure scheduling quantity matches resource performance):

[0226]

[0227] Wherein, ΔP min and ΔF min These are the smallest scheduling units for peak shaving and frequency regulation, respectively, used to finely control the allocation accuracy of scheduling quantities.

[0228] Example 3, referring to Figure 2 This embodiment also provides a distribution network distributed resource participation multi-power auxiliary service dispatching system, including:

[0229] The data acquisition and processing module is used to acquire real-time operational data of several types of distributed resource data in the target power grid, and to perform the first preprocessing on the real-time operational data;

[0230] The classification set acquisition module is used to granularize the real-time running data after the first preprocessing to obtain a distributed resource granular set with different characteristics;

[0231] The partitioning module is used to preset peak shaving and frequency regulation specific indicators, evaluate the distributed resources in the distributed resource granular set based on the peak shaving and frequency regulation specific indicators, and divide the distributed resources into peak shaving resources, frequency regulation resources and reserve resources.

[0232] The priority acquisition module is used to acquire the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching scheduling principle, and determine the priority of peak-shaving and frequency regulation resources.

[0233] The scheduling module is used to build a multi-objective optimization model, determine the scheduling volume of various types of distributed resource data, and perform scheduling of distributed resources in the power ancillary services market.

[0234] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0235] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows: Figure 2 As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for dispatching distributed resources in a power distribution network using diversified power auxiliary services. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0236] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0237] Acquire real-time operational data of several types of distributed resource data in the target power grid, and perform first preprocessing on the real-time operational data;

[0238] The real-time running data after the first preprocessing is granularized and classified to obtain a set of distributed resource granules with different characteristics;

[0239] Preset peak shaving and frequency regulation specific indicators, evaluate distributed resources in the distributed resource granular set based on peak shaving and frequency regulation specific indicators, and classify distributed resources into peak shaving resources, frequency regulation resources and reserve resources;

[0240] Obtain the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching and scheduling principle, and determine the priority of peak-shaving and frequency regulation resources;

[0241] A multi-objective optimization model is constructed to determine the scheduling volume of various types of distributed resource data, and to schedule distributed resources in the power ancillary services market.

[0242] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0243] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0244] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for dispatching distributed resources in a power distribution network using multi-faceted power ancillary services, characterized in that, include: Acquire real-time operational data of several types of distributed resource data in the target power grid, and perform a first preprocessing on the real-time operational data; The preprocessed real-time running data is granularized and classified to obtain a set of distributed resource granules with different characteristics; Preset peak shaving and frequency regulation specific indicators, evaluate the distributed resources in the distributed resource granular set based on the peak shaving and frequency regulation specific indicators, and divide the distributed resources into peak shaving resources, frequency regulation resources and reserve resources; Obtain the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching and scheduling principle, and determine the priority of peak-shaving and frequency regulation resources; A multi-objective optimization model is constructed to determine the scheduling volume of various types of distributed resource data, and to schedule distributed resources in the power ancillary services market.

2. The method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 1, characterized in that, The first preprocessing of the real-time running data includes: The real-time running data is processed for missing values ​​and outliers are removed to obtain the first cleaned data; The first cleaned data is normalized using the max-min normalization method to obtain the first normalized dataset; Missing values ​​are filled using linear interpolation or by filling in the missing values ​​based on the mean of historical data.

3. A method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 2, characterized in that, The granular classification of the first preprocessed real-time running data includes: The first feature vector is constructed based on the response time, response accuracy, and response duration of each distributed resource; The first feature vector is mapped to the hypersphere feature space, and Mahalanobis distance is used to measure the similarity between samples; The idea of ​​density peak clustering is introduced to optimize the selection of initial cluster centers, and iterative clustering is performed by combining the improved K-means+ algorithm to obtain multiple sets of first resource particles.

4. The method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 3, characterized in that, The response time includes instruction delay time, startup time, rise time, and settling time; The response accuracy is calculated using the error rate between the actual response power and the command power. The duration of the response is dynamically assessed based on available energy, regulated power, and charge / discharge efficiency factor.

5. A method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 4, characterized in that, The peak-shaving specific indicators include peak-shaving capacity and peak-shaving rate, and the frequency modulation specific indicators include frequency response coefficient and frequency modulation capacity margin. Among them, peak shaving capacity represents the total amount of active power that can be adjusted within the scheduling period, and peak shaving rate represents the change in active power per unit time. The frequency response coefficient is the ratio of the change in active power to the system frequency deviation, and the frequency regulation capacity margin is the available frequency regulation power reserve after the resources meet the current operating requirements.

6. A method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 5, characterized in that, The evaluation of distributed resources based on peak shaving and frequency modulation specific indicators includes: Construct a first evaluation matrix, where rows correspond to each distributed resource and columns correspond to each evaluation indicator; The first weight vector of each evaluation indicator was determined using the analytic hierarchy process. Calculate the first comprehensive score of each distributed resource, and divide the resources based on a preset first threshold and a second threshold, wherein the second threshold is greater than the first threshold; Specifically, when the first comprehensive score is greater than or equal to the second threshold, it is classified as a frequency modulation resource; when it is less than the second threshold but greater than or equal to the first threshold, it is classified as a peak shaving resource; otherwise, it is classified as a reserve resource.

7. A method for dispatching distributed resources in a power distribution network with participation in multi-faceted power ancillary services as described in claim 6, characterized in that, The construction of the multi-objective optimization model includes: The first optimization objective is to minimize the total scheduling cost, and the second optimization objective is to maximize the weighted sum of the total performance weight and the scheduling amount. A first set of constraints is set, which includes: peak shaving and frequency regulation demand satisfaction constraints, resource capacity constraints, and scheduling quantity accuracy constraints. Among them, the minimum scheduling unit for peak shaving service is defined as the first unit value, and the minimum scheduling unit for frequency regulation service is defined as the second unit value in the scheduling quantity precision constraint.

8. A distributed resource participation multi-electronic power auxiliary service dispatching system for distribution networks, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition and processing module is used to acquire real-time operational data of several types of distributed resource data in the target power grid, and to perform a first preprocessing on the real-time operational data; The classification set acquisition module is used to granularize the first preprocessed real-time running data to obtain a distributed resource granular set with different characteristics; The partitioning module is used to preset peak shaving and frequency regulation specific indicators, evaluate the distributed resources in the distributed resource granular set based on the peak shaving and frequency regulation specific indicators, and partition the distributed resources into peak shaving resources, frequency regulation resources and reserve resources. The priority acquisition module is used to acquire the peak-shaving and frequency regulation needs and market information of the target power grid, determine the priority of peak-shaving and frequency regulation resources, preset the optimal matching scheduling principle, and determine the priority of peak-shaving and frequency regulation resources. The scheduling module is used to build a multi-objective optimization model, determine the scheduling volume of various types of distributed resource data, and perform scheduling of distributed resources in the power ancillary services market.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for dispatching distributed resources in a power distribution network with participation in multiple power auxiliary services as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for dispatching distributed resources in a power distribution network with participation in multiple power auxiliary services as described in any one of claims 1 to 7.

Images