CN-115719130-B - New energy resource grading evaluation method based on clustering algorithm

Abstract

The invention discloses a new energy resource grading evaluation method based on a clustering algorithm, which specifically comprises the following steps of 1, solving the monthly power generation theoretical hours of each new energy station, 2, analyzing and correcting data acquired in the step 1, constructing an evaluation index of each new energy station, 3, analyzing the new energy station evaluation indexes obtained in the step 2 to obtain a new energy evaluation index of a transformer substation, and 4, aggregating the transformer substation evaluation indexes obtained in the step 3 to obtain a new energy evaluation index of a provincial level and outputting the new energy evaluation index. The invention can provide theoretical basis and engineering guidance for planning new energy installation, and is favorable for promoting the high-quality development of new energy.

Inventors

• LV JINLI
• ZHANG XIAOQI
• Duan Naixin
• GE PENGJIANG
• WANG KANGPING
• ZHAO XIN
• JIN JILIANG

Assignees

• 国家电网有限公司西北分部

Dates

Publication Date

20260505

Application Date

20221115

Claims (1)

1. 1. The new energy resource grading evaluation method based on the clustering algorithm is characterized by comprising the following steps of: Step 1, solving the theoretical hours of monthly power generation of each new energy station, wherein the specific process of the step 1 is that related data of the new energy station is obtained from an EMS scheduling system, wherein the related data comprises grid-connected capacity indexes Index of power generation Electricity limitation indicator Average wind speed index Average temperature index Index of radiant quantity Index of number of hours of sunshine The theoretical number of hours of power generation per station is calculated by the following formula (1): (1); Wherein, the Is the theoretical number of hours of power generation for a wind farm or photovoltaic power plant, Is the total power generation amount in the current month, Is the total electricity limiting amount in the current month, The actual grid-connected capacity of the power station is obtained; Step 2, analyzing and correcting the data obtained in the step 1 to construct an evaluation index of each new energy station, wherein the specific process of the step 2 is as follows: step 2.1, detecting outliers by using an abnormal data identification method based on the mahalanobis distance; (2); Wherein, the The distance of the mahalanobis is indicated, The vector of the samples is represented and, The mean value of the sample is represented, Representing covariance matrix between samples when When the matrix is a unit matrix, the Mahalanobis distance is equal to the Euclidean distance; Step 2.2, setting a center-to-center distance MINDISTANCE, detecting a calculation result obtained in the formula (2), and when the center-to-center distance of the calculation result is smaller than MINDISTANCE, indicating that the theoretical number of power generation hours per month is abnormal, modifying the theoretical number of power generation hours per month, and entering step 2.4, otherwise entering step 2.3; Step 2.3, setting a center-to-center distance maxDistance, detecting the result obtained in step 2.2, and when the theoretical number of hours of power generation per month is greater than maxDistance, indicating that the theoretical number of hours of power generation per month is abnormal, modifying the theoretical number of hours of power generation per month, and entering step 2.4, otherwise entering step 2.5; Step 2.4, if abnormal data exist, correcting the abnormal data by using an average value correction method, wherein the method is specifically shown as a formula (3): (3) Wherein, the In the event of an abnormal data set, And Two data adjacent to the abnormal data; and 2.5, weighting each new energy station index obtained after the correction in the step 2.4, wherein the new energy station index is shown in the following formula (4): (4) Wherein, the Is an index of the average wind speed, Is an index of the average temperature of the water, Is an index of the radiation quantity, and is used for measuring the radiation quantity, Is an index of the number of hours of sunlight, The correction factor, q, represents the comprehensive evaluation index value of the new energy station; Step 3, analyzing the new energy station evaluation indexes obtained in the step 2 to obtain new energy evaluation indexes of a transformer substation, wherein the specific process of the step 3 is as follows: step 3.1, step 2.5 The data adopts a neuron algorithm to calculate a discrimination function value of each input, and a specific neuron with the smallest discrimination function value is considered as a winner, wherein the discrimination function of each neuron j is as follows: (5) Wherein, the input space is D dimension, and the input is The connection weight between the input unit i and the neuron j at the calculation layer is Where N is the total number of neurons; Step 3.2, winning neurons Updating and winning neurons The calculation formula of the update degree of the adjacent nodes is as follows: (6); Wherein, the Representing neurons And the winning neuron Squaring the distance in the output topology space; Is the width of the neighborhood, Decay with time; Step 3.3, adjusting the connection weights of the related excitatory neurons ; Step 3.4, continuing to return to step 3.1 until the feature map tends to be stable; Step 4, aggregating the substation level evaluation indexes obtained in the step 3 to obtain provincial new energy evaluation indexes and outputting the provincial new energy evaluation indexes, wherein the specific process of the step 4 is as follows: Step 4.1, setting cluster parameters, namely, a drift function g (x), a drift vector N h (x), selecting sample points, namely field station data as N, radius as h, selecting cluster density distance as s and clustering number as x The clustering center is i The threshold value of the data density is : (7) Wherein k' (x) represents the derivative of the profile function of the kernel function used in the drifting algorithm; Step 4.2, selecting a circle with radius h in an n-dimensional space R n formed by input station level data, and recording the circle center as o; step 4.3, recording that the station level data in the radius h belongs to a set N, and defining that the station level data in the set N belongs to a cluster C; Step 4.4, calculating a value N h (x) of the drift vector in the cluster C; (8); Wherein xi represents the ith station-level sample data, g (x) represents the derivative of the kernel function outline function, h represents the bandwidth of the mean shift kernel function, namely the neighborhood search radius, and the space radius of the corresponding set N, and x is the circle center of the current neighborhood in the mean shift iteration process, namely the clustering center iteration point, and is the core independent variable of the formula; Step 4.5 comparing the calculated value of the drift phasor N h (x) with a threshold value representing data density The comparison is made if N (x) | < Ending the iterative process, otherwise, re-deriving a new circle center o', repeating the processes of the steps 4.3-4.5, wherein all encountered data points belong to the cluster C during iteration; Step 4.6, if the distance between the maximum density point of the current cluster C and the rest existing cluster density points is larger than the cluster density distance s, adding one class; Step 4.7, repeating the steps 4.2-4.6 until all points are marked; step 4.8, according to the access times of each station level data, the class of the point finally determines the class i of each province according to the class with the largest access times And the number of clusters x 。

Description

New energy resource grading evaluation method based on clustering algorithm Technical Field The invention belongs to the technical field of power system management and evaluation, and relates to a new energy resource grading evaluation method based on a clustering algorithm. Background At present, the new energy installation planning and the power generation grid-connected allocation in China lack of change and flow rule analysis, and the planning analysis on the reasonable power generation of the new energy cannot be performed according to the resource change and the flow rule in one year of resources, so that the current new energy development level evaluation is on one side, the new energy resource grading evaluation method is imperfect and is not beneficial to guiding the future work of a power system, and the low-carbon, safe and efficient modern energy system is built. Disclosure of Invention The invention aims to provide a new energy resource grading evaluation method based on a clustering algorithm, which can provide theoretical basis and engineering guidance for new energy installation planning and is beneficial to promoting the high-quality development of new energy. The technical scheme adopted by the invention is that the new energy resource grading evaluation method based on the clustering algorithm specifically comprises the following steps: step1, solving the monthly power generation theoretical hours of each new energy station; step 2, analyzing and correcting the data obtained in the step 1, and constructing an evaluation index of each new energy station; step 3, analyzing the new energy station evaluation indexes obtained in the step 2 to obtain a substation-level new energy evaluation index; And 4, aggregating the substation level evaluation indexes obtained in the step 3 to obtain provincial level new energy evaluation indexes and outputting the provincial level new energy evaluation indexes. The invention is also characterized in that: The specific process of the step1 is as follows: acquiring related data of a new energy station from an EMS scheduling system, wherein the related data comprises grid-connected capacity indexes Index of power generationElectricity limitation indicatorAverage wind speed indexAverage temperature indexIndex of radiant quantityIndex of number of hours of sunshineThe theoretical number of hours of power generation per station is calculated by the following formula (1): (1); Wherein, the Is the theoretical number of hours of power generation for a wind farm or photovoltaic power plant,Is the total power generation amount in the current month,Is the total electricity limiting amount in the current month,And (5) the actual grid-connected capacity of the power station. The specific process of the step 2 is as follows: step 2.1, detecting outliers by using an abnormal data identification method based on the mahalanobis distance; (2); Wherein, the The distance of the mahalanobis is indicated,The vector of the samples is represented and,The mean value of the sample is represented,Representing covariance matrix between samples whenWhen the matrix is a unit matrix, the Mahalanobis distance is equal to the Euclidean distance; Step 2.2, setting a center-to-center distance MINDISTANCE, detecting a calculation result obtained in the formula (2), and when the center-to-center distance of the calculation result is smaller than MINDISTANCE, indicating that the theoretical number of power generation hours per month is abnormal, modifying the theoretical number of power generation hours per month, and entering step 2.4, otherwise entering step 2.3; Step 2.3, setting a center-to-center distance maxDistance, detecting the result obtained in step 2.2, and when the theoretical number of hours of power generation per month is greater than maxDistance, indicating that the theoretical number of hours of power generation per month is abnormal, modifying the theoretical number of hours of power generation per month, and entering step 2.4, otherwise entering step 2.5; Step 2.4, if abnormal data exist, correcting the abnormal data by using an average value correction method, wherein the method is specifically shown as a formula (3): (3) ; Wherein, the In the event of an abnormal data set,AndTwo data adjacent to the abnormal data; and 2.5, weighting each new energy station index obtained after the correction in the step 2.4, wherein the new energy station index is shown in the following formula (4): (4); Wherein, the Is an index of the average wind speed,Is an index of the average temperature of the water,Is an index of the radiation quantity, and is used for measuring the radiation quantity,Is an index of the number of hours of sunlight,And the correction factor, q, represents the comprehensive evaluation index value of the new energy station. The specific process of the step 3 is as follows: step 3.1, step 2.5 The data adopts a neuron algorithm to calculate a discrimination function