CN-121997257-A - Virtual power plant distributed resource aggregation data processing method, system and storage medium

CN121997257ACN 121997257 ACN121997257 ACN 121997257ACN-121997257-A

Abstract

The invention provides a method, a system and a storage medium for processing distributed resource aggregation data of a virtual power plant, which are used for processing real-time and historical data of each distributed resource based on time exponential decay weights, constructing time weighted feature vectors containing indexes such as rated capacity, power fluctuation rate and the like, dividing the resource into a plurality of resource clusters through a clustering algorithm, calculating cluster center features considering historical scheduling compliance and data confidence, evaluating the deviation degree of each resource from the cluster center of the corresponding resource on multidimensional indexes, calculating the deviation degree discrete coefficient in the cluster according to the evaluation, further determining an adjustment threshold, rejecting the resource with the deviation degree exceeding the threshold, distributing aggregation weights according to the deviation degree for the rest resources, and aggregating the weighted data of reserved resources to obtain the total aggregation state of the virtual power plant.

Inventors

HU XIN
LIU ZHENZHEN
HUANG JINGYIN
HUANG KANGQIAN
ZHAN LICHAO
SHANG JINGXIANG
YANG FAN
LIU JIAJUN
LIN XIJUN
ZHENG YINGKAI

Assignees

广东电力交易中心有限责任公司
南方电网数字电网集团(广东)有限公司

Dates

Publication Date: 20260508
Application Date: 20260119

Claims (9)

1. The distributed resource aggregation data processing method for the virtual power plant is characterized by comprising the following steps of: acquiring real-time and historical data of each distributed resource, constructing a time weighted feature vector for each distributed resource based on rated capacity, historical power fluctuation rate, response delay and communication link quality; Calculating the multi-dimensional adjustment potential energy deviation degree between the current running state of each distributed resource and the cluster center characteristic of the affiliated resource cluster according to the active power deviation, the adjustment response time mismatch and the state unexpected switching cost; If the multidimensional adjustment potential energy deviation degree of the distributed resources exceeds the adjustment threshold value of the belonging resource cluster, eliminating the resources in the current aggregation, otherwise, reserving the resources and distributing aggregation weights for the resources, wherein the aggregation weights and the multidimensional adjustment potential energy deviation degree are in negative correlation, and aggregating the weighted data of all reserved resources to obtain the total aggregation state of the virtual power plant.
2. The method of claim 1, wherein constructing a time-weighted feature vector for each distributed resource based on rated capacity, historical power fluctuation rate, response delay, and communication link quality comprises: the rated capacity is taken as a static characteristic dimension, the value is the nominal rated capacity, the historical power fluctuation rate, the response delay and the communication link quality are taken as dynamic characteristic dimensions, and the dynamic characteristic dimensions The calculation formula of (2) is as follows: Wherein, the For the value of the jth dynamic feature dimension at the current instant t, For the latest observation of the jth dynamic feature at the current instant t, As the characteristic value of the previous moment in time, Is the smoothing factor of the j-th dynamic feature dimension, and The smoothing factor The value of (2) is preset according to the time-varying characteristic of the corresponding characteristic, and the smoothing factor of the response delay and the communication link quality is larger than that of the historical power fluctuation rate; and combining the static characteristic dimension value and the three dynamic characteristic dimension values according to the preset sequence of rated capacity, historical power fluctuation rate, response delay and communication link quality to form a four-dimensional vector, and obtaining the time weighted characteristic vector.
3. The method of claim 1, wherein the partitioning of the distributed resources into at least two resource clusters using a preset clustering algorithm comprises: dividing by adopting a K-means++ clustering algorithm, wherein the number K of preset resource clusters is 5; In the K-means++ clustering algorithm, the distance between the computing resources is a weighted Euclidean distance, wherein the weights of the four dimensions of rated capacity, historical power fluctuation rate, response delay and communication link quality are respectively 0.2, 0.4, 0.3 and 0.1.
4. The method of claim 1, wherein said calculating cluster center characteristics for each resource cluster comprises: Taking the product of the historical scheduling compliance and the data confidence as the weight of weighted average; The historical scheduling compliance degree is the ratio of the number of times of successfully responding to scheduling instructions by the distributed resource to the total number of times of scheduling instructions in the past 30 days; the data confidence is quantized according to the completeness and timeliness of data uploading, and the value of the data confidence is a floating point number between 0 and 1; Cluster center feature The calculation formula of (2) is as follows: Wherein For the feature vector of the ith resource in cluster k, Is the weight of the ith resource in cluster k.
5. The method of claim 1, wherein the calculating a multi-dimensional adjustment potential energy offset between the current operating state of each distributed resource and the cluster center feature of the belonging cluster of resources based on the active power offset, the adjustment response time mismatch, and the state unexpected handoff cost comprises: the calculation formula of the multidimensional adjustment potential energy deviation degree is as follows: , Wherein the method comprises the steps of As the absolute value of the difference between the current active power actual value and the scheduling target value, Is the rated capacity of the resource; the absolute value of the difference between the actual response time and the standard response time defined for the cluster center feature, Is a reference time constant; for the cost of unexpected state switching, when unscheduled start-stop or mode switching occurs to the resource, the value is 1, otherwise, the value is 0; 、 And Is a preset weight coefficient.
6. The method of claim 1, wherein calculating intra-cluster bias degree discrete coefficients for a resource cluster, and deriving the adjustment threshold using the bias degree discrete coefficients, comprises: calculating the average value of all distributed resource multidimensional adjustment potential energy deviation degrees in a cluster And standard deviation By coefficient of variation As the intra-cluster deviation degree discrete coefficient; The calculation formula of the adjustment threshold T is: Wherein As a basis for the threshold value, To adjust the sensitivity coefficient.
7. The method of claim 1, wherein the assigning an aggregate weight to the resource, the aggregate weight being inversely related to the degree of deviation of the multi-dimensional adjustment potential energy, comprises: And calculating the aggregation weight by adopting a Gaussian function, wherein a specific calculation formula is as follows: Wherein For aggregating weights, D is the multidimensional adjustment potential energy deviation degree of the resources, s is a preset parameter for controlling the weight distribution width, and the value of s is 0.5 times of the adjustment threshold value of the resource cluster.
8. A virtual power plant distributed resource aggregate data processing system, comprising the following elements: The device comprises a feature extraction unit, a clustering unit and a clustering unit, wherein the feature extraction unit is used for acquiring real-time and historical data of each distributed resource, constructing a time weighted feature vector for each distributed resource based on rated capacity, historical power fluctuation rate, response delay and communication link quality; The system comprises a calculation unit, a multi-dimensional adjustment potential energy deviation degree calculation unit and a power control unit, wherein the calculation unit is used for calculating the multi-dimensional adjustment potential energy deviation degree between the current running state of each distributed resource and the cluster center characteristic of the affiliated resource cluster according to the active power deviation, the adjustment response time mismatch and the state unexpected switching cost; The aggregation unit is used for eliminating the resources in the aggregation if the multidimensional adjustment potential energy deviation degree of the distributed resources exceeds the adjustment threshold value of the resource cluster, otherwise, reserving the resources and distributing aggregation weights for the resources, wherein the aggregation weights and the multidimensional adjustment potential energy deviation degree are in a negative correlation relationship, and aggregating the weighted data of all reserved resources to obtain the total aggregation state of the virtual power plant.
9. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the method according to any of claims 1-7.

Description

Virtual power plant distributed resource aggregation data processing method, system and storage medium Technical Field The application belongs to the field of data processing, and particularly relates to a method, a system and a storage medium for processing distributed resource aggregation data of a virtual power plant. Background The virtual power plant is an advanced energy management form, and geographically dispersed distributed energy sources such as wind power, photovoltaic, an energy storage system, controllable loads and the like are aggregated through an information communication technology to form a unified whole to participate in power market and power grid dispatching. And evaluating and processing the aggregate data of the distributed resources, which is a precondition for ensuring the safe, stable and economic operation of the virtual power plant. However, the running states of the distributed resources are quite different, the existing aggregation method usually classifies or clusters the resources firstly, but the existing aggregation method is mostly dependent on static physical properties of the resources such as types and rated capacities, real-time characteristics of the resources such as power fluctuation conditions, communication link quality and response delay are ignored, and the clustering result cannot truly reflect the actual running state cluster of the resources under a specific time section. And when the state parameter of a certain resource, such as power deviation, exceeds a preset constant limit value, the resource is removed from the aggregation calculation, and the change of the state discrete degree in different resource clusters cannot be adapted. For example, a cluster of resources consisting of fans has itself a relatively high volatility, while a cluster consisting of stored energy is relatively stable, and it is clearly unreasonable to use the same reject criteria. And for the screened resources, most of the resources are aggregated by adopting a simple weighting mode based on equal weight or capacity, and the weight distribution is not carried out according to the real-time deviation degree of the resources and the cluster center characteristics, so that the influence of the more stable and ideal resources on the final result is not great, and the reliability and the usability of the aggregation state are reduced. Disclosure of Invention In order to solve the above problems, the present invention provides a distributed resource aggregation data processing method for a virtual power plant, which includes the following steps: acquiring real-time and historical data of each distributed resource, constructing a time weighted feature vector for each distributed resource based on rated capacity, historical power fluctuation rate, response delay and communication link quality; Calculating the multi-dimensional adjustment potential energy deviation degree between the current running state of each distributed resource and the cluster center characteristic of the affiliated resource cluster according to the active power deviation, the adjustment response time mismatch and the state unexpected switching cost; If the multidimensional adjustment potential energy deviation degree of the distributed resources exceeds the adjustment threshold value of the belonging resource cluster, eliminating the resources in the current aggregation, otherwise, reserving the resources and distributing aggregation weights for the resources, wherein the aggregation weights and the multidimensional adjustment potential energy deviation degree are in negative correlation, and aggregating the weighted data of all reserved resources to obtain the total aggregation state of the virtual power plant. Optionally, the constructing a time weighted feature vector for each distributed resource based on the rated capacity, the historical power fluctuation rate, the response delay, and the communication link quality includes: the rated capacity is taken as a static characteristic dimension, the value is the nominal rated capacity, the historical power fluctuation rate, the response delay and the communication link quality are taken as dynamic characteristic dimensions, and the dynamic characteristic dimensions The calculation formula of (2) is as follows: Wherein, the For the value of the jth dynamic feature dimension at the current instant t,For the latest observation of the jth dynamic feature at the current instant t,As the characteristic value of the previous moment in time,Is the smoothing factor of the j-th dynamic feature dimension, andThe smoothing factorThe value of (2) is preset according to the time-varying characteristic of the corresponding characteristic, and the smoothing factor of the response delay and the communication link quality is larger than that of the historical power fluctuation rate; and combining the static characteristic dimension value and the three dynamic characteristic dimensio