Typical feature mining and load evaluation method, system, equipment, medium and product

Abstract

The invention discloses a typical feature mining and load evaluation method, system and device, a medium and a product, and relates to the technical field of power systems, and the method comprises the steps: obtaining a plurality of types of typical demand features of a power system in a period; clustering each type of typical demand features to obtain a clustering result; according to the number of the category groups, weighting each clustering category to obtain a weighting result; constructing a weighted shortest distance optimization model according to an empowerment result, and calculating an estimated typical value of each type of typical demand features; calculating a demand side load mode in the demand time period; performing similarity matching calculation on the demand side load mode and the demand load curve, and constructing a load contribution evaluation space according to the similarity matching calculation; and finally, evaluating the load contribution by adopting a TOPSIS model to obtain the overall contribution degree of the load. According to the method, the contribution condition of the effective information of each feature sample to the typical feature value can be effectively determined, so that the typical demand time and the load demand curve are reasonably checked.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a typical feature mining and load evaluation method, system, equipment, medium and product. Background Art

[0002] The new power system with new energy as the main body has significant high randomness problems on both the source and load sides, which has aggravated the imbalance trend between supply and demand, and the pressure on the flexible resource regulation of the power system has become more serious. The regulation and effect evaluation of demand-side resources have advantages in promoting the balance of supply and demand, but how to reasonably characterize the contribution of demand-side resources to the power balance regulation needs of the power grid has become a key issue.

[0003] Current research mainly focuses on the effect of demand response, evaluating contributions through electricity price guidance or baseline calculation of response capacity, but there are problems such as momentary fluctuations and passive user participation. Current time period verification research is easily affected by subjective factors, and is mostly conducted at the power grid system level, which requires high computing resources. The volatility of new energy output and load power consumption leads to periodic differences in demand periods, and it is necessary to mine relatively stable demand periods and load demand curves. For the mining of typical values ​​of random feature samples, existing research is mainly based on statistical features or machine learning, but the penetration of new energy and the increase in load randomness make the statistical features unclear, and current research does not involve effective information of abnormally distributed samples. Summary of the invention

[0004] The purpose of this application is to provide a typical feature mining and load evaluation method, system, equipment, medium and product, which can effectively clarify the contribution of the effective information of each feature sample to the typical feature value, so as to reasonably determine the typical demand time and load demand curve.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a typical feature mining and load evaluation method, the typical feature mining and load evaluation method comprising:

[0007] Acquire several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend;

[0008] Clustering each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category;

[0009] Assigning weights to each of the cluster categories according to the number of category groups to obtain a weighted result;

[0010] Constructing a weighted shortest distance optimization model according to the weighted results, and calculating the estimated typical value of each type of typical demand characteristics by using the weighted shortest distance optimization model;

[0011] Calculating a demand-side load pattern during a demand period based on the estimated typical value;

[0012] Performing a similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result;

[0013] Constructing a load contribution evaluation space according to the similarity matching calculation result;

[0014] The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

[0015] In a second aspect, the present application provides a typical feature mining and load evaluation system, the typical feature mining and load evaluation system comprising:

[0016] Demand feature calculation module, used to:

[0017] Acquire several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend;

[0018] Clustering each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category;

[0019] Assigning weights to each of the cluster categories according to the number of category groups to obtain a weighted result;

[0020] Constructing a weighted shortest distance optimization model according to the weighted results, and calculating the estimated typical value of each type of typical demand characteristics by using the weighted shortest distance optimization model;

[0021] DLC matching module for:

[0022] Calculating a demand-side load pattern during a demand period based on the estimated typical value;

[0023] Performing a similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result;

[0024] Load contribution evaluation module, used for:

[0025] Constructing a load contribution evaluation space according to the similarity matching calculation result;

[0026] The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

[0027] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the typical feature mining and load evaluation methods described above.

[0028] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the typical feature mining and load evaluation methods described above.

[0029] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the typical feature mining and load evaluation methods described above.

[0030] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0031] The present application provides a typical feature mining and load evaluation method, system, equipment, medium and product, the method comprising: obtaining several types of typical demand characteristics of a power system within a cycle; the typical demand characteristics include: power demand period and power demand trend; clustering each type of the typical demand characteristics respectively to obtain clustering results; the clustering results include: cluster categories and the number of category groups within each cluster category; weighting each of the cluster categories according to the number of category groups to obtain weighting results; constructing a weighted shortest distance optimization model according to the weighting results, and using the weighted shortest distance optimization model to calculate the estimated typical value of each type of the typical demand characteristics; calculating the demand side load pattern within the demand period according to the estimated typical value; performing similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result; constructing a load contribution evaluation space according to the similarity matching calculation result; using the TOPSIS model to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution. This application determines the demand characteristics based on the typical demand characteristics calculation method of clustering weights, realizes the trend similarity matching calculation between the demand-side load pattern and the DLC during the demand period, and then constructs the load contribution evaluation space to obtain the overall load contribution. It can effectively evaluate the load contribution and realize the support level evaluation of the demand-side electricity consumption behavior for the system power balance, thereby effectively alleviating the pressure on the flexibility resource regulation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 A flowchart of a typical feature mining and load evaluation method provided in one embodiment of the present application;

[0034] Figure 2 A distribution diagram of the start and end times of a monthly demand period provided in an embodiment of the present application;

[0035] Figure 3 A distribution diagram of the maximum upward trend and the maximum downward trend of the monthly net load provided in an embodiment of the present application;

[0036] Figure 4 A schematic diagram of the trend segment weighting principle provided in an embodiment of the present application;

[0037] Figure 5 A schematic diagram of typical value estimation based on group effect contribution provided in one embodiment of the present application;

[0038] Figure 6 A typical value calculation flow chart based on group effect weighting provided in one embodiment of the present application;

[0039] Figure 7 A clustered data graph of the start and end times of a monthly demand period provided in an embodiment of the present application;

[0040] Figure 8 A monthly typical demand period diagram provided for an embodiment of the present application;

[0041] Fig. 9 A diagram of the number of categories under the time dimension provided in an embodiment of the present application;

[0042] Fig.10 The net load curve and typical curve diagram for February provided in one embodiment of the present application;

[0043] Fig.11 A typical value load curve provided in an embodiment of the present application and a comprehensive average distance diagram of the validation set;

[0044] Fig.12 A load aggregator evaluation result diagram provided in an embodiment of the present application;

[0045] Fig.13 A flowchart of the overall structure of the present application is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] The new power system with new energy as the main body has significant high randomness problems on both the source and load sides, which has aggravated the imbalance trend between supply and demand, and the pressure on the flexible resource regulation of the power system has become more serious. The regulation and effect evaluation of demand-side resources have advantages in promoting the balance of supply and demand, but how to reasonably characterize the contribution of demand-side resources to the power balance regulation needs of the power grid has become a key issue.

[0048] Current research mainly focuses on the effect of demand response, and evaluates the contribution through electricity price guidance or baseline calculation of response capacity, but there are problems such as momentary fluctuations and passive participation of users. The new energy power system is affected by the weather, and the system power balance regulation demand is highly random and volatile, making it difficult to mine the typicality of load demand time periods and demand load curves. The current time period verification research is easily affected by subjective factors, and is mostly carried out at the power grid system level, which requires high computing resources. The volatility of new energy output and load power consumption leads to periodic differences in demand time periods, and it is necessary to mine relatively stable demand time periods and load demand curves. For the typical value mining of random feature samples, existing research is mainly based on statistical features or machine learning, but the penetration of new energy and the increase in load randomness make the statistical features unclear, and the typical features are difficult to verify, and the current research does not involve the effective information of abnormal distribution samples. Therefore, how to effectively clarify the contribution of the effective information of each feature sample to the typical feature value, so as to reasonably verify the typical demand time and load demand curve, has become the key to effectively evaluate the load contribution.

[0049] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0050] In an exemplary embodiment, Figure 1 As shown, a typical feature mining and load evaluation method is provided, which is executed by a computer device, specifically, it can be executed by a computer device such as a terminal or a server alone, or it can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to a server as an example for explanation, including the following steps S1 to S8. Among them:

[0051] S1. Obtain several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend. The power demand period includes: the start time sequence of the valley and peak demand periods; the power demand trend includes: the maximum climbing trend during the valley, the maximum descending trend during the valley, the maximum climbing trend during the peak, and the maximum descending trend during the peak.

[0052] The obtaining of several types of typical demand characteristics of the power system within a cycle specifically includes: obtaining the differences in demand characteristics of the power system between months and the adaptability of demand-side users to typical grid demands within a monthly cycle to obtain typical data; according to the typical data, calculating the power demand period based on the response capability of demand-side users to provide 5% of the maximum load; according to the typical data, determining the power demand trend based on the distribution of trend characteristics within the power demand period of the monthly cycle.

[0053] This embodiment is based on the differences in system demand characteristics between months, while taking into account the adaptability of demand-side users to the typical demand of the power grid within the monthly cycle, and adopts a monthly cycle to analyze the typical demand characteristics of the system. Typical demand characteristics mainly include demand time periods, start and end time sequences of valley and peak demand periods within the monthly cycle, etc., among which the maximum climbing trend and maximum descending trend of daily valley and peak hours will be further mined by analyzing the distribution of power system demand characteristics within the monthly cycle. These trends are obtained by extracting the load data of daily valley and peak hours within the month, and then analyzing their maximum change trends. Specifically, a specific method, such as normal distribution fitting analysis, is used to reflect the distribution of demand trends, so as to clarify the typical power regulation demand within the monthly cycle. The maximum climbing trend and maximum descending trend of daily valley and peak hours are calculated based on the change in the load data of the power system during valley and peak hours and the time required for the change, and these data are all derived from the actual operation records of the power grid system.

[0054] The typical distribution of demand characteristics was studied in the double months (2, 4, 6, 8, 10, 12) of a regional power system in a year. The power demand period T was calculated based on the response capability of the demand side to provide 5% of the maximum load. k1 and T k2 , the distribution of the trend characteristics in the monthly cycle demand period is used to reflect the distribution of the demand trend, the maximum climbing trend and the maximum descending trend in the monthly valley and peak hours are extracted, and the distribution of the maximum trend is analyzed by normal distribution fitting, such as Figure 2 and Figure 3 shown.

[0055] S2. Cluster each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category.

[0056] Each type of the typical demand characteristics is clustered separately to obtain clustering results, specifically including: for one type of the typical demand characteristics: using the silhouette coefficient method to determine the number of clusters of the typical demand characteristics; according to the number of clusters, clustering the typical demand characteristics through the K-means clustering algorithm to obtain clustering results.

[0057] S3. Weight each of the cluster categories according to the number of category groups to obtain a weighted result.

[0058] S4. Construct a weighted shortest distance optimization model according to the weighted results, and use the weighted shortest distance optimization model to calculate the estimated typical value of each type of typical demand characteristics.

[0059] In this embodiment, the basic idea of ​​typical estimation of group effect is as follows: Figure 4 As shown in the figure, the demand feature samples are clustered by the K-means clustering algorithm, and the silhouette coefficient method (SC) is used as a typical clustering quality evaluation index to determine the number of clusters. The number of category groups is determined according to the clustering category, and each category is weighted. The weighted shortest distance optimization model is constructed using the following formula to obtain the estimated typical value of the demand feature, as shown in Figure 5 shown.

[0060]

[0061] Among them, x m is the demand feature sample in the sequence {X}, x * is the contribution function Φ of the sample points m (x m ) under the comprehensive value, e m For a certain demand feature sample x m Distance Assumption Typical Value x d The distance between m is the mth demand feature in the feature sequence, C k is the kth cluster center of the feature sequence, w k is the probability of the k-th sample, N k is the number of the kth class, M is the number of samples, x k,i ∈{X k} is the total feature sample X belonging to category C k The feature sample, x k,i is the k-th feature X k The i-th feature sample in , F is the comprehensive shortest distance objective function considering all category samples under the weighted situation, G is the distance between the category sample and the typical value under the weighted situation; e k,i For a certain demand feature sample x k Distance Assumption Typical Value x iThe distance between them, C is the cluster center of the feature sequence, and F is the comprehensive shortest distance objective function considering all categories of samples under the weighted situation.

[0062] When determining the number of clusters, this embodiment uses the Silhouette Coefficient (SC) method as a clustering quality evaluation indicator. The silhouette coefficient indicates the closeness between each type of sample and the degree of discreteness between each type after load clustering. The larger the silhouette coefficient, the better the clustering effect. Therefore, by calculating the silhouette coefficient under different numbers of clusters, and selecting the number of clusters corresponding to the number of times with a larger silhouette coefficient as the optimal number of clusters. At the same time, it is also necessary to set the upper and lower limits of the number of categories based on the silhouette coefficient method to avoid excessive segmentation of cluster data, which is not conducive to discovering common characteristics between categories. This method can ensure the rationality and effectiveness of the clustering results, thereby providing an accurate basis for the subsequent calculation of typical demand characteristics.

[0063] For the demand feature sequence, once the clustering category is obtained, the number of category groups is determined according to the clustering category, and each category is weighted by the group effect. The category group effect here is the frequency of occurrence of the category group in the feature dimension, that is, the contribution trend of the category to the typical value.

[0064] After obtaining the number of categories after clustering the demand features, we can calculate the probability of each type of sample appearing in the feature dimension. If the probability of a certain category appearing is low, it means that the effective information contained in this category contributes less to the typical value, and the impact of this category on the typical value should be reduced; on the contrary, if the probability of a certain category appearing is high, the contribution of the effective information contained in this category to the typical value is high, and the impact of this category on the typical value should be considered more.

[0065] S5. Calculate the demand-side load pattern within the demand period according to the estimated typical value.

[0066] S6. Perform similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result.

[0067] The demand side load pattern within the demand period is calculated according to the estimated typical value; a similarity matching calculation is performed on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result, specifically including: determining the valley demand period and the peak demand period according to the estimated typical value with the maximum load regulation capacity provided by the demand side resources as the target; calculating the trend feature weight according to the valley demand period and the peak demand period; calculating the matching ability of the demand load curve and the load curve to be evaluated according to the trend feature weight to obtain a similarity matching calculation result.

[0068] Taking the maximum load regulation capacity provided by demand-side resources as the target, determine the low demand period T k1and peak demand period T k2 In order to reduce the regulation pressure of flexible resources within the time period, ideally, the demand-side resource load pattern is required to show a trend opposite to the net load curve within the demand period. The load pattern with the opposite trend is defined as the demand load curve (DCL).

[0069] The imbalance of power supply and demand in the system is likely to lead to insufficient periodic regulation capacity. Determining this period will help the system's adjustable resources alleviate the regulation pressure of power balance. This application takes the maximum load regulation capacity provided by demand-side resources as the goal and determines the demand period:

[0070]

[0071] Where: T k1 T is the low demand period. k2 is the peak demand period, t i is the demand period interval; JH is the net load curve; load is the total load on the demand side, α is the proportion of the regulation capacity provided by the demand side resources to the total load on the demand side. According to the current grid requirements for demand side resources, α is set to 5%.

[0072] In order to reduce the regulation pressure of flexible resources during the period, ideally, the demand-side resource load pattern is required to show an opposite trend to the net load curve during the demand period. The load pattern with the opposite trend is defined as the demand load curve (DLC), as shown in the following formula (the demand load curve is a graph that describes the change of power load over time and is an important basis for power system planning and scheduling):

[0073]

[0074] Among them, L ki is the demand load curve; is the net load during the demand period; L1 is the load demand curve during the off-peak period; L2 is the load demand curve during the peak period; J is the net load.

[0075] For the problem of periodic power balance regulation, if the power consumption pattern of the demand-side load is similar to DLC during the demand period, it can be considered to be helpful in smoothing the net load fluctuation, and vice versa. Furthermore, considering the amplitude difference factor of DLC fluctuation during the period, if the load power consumption curve is similar to DLC and the corresponding amplitude difference is close, the more the user load contributes, the more beneficial it can be considered to be in smoothing the net load fluctuation, which can greatly alleviate the supply and demand regulation needs of the power grid.

[0076] According to the climbing trend and descending trend of the demand load curve, the load demand curve is divided into two trends of double rising and double falling, and four trend segments. The trend characteristics are expressed in the form of incremental slope, and then the weight of the trend period is calculated. The specific process is shown in Figure 6 .

[0077] This application describes the morphological similarity based on the idea of ​​trend segmentation, and combines the distance method to calculate the numerical similarity, and proposes a trend segment weighting (TSW) method to calculate the final trend correlation.

[0078] The demand load curve is divided into sections according to its climbing and descending trends. The load curve is divided into four trend sections, including two trends: double rising (period t1, t3) and double descending (period t2, t4). In order to quantify the load change trend of the "duration-trend" relationship in each trend section, the trend characteristics are expressed in the form of incremental slope. See the following formula for details:

[0079]

[0080] Among them, L ti For the tth i Load curve for a trend period, i = {1,3,L,I}.

[0081] The characteristic trend vector is:

[0082] TR=[tr1,tr2,L,tr i ];

[0083] (3) According to the trend characteristics of each trend segment within the demand period, the weight of the trend period is calculated. The more obvious the trend characteristics are, the greater the weight is. :

[0084] W tr =TR / sum(TR);

[0085] Among them, W tr is the weight vector of the trend segment.

[0086] After normalizing the load curve to be evaluated and DLC, the Euclidean distance is used to calculate the similarity between the load to be evaluated and DLC in the trend segment. The Euclidean distance weighted by the trend segment is considered to be the sum of the product of the weight and the Euclidean distance in each trend segment, and then the matching ability F1 of the demand load curve and the load curve to be evaluated is calculated to realize the trend similarity matching between the demand side load pattern and DLC in the demand period.

[0087] In this application, the load curve data to be evaluated is obtained in real time or periodically from the power system's monitoring and data acquisition system (SCADA), smart grid management system, or certified power metering equipment during the daily operation of the power system. The specific acquisition time depends on the evaluation requirements and may cover multiple continuous operation cycles to ensure the comprehensiveness and representativeness of the data. These load curve data provide basic support for subsequent key steps such as load characteristic analysis, time division, and energy storage optimization configuration, thereby effectively supporting the optimal scheduling and energy efficiency improvement of the power system under periodic supply and demand imbalances.

[0088] Calculate demand periods: Based on the maximum load regulation capability provided by demand-side resources, use the following formula to determine the low demand period and peak demand period.

[0089] The imbalance of power supply and demand in the system is likely to lead to insufficient periodic regulation capacity. Determining this period will help the system's adjustable resources alleviate the regulation pressure of power balance. This application takes the maximum load regulation capacity provided by demand-side resources as the goal and determines the demand period:

[0090]

[0091] Where: T k1 T is the low demand period. k2 is the peak demand period, t i is the demand period interval; JH is the net load curve; load is the total load on the demand side, α is the proportion of the regulation capacity provided by the demand side resources to the total load on the demand side. According to the current grid requirements for demand side resources, α is set to 5%.

[0092] In order to reduce the regulation pressure of flexible resources during the period, ideally, the demand-side resource load pattern is required to show an opposite trend to the net load curve during the demand period. The load pattern with the opposite trend is defined as the demand load curve (DLC), which is specifically expressed as

[0093]

[0094] Among them, L ki is the demand load curve; is the net load during the demand period.

[0095] Calculate the trend feature weight, according to the low demand period and peak demand period determined by the demand period calculation submodule, and use the following formula to calculate the trend feature weight.

[0096] For the problem of periodic power balance regulation, during the demand period, if the demand side load power consumption pattern is similar to DLC, it can be considered to be helpful in smoothing the net load fluctuation, and vice versa. Further, considering the amplitude difference factor of DLC fluctuation during the period, if the load power consumption curve is similar to DLC and the corresponding amplitude difference is similar, the more the user load contributes, the more beneficial it can be considered to be in smoothing the net load fluctuation, which can greatly alleviate the power grid supply and demand regulation needs. The matching ability of the demand side load power consumption pattern and DLC is measured as follows:

[0097]

[0098] Among them, F1 is the matching value between load curve and DLC; f trend is the trend similarity measure after load normalization, f 1,m is the amplitude difference ratio between the load curve and DLC, TR is the characteristic trend vector, L is the load curve value, L k is the load demand curve of the kth trend period, k=1 is the load demand curve of the valley period, and k=2 is the load demand curve of the peak period. The characteristic trend vector is a quantitative representation used to describe the load change trend of the demand load curve in each trend segment. It is based on the idea of ​​piecewise linear representation of the load curve, and extracts the trend characteristics of the load curve by calculating the slope of the load increment in each trend segment. These characteristic trend vectors play an important role in subsequent steps such as weight calculation and trend similarity matching, which helps to accurately characterize the power balance regulation demand of the power system in the demand period.

[0099] The demand load curve is divided into sections according to its climbing and descending trends. The load curve is divided into four trend sections, including two trends: double rising (period t1, t3) and double descending (period t2, t4). In order to quantify the load change trend of the "duration-trend" relationship in each trend section, the trend characteristics are expressed in the form of incremental slope, as shown in the following formula:

[0100]

[0101] in, For the tth i Load curve for a trend period, i = {1,3,L,I}.

[0102] The characteristic trend vector is:

[0103] TR=[tr1,tr2,L,tr i ];

[0104] According to the trend characteristics of each trend segment within the demand period, the weight of the trend period is calculated:

[0105] W tr =TR / sum(TR);

[0106] W tr is the weight vector of the trend segment.

[0107] Calculate the load curve trend matching capability: It is used to calculate the matching capability between the demand load curve and the load curve to be evaluated based on the trend feature weight calculated by the trend feature weight calculation submodule, using the following formula;

[0108] Normalize the load curve to be evaluated and DLC:

[0109] x' i =(x i -min(X)) / (max(X)-min(X));

[0110] Where X is the corresponding load value in the load curve. i} is the normalized data, {x i} is the original data.

[0111] The Euclidean distance is used to calculate the similarity between the load to be evaluated and the DLC in the trend segment:

[0112]

[0113] in, For the tth i The peak / valley load curve of the trend period, k = 1, is the load demand curve of the valley period, and k = 2 is the load demand curve of the peak period. i It is the Euclidean distance between the normalized demand load curve and the load to be evaluated.

[0114] The distance vector is:

[0115] D=[d1,d2,L,d i ];

[0116] The Euclidean distance considering the trend segment weighting during the demand period is the sum of the product of the weight in each trend segment and the Euclidean distance:

[0117] f trend =W tr ·D T ;

[0118] Therefore, the matching ability between the demand load curve and the load curve to be evaluated is obtained as follows:

[0119] F1=W tr ·D T ·f 1,m ;

[0120] {x' i} is the normalized data, {x i} is the original data, d i is the Euclidean distance between the normalized demand load curve and the load to be evaluated, D is the distance vector, and f trend is the Euclidean distance considering the weight of the trend segment

[0121] S7. Constructing a load contribution evaluation space according to the similarity matching calculation result.

[0122] S8. Use the TOPSIS model to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

[0123] The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution, specifically including: calculating the contribution of the stable component of the load during the demand period to the net load mean, which is used to measure the peak-valley support capacity of the demand-side load for the net load; calculating the relationship between the ratio of the load mean during the demand period to the load mean during the entire period and the ratio of the net load mean during the demand period to the net load mean during the entire period, which is used to measure the load's ability to reverse peak and valley.

[0124] According to the demand characteristics, the contribution of the stable component of the load to the mean net load during the demand period is calculated to measure the peak and valley support capacity of the demand-side load for the net load;

[0125] Peak-valley support capacity refers to the peak-valley support capacity of the demand-side load for the net load. It is measured by the contribution of the stable component of the load during the demand period to the mean net load, as expressed by the formula:

[0126]

[0127] In the above formula, is the ideal mean demand, F 2,1 The bigger the better, F 2,2 The smaller the better, F 2,1 is the peak-to-valley support capacity during the valley period, F 2,2 is the peak-to-valley support capacity during peak hours, l k1 is the load curve during the low demand period, l k2 is the load curve during peak demand period, is the net load curve during the low demand period, is the net load curve during peak demand period.

[0128] Peak-valley support capacity refers to the support capacity of the demand-side load to the net load during peak and valley periods, which is measured by calculating the contribution of the steady component of the load during the demand period to the mean net load. Specifically, there are two equations for measuring the peak (F 2.1 ) and valley value (F 2.2 ) period. Among them, F 2.1It is the ratio of the minimum load to the ideal average demand (i.e. the net load average minus the minimum net load value) during the demand period. The larger the better. 2.2 It is the ratio of another minimum load (considering different time periods or conditions) during the demand period to (minimum net load value minus net load average), the smaller the better. These two equations help evaluate the ability of demand-side load in alleviating the imbalance between supply and demand of the power grid.

[0129] The relationship between the ratio of the average load in the demand period to the average load in the entire period and the ratio of the average net load in the demand period to the average net load in the entire period is calculated according to the demand characteristics, which is used to measure the load peak-valley capability.

[0130] Peak-valley capacity. The relationship between the ratio of the average load in the demand period to the average load in the entire period and the ratio of the average net load in the demand period to the average net load in the entire period is expressed as follows:

[0131]

[0132] Where JH is the net load curve, For the T ki Net load curve for each period, T k1 T is the low demand period. k2 is the peak demand period, t i is the demand period interval, For the T k The load curve of each time period, k = 1, is the load demand curve of the valley period, k = 2 is the load demand curve of the peak period, is the net load mean ratio; f 3,i is the ratio of the load demand period average to the full period average. The first equation F 3,i It represents the load evaluation index of a certain period, which is equal to the net load J of that period. s,i Subtract a ratio obtained by dividing a certain value of the net load by its mean value (i.e., the net load mean ratio mean(JHr,) / JHmean). The second equation f 3,i is a ratio that represents the ratio of the average value of load demand during the period to the average value of load during the entire period (i.e., the ratio of the mean value of load demand during the period to the mean value of load demand during the entire period, mean(lr.) / mean(l)). These equations and ratios are used to evaluate the changes in power load during different periods of time and their impact on the power balance of the power grid.

[0133] The load DCL matching capacity is calculated based on the trend similarity matching between the demand-side load pattern and the DLC during the demand period. Together with the load peak-valley support capacity and load reverse peak-valley capacity obtained above, the three construct the load contribution evaluation space, and the TOPSIS model is used to evaluate the load contribution, which is used to evaluate the overall load contribution.

[0134] The load DCL matching capacity is obtained by matching calculation:

[0135] Calculate demand periods: Based on the maximum load regulation capability provided by demand-side resources, use the following formula to determine the low demand period and peak demand period.

[0136] The imbalance of power supply and demand in the system is likely to lead to insufficient periodic regulation capacity. Determining this period will help the system's adjustable resources alleviate the regulation pressure of power balance. This application takes the maximum load regulation capacity provided by demand-side resources as the goal and determines the demand period:

[0137]

[0138] Where: T k1 T is the low demand period. k2 is the peak demand period, t i is the demand period interval; JH is the net load curve; load is the total load on the demand side, α is the proportion of the regulation capacity provided by the demand side resources to the total load on the demand side. According to the current grid requirements for demand side resources, α is set to 5%.

[0139] In order to reduce the regulation pressure of flexible resources during the period, ideally, the demand-side resource load pattern is required to show an opposite trend to the net load curve during the demand period. The load pattern with the opposite trend is defined as the demand load curve (DLC), which is specifically expressed as

[0140]

[0141] Among them, L ki is the demand load curve; is the net load during the demand period.

[0142] Calculate the trend feature weight, according to the low demand period and peak demand period determined by the demand period calculation submodule, and use the following formula to calculate the trend feature weight.

[0143] For the problem of periodic power balance regulation, during the demand period, if the demand side load power consumption pattern is similar to DLC, it can be considered to be helpful in smoothing the net load fluctuation, and vice versa. Further, considering the amplitude difference factor of DLC fluctuation during the period, if the load power consumption curve is similar to DLC and the corresponding amplitude difference is similar, the more the user load contributes, the more beneficial it can be considered to be in smoothing the net load fluctuation, which can greatly alleviate the power grid supply and demand regulation needs. The matching ability of the demand side load power consumption pattern and DLC is measured as follows:

[0144]

[0145] Among them, F1 is the matching value between load curve and DLC; f trend is the trend similarity measure after load normalization, f 1,m The amplitude difference ratio TR between the load curve and DLC is the characteristic trend vector.

[0146] The demand load curve is divided into sections according to its climbing and descending trends. The load curve is divided into four trend sections, including two trends: double rising (period t1, t3) and double descending (period t2, t4). In order to quantify the load change trend of the "duration-trend" relationship in each trend section, the trend characteristics are expressed in the form of incremental slope, as shown in the following formula:

[0147]

[0148] Among them, L ti For the tth i Load curve for a trend period, i = {1,3,L,I}.

[0149] The characteristic trend vector is:

[0150] TR=[tr1,tr2,L,tr i ];

[0151] According to the trend characteristics of each trend segment within the demand period, the weight of the trend period is calculated:

[0152] W tr =TR / sum(TR);

[0153] W tr is the weight vector of the trend segment.

[0154] Calculate the load curve trend matching capability: It is used to calculate the matching capability between the demand load curve and the load curve to be evaluated based on the trend feature weight calculated by the trend feature weight calculation submodule, using the following formula;

[0155] Normalize the load curve to be evaluated and DLC:

[0156] x' i =(x i -min(X)) / (max(X)-min(X));

[0157] The Euclidean distance is used to calculate the similarity between the load to be evaluated and the DLC in the trend segment:

[0158]

[0159] The distance vector is:

[0160] D=[d1,d2,L,d i];

[0161] The Euclidean distance considering the trend segment weighting during the demand period is the sum of the product of the weight in each trend segment and the Euclidean distance:

[0162] f trend =W tr ·D T ;

[0163] Therefore, the matching ability between the demand load curve and the load curve to be evaluated is obtained as follows:

[0164] F1=W tr ·D T ·f 1,m ;

[0165] {x' i} is the normalized data, {x i} is the original data, d i is the Euclidean distance between the normalized demand load curve and the load to be evaluated, D is the distance vector, and f trend is the Euclidean distance that takes into account the weight of the trend segment.

[0166] This embodiment provides a typical feature mining and load evaluation method, including the following steps: according to the distribution of the power system demand characteristics within the monthly cycle, the complexity distribution of the monthly demand characteristics of the power system is characterized, the daily demand characteristics within the monthly cycle are extracted, and the demand characteristics are determined by the typical demand feature calculation method based on clustering weighting; according to the demand characteristics determined by the typical demand feature calculation method based on clustering weighting, the trend characteristics of the demand curve are calculated and analyzed, and the trend similarity matching calculation between the demand side load pattern and DLC within the demand period is completed; according to the trend similarity matching calculation between the demand side load pattern and DLC within the demand period, the load DCL matching capability is obtained, the load contribution evaluation space is constructed, the load contribution is evaluated by the TOPSIS model, and then the overall load contribution is obtained. The present application can effectively evaluate the load contribution, realize the support level evaluation of the demand side power consumption behavior to the system power balance, and thus effectively alleviate the pressure on the flexibility resource regulation of the power system.

[0167] Based on the same inventive concept, the embodiment of the present application further provides a typical feature mining and load evaluation system for implementing the typical feature mining and load evaluation method involved above, the typical feature mining and load evaluation system comprising:

[0168] Demand feature calculation module, used to:

[0169] Acquire several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend;

[0170] Clustering each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category;

[0171] Assigning weights to each of the cluster categories according to the number of category groups to obtain a weighted result;

[0172] A weighted shortest distance optimization model is constructed according to the weighted results, and the weighted shortest distance optimization model is used to calculate the estimated typical value of each type of typical demand characteristics.

[0173] DLC matching module for:

[0174] Calculating a demand-side load pattern during a demand period based on the estimated typical value;

[0175] A similarity matching calculation is performed on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result.

[0176] Load contribution evaluation module, used for:

[0177] Constructing a load contribution evaluation space according to the similarity matching calculation result;

[0178] The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

[0179] Specifically,

[0180] The demand characteristic calculation module is used to extract the daily demand characteristics within the monthly cycle based on the monthly net load data input, and determine the demand characteristics using a typical demand characteristic calculation method based on clustering weighting.

[0181] The demand feature calculation module specifically includes:

[0182] According to the idea of ​​"group effect", based on the input of monthly net load data, the daily demand characteristics within the monthly cycle are extracted, and the typical characteristic values ​​are calculated using the method of minimum comprehensive distance. At the same time, in order to take into account the contribution of the differential effective information of each sample point of demand characteristics to the typical load value, the cluster-category weighting method is used to weight each characteristic sample, and the demand characteristic samples are clustered by the K-means clustering algorithm. The silhouette coefficient method (SC) is used as a typical clustering quality evaluation index to determine the number of clusters, and the number of category groups is determined according to the cluster category. The weighted shortest distance optimization model is constructed using the formula to obtain the estimated typical value.

[0183] The DLC matching module is used to calculate and analyze the trend characteristics of the demand curve according to the demand characteristics determined by the typical demand characteristic calculation method based on clustering weighting, and complete the trend similarity matching calculation between the demand side load pattern and the DLC during the demand period.

[0184] The DLC matching module specifically includes:

[0185] The demand period calculation submodule is used to determine the valley demand period and the peak demand period according to the maximum load regulation capacity provided by the demand side resources using the following formula;

[0186]

[0187] Where: T k1 T is the low demand period. k2 is the peak demand period, t i is the demand period interval; JH is the net load curve; load is the total load on the demand side, α is the proportion of the regulation capacity provided by the demand side resources to the total load on the demand side. According to the current grid requirements for demand side resources, a is set to 5%, L ki is the demand load curve; JH Tki is the net load during the demand period.

[0188] The trend feature weight calculation submodule is used to calculate the trend feature weight using the following formula according to the low demand period and peak demand period determined by the demand period calculation submodule.

[0189]

[0190]

[0191] TR=[tr1,tr2,L,tr i ];

[0192] W tr =TR / sum(TR);

[0193] Among them, F1 is the matching value between load curve and DLC; f trend is the trend similarity measure after load normalization, f 1,m is the amplitude difference ratio between the load curve and DLC, For the tth i The load curve of a trend period, i = {1, 3, L, I}, TR is the characteristic trend vector, W tr is the weight vector of the trend segment.

[0194] The load curve trend matching submodule is used to calculate the matching ability between the demand load curve and the load curve to be evaluated according to the trend feature weight calculated by the trend feature weight calculation submodule using the following formula;

[0195] x' i =(x i -min(X)) / (max(X)-min(X));

[0196]

[0197] D=[d1,d2,L,d i ];

[0198] f trend =W tr ·D T ;

[0199] F1=W tr ·D T ·f 1,m ;

[0200] {x' i} is the normalized data, {x i} is the original data, d i is the Euclidean distance between the normalized demand load curve and the load to be evaluated, D is the distance vector, and f trend is the Euclidean distance that takes into account the weight of the trend segment.

[0201] The load contribution evaluation module is used to construct a load contribution evaluation space based on the trend similarity matching calculation between the demand-side load pattern and the DLC during the demand period, and use the TOPSIS model to evaluate the load contribution, thereby obtaining the overall load contribution.

[0202] The load contribution evaluation module specifically includes:

[0203] The peak-valley support capacity module calculates the contribution of the stable component of the load to the mean net load during the demand period according to the demand characteristic calculation module, and is used to measure the peak-valley support capacity of the demand-side load to the net load.

[0204] The peak-valley capacity module calculates the relationship between the ratio of the load average during the demand period to the load average during the entire period and the ratio of the net load average during the demand period to the net load average during the entire period according to the demand characteristic calculation module, which is used to measure the load peak-valley capacity.

[0205] The load overall contribution module obtains the load DCL matching capacity according to the DLC matching module, the peak-valley support capacity module obtains the load peak-valley support capacity, and the reverse peak-valley capacity module obtains the load reverse peak-valley capacity. These three construct the load contribution evaluation space, and the TOPSIS model is used to evaluate the load contribution, which is used to evaluate the overall load contribution.

[0206] In an exemplary embodiment, the complexity distribution of the monthly demand characteristics of the power system is characterized.

[0207] In order to make the selected monthly cycle representative, this application selects the double months (2, 4, 6, 8, 10, 12) in a certain area's power system to carry out research on the typical distribution of demand characteristics.

[0208] Use normal distribution to fit the sample sequence and analyze the typical distribution of the start and end time series of the demand period within a month, such as Figure 2 shown. Figure 2 In the figure, a and b are the distribution of the start and end time of the valley demand period, respectively, and c and d are the distribution of the start and end time of the peak demand period, respectively. In the selected 6 months, the distribution range of the peak and valley start time is different and the mean difference is large, which fails to show an obvious centralized distribution; the distribution of the valley end time between months is relatively centralized, but there is a significant difference in the mean. In short, there are obvious differences in the mean and centralized distribution of the peak and valley start and end times between months, and there are obvious distribution outliers.

[0209] According to the calculation results that the change trend of net load in the demand period can directly affect the DLC matching capacity, the distribution of the demand trend is reflected by the distribution of the trend characteristics in the monthly cycle demand period. This application extracts the maximum climbing trend and the maximum descending trend during the daily valley and peak hours in the month, and uses the normal distribution fitting to analyze the distribution of the maximum trend, such as Figure 3 shown.

[0210] Figure 3 There are large differences in the mean and dispersion of the distribution of the maximum upward and downward trends in demand periods between months. Figure 3 In the figure, a and b are the maximum rising and falling trend distributions during valley hours, respectively, and c and d are the maximum rising and falling trend distributions during peak hours. The distribution of rising and falling trends during valley hours is relatively concentrated. Except for October, the distribution of rising and falling trends during valley hours in other months is obviously different, which means that the rising and falling trends in October are more symmetrical. At the same time, the mean and discreteness of the maximum rising and falling trends during peak hours are very different between months, indicating that there are random climbing and descending trends between months. Therefore, it is necessary to accurately calculate the typical net load curve during the monthly demand period to clarify the typical power regulation demand during the monthly cycle.

[0211] In an exemplary embodiment, a typical value calculation method based on group effect weighting is also provided:

[0212] Based on the distribution analysis of the power system demand characteristics within the monthly cycle, the daily demand characteristics within the monthly cycle are extracted.

[0213] According to the basic idea of ​​typical estimation of group effect, the demand feature samples are clustered by K-means clustering algorithm, and the number of clusters with larger silhouette coefficient is selected as the optimal number of clusters. The upper and lower limits of the number of categories are set based on the silhouette coefficient method, and the number of category groups is determined according to the clustering category, and each category is weighted.

[0214] According to the clustering-based weight verification, a weighted shortest distance optimization model is constructed, typical values ​​are estimated, and the typical value calculation based on group effect weighting is completed.

[0215] In an exemplary embodiment, a representative value calculation system based on group effect weighting includes:

[0216] The demand feature extraction module is used to extract the daily demand features within the monthly cycle based on the distribution analysis of the demand features of the power system within the monthly cycle.

[0217] The weight verification module is used to cluster the demand feature samples through the K-means clustering algorithm according to the basic idea of ​​typical estimation of group effect, evaluate the clustering effect and calculate the optimal number of clusters, select the number of clusters with the largest silhouette coefficient as the best number of clusters, set the upper and lower limits of the number of categories based on the silhouette coefficient method, determine the number of category groups according to the clustering category, and assign weights to each category.

[0218] The typical value calculation module is used to consider the comprehensive shortest distance of all category samples under weighted conditions according to the demand feature extraction module and the weight verification module, build a weighted shortest distance optimization model, estimate the typical value, and complete the typical value calculation based on group effect weighting.

[0219] In an exemplary embodiment, the data of this application comes from a provincial power grid data center in China. The wind and solar energy and load in this area are growing rapidly, and the imbalance between supply and demand needs to be adjusted. The data includes the output of new energy, total load and the data of three industrial load aggregators for the whole year (15-minute resolution), and the changes in grid demand and the contribution of industrial user load patterns to the balance of supply and demand are studied.

[0220] Example Analysis 1 Results and Analysis of Typical Value Estimation

[0221] Example Analysis 101 Typical Values ​​of Demand Period

[0222] In order to describe the contribution of the discrete distribution of the start and end time of each monthly demand to its typical value, the demand characteristic X m,s , X m,e Cluster analysis is used to reflect the differences in the distribution of monthly characteristics, such as Figure 7 As shown in the figure: the fewer the number of clusters, the more obvious the feature category information is; the more the number of clusters, the more discrete the feature sample distribution is and the more vague the category information is. Figure 7 There are obvious differences in the number of clusters in the start and end times of the selected monthly demand, indicating that there are obvious differences in the distribution of demand feature categories between months.

[0223] After obtaining the category information of the feature sample, the required feature X m,s , X m,e The typical value of is calculated, and the interval of peak and valley demand periods is calculated, where the valley and peak demand periods of the selected month are as follows: Figure 8 shown. Figure 8 The changes in the mid-peak period are more obvious. The peak demand period in June and August lasts longer, which is consistent with the situation that the temperature control load lasts longer in summer; the peak demand period in February, April, October, and December is relatively short, which is related to the reduction in temperature control load demand and the diverse heating methods in the region. Compared with the peak demand period, the difference in the valley demand period is relatively small (because the selected resolution is 15 minutes, the difference in the valley demand period between months is about 0.5-1 hour). This is because the electricity consumption is reduced at night and there is a certain amount of wind energy resources. In addition, Figure 8 The starting time of the valley demand in April, October and December is the same, Figure 7 It can be seen that the distribution of the starting time of valley demand in April is more complex, clustered into 8 categories, while October and December have more obvious category information, clustered into 2 categories. The distribution of the starting time of valley demand in these three months is quite different, and the weight information of each category is different, and its contribution to the typical value is also different. Therefore, although the distribution of characteristic samples is quite different, there will still be similar typical characteristic values.

[0224] Example Analysis 102 Typical Load Curve

[0225] According to the determined demand period, calculate the typical load curve in the demand period. This application selects the net load curve in the February valley demand period for category analysis in the time dimension, such as Fig. 9 As shown in the figure, the number of categories at each moment is different, that is, the distribution of load characteristics in the time period dimension is more complex.

[0226] The proposed typical value estimation method, mode method and mean method are used to calculate the typical load curve, such as Fig.10As shown in the figure, A is the classic method proposed in this paper, B is the majority method, and C is the mean method. Fig.10 Among them, method C can be regarded as the group effect calculation of the typical load curve of equal probability samples. The typical load curve is relatively smooth, but it ignores the effective information of the difference of the sample group on the typical value; method A assigns weights to the feature samples according to the clustering classification of the feature samples, and the calculated typical load curve has some fluctuations due to the influence of the clustered feature samples; method B pays too much attention to the category with the largest number of clustered samples and ignores the influence of other groups. Therefore, it shows violent fluctuations and is not suitable for estimating typical values ​​in this study.

[0227] Example Analysis 103 Typicality Verification

[0228] According to the need to verify the effectiveness of the typical value calculation method proposed in this article, this application selects the peak demand characteristics in April as the object, and uses three methods A, B, and C to compare the typical value calculation. The validation set is constructed by sampling with replacement, and samples are randomly selected to calculate the distance from the three typical values. After repeating the experiment 10 times, the comprehensive distance is calculated. The smaller the comprehensive distance, the stronger the representativeness. The obtained comprehensive average distance between the typical value and the validation sample set is shown in Table 1. In Table 3, the last column is the number of minimum occurrences of the comprehensive average distance between the three typical values ​​and the validation feature set after 10 experiments. The minimum number of comprehensive average distances of method A is 6 times, method B is 1 time, and method C is 3 times. Therefore, in the verification method proposed in this article, the typical value estimated by method A is more representative.

[0229] Table 1 Comprehensive average distance between typical values ​​and validation sample set

[0230]

[0231] Compared with the demand moment characteristics, the characteristics of the load curve pattern are more complex, so the number of experiments for the verification of the typical load curve is set to 30 times, and the other settings remain unchanged. The comprehensive average distances between the three typical load curves and the verification load curve are calculated respectively, and the results are shown in 11. Fig.11 In the figure, after 30 tests, the comprehensive average distances between the load curve extracted in each experiment and the three typical load curves are different. The number of times the comprehensive distance calculated by method B is the largest is obviously too many, so this method is not representative of the monthly load curve. At the same time, it can be seen that the number of times the comprehensive distance of method A is the smallest is more than that of method B. By counting the number of times the comprehensive distance of the three typical methods is the smallest, it can be seen that method A has a minimum comprehensive distance of 18 times, method B has a minimum distance of 1 time, and method C has a minimum distance of 11 times, so the typical load curve calculated by method A is more representative.

[0232] Case Analysis 2 Monthly Load Evaluation

[0233] Case Analysis 201DLC Trend Matching

[0234] This application uses the Euclidean distance and the method proposed in this paper to calculate the trend similarity f of load characteristics DLC trend , analyze the similarity measurement results. Select industrial load aggregator I as the research object, and still use the bimonthly demand load curve to study and analyze the f in the peak and valley demand period trend As shown in Table 2. In Table 2, compared with the Euclidean distance method, the f calculated by the proposed method is trend are smaller than f calculated by the Euclidean distance method trend This is because the proposed method gives a larger weight to the trend segment with a surge in the trend, while the Euclidean distance method does not highlight the need for a surge in the trend. Relatively speaking, the weight of the Euclidean distance of the trend segment with a stable trend is higher than that of the proposed method, while the length of the trend segment with a surge in the trend is relatively short. Therefore, the trend similarity of the TSW method proposed in this paper is lower than that of the Euclidean distance in Table 4.

[0235] Table 2 DLC trend similarity

[0236]

[0237] Example Analysis 202 Monthly Load Evaluation Results

[0238] According to the load evaluation characteristics of the calculated aggregator, the TOPSIS model is used to evaluate the friendliness of the industrial load aggregator to the power grid. In order to compare the impact of the TSW method and the Euclidean distance method, the DLC matching capability characteristics are calculated respectively. The evaluation results are shown in Fig.12 . Fig.12 In the figure, (a) is the DLC matching ability calculated by the TSW trend similarity method proposed in this paper, and (b) is the DLC matching ability calculated by the Euclidean distance.

[0239] As shown in (a), except for June, aggregator I ranked third, and its friendliness ranking in other months was 2. Except for August, when aggregator II ranked the lowest, its ranking in other months was 1. This reflects that industrial aggregators contribute differently to grid demand in different months. Each aggregator can adjust its own power load mode by introducing auxiliary devices such as energy storage and energy management management system according to the characteristics of grid demand to improve its contribution to grid demand. In Figure (b), except for August, aggregator II has a clear lead in the evaluation results, while the comprehensive evaluation results of aggregators I and III are relatively low. This is because the DLC matching ability characteristics based on the TSW method are smaller than those of the Euclidean distance method. The trend segmentation weighting method is used to optimize the importance of the DLC matching ability characteristics in the TOPSIS comprehensive evaluation index, making the comprehensive evaluation results more reasonable.

[0240] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0241] The power imbalance of supply and demand in a high-proportion renewable energy power system is greatly affected by the randomness of resources on both sides of the "source and load", which makes it difficult to accurately characterize the power balance regulation demand of the power grid, and it is difficult for users on the demand side to clearly understand the contribution of their own power consumption patterns to the power balance of the power grid. In order to reasonably evaluate the support level of demand-side power consumption behavior for system power balance, a study is conducted on the demand characteristics that are difficult to regulate power balance. According to the analysis of the distribution of power system demand characteristics within the monthly cycle, the complexity distribution of the monthly demand characteristics of the power system is characterized, the daily demand characteristics within the monthly cycle are extracted, and the demand characteristics are determined by the typical demand characteristic calculation method based on clustering weighting; according to the demand characteristics determined by the typical demand characteristic calculation method based on clustering weighting, the trend characteristics of the demand curve are calculated and analyzed, and the trend similarity matching calculation between the demand side load pattern and DLC during the demand period is completed; according to the trend similarity matching calculation between the demand side load pattern and DLC during the demand period, the load DCL matching capacity is obtained, the load contribution evaluation space is constructed, and the load contribution is evaluated by the TOPSIS model, and then the overall load contribution is obtained. This application determines the demand characteristics based on the typical demand characteristics calculation method of clustering weights, realizes the trend similarity matching calculation between the demand-side load pattern and the DLC during the demand period, and then constructs the load contribution evaluation space to obtain the overall load contribution. It can effectively evaluate the load contribution and realize the support level evaluation of the demand-side electricity consumption behavior for the system power balance, thereby effectively alleviating the pressure on the flexibility resource regulation of the power system.

[0242] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a typical feature mining and load evaluation method is implemented.

[0243] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0244] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0245] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0246] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0247] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A typical feature mining and load evaluation method, characterized in that: The typical feature mining and load evaluation method includes: Acquire several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend; Clustering each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category; Assigning weights to each of the cluster categories according to the number of category groups to obtain a weighted result; Constructing a weighted shortest distance optimization model according to the weighted results, and calculating the estimated typical value of each type of typical demand characteristics by using the weighted shortest distance optimization model; Calculating a demand-side load pattern during a demand period based on the estimated typical value; Performing a similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result; constructing a load contribution evaluation space according to the similarity matching calculation result; The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

2. The typical feature mining and load evaluation method according to claim 1 is characterized in that: The power demand period includes: the start time sequence of the valley and peak demand periods; the power demand trend includes: the maximum climbing trend during the valley period, the maximum descending trend during the valley period, the maximum climbing trend during the peak period, and the maximum descending trend during the peak period.

3. The typical feature mining and load evaluation method according to claim 2 is characterized in that: The obtaining of several types of typical demand characteristics of the power system within a cycle specifically includes: Obtain the differences in power system demand characteristics between months and the adaptability of demand-side users to typical grid demands within a monthly cycle to obtain typical data; According to the typical data, the power demand period is calculated based on the response capability of the demand-side user to provide 5% of the maximum load; According to the typical data, the power demand trend is determined by the distribution of trend characteristics within the power demand period of the monthly cycle.

4. The typical feature mining and load evaluation method according to claim 1, characterized in that: Each type of typical demand features is clustered to obtain clustering results, including: For a class of typical demand characteristics: Using the silhouette coefficient method to determine the number of clusters of the typical demand characteristics; According to the cluster number, the typical demand characteristics are clustered by using a K-means clustering algorithm to obtain a clustering result.

5. The typical feature mining and load evaluation method according to claim 1, characterized in that: Calculating a demand-side load pattern during a demand period based on the estimated typical value; Performing similarity matching calculation on the demand side load pattern and the demand load curve to obtain similarity matching calculation results, specifically including: Determine the valley demand period and the peak demand period based on the estimated typical value and taking the maximum load regulation capacity provided by the demand-side resources as a target; Calculate the trend feature weight according to the valley demand period and the peak demand period; The matching capability between the demand load curve and the load curve to be evaluated is calculated according to the trend characteristic weight, and a similarity matching calculation result is obtained.

6. The typical feature mining and load evaluation method according to claim 1, characterized in that: The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution, specifically including: Calculate the contribution of the stable component of the load to the mean net load during the demand period, which is used to measure the peak-valley support capacity of the demand-side load for the net load; The relationship between the ratio of the average load during the demand period to the average load during the entire period and the ratio of the average net load during the demand period to the average net load during the entire period is calculated to measure the load's ability to reverse peak and valley conditions.

7. A typical feature mining and load evaluation system, characterized in that: The typical feature mining and load evaluation system includes: Demand feature calculation module, used to: Acquire several types of typical demand characteristics of the power system within a cycle; the typical demand characteristics include: power demand period and power demand trend; Clustering each type of typical demand features to obtain clustering results; the clustering results include: clustering categories and the number of category groups in each clustering category; Assigning weights to each of the cluster categories according to the number of category groups to obtain a weighted result; Constructing a weighted shortest distance optimization model according to the weighted results, and calculating the estimated typical value of each type of typical demand characteristics by using the weighted shortest distance optimization model; DLC matching module for: Calculating a demand-side load pattern during a demand period based on the estimated typical value; Performing a similarity matching calculation on the demand side load pattern and the demand load curve to obtain a similarity matching calculation result; Load contribution evaluation module, used for: constructing a load contribution evaluation space according to the similarity matching calculation result; The TOPSIS model is used to evaluate the load contribution according to the load contribution evaluation space to obtain the overall load contribution.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the typical feature mining and load evaluation method described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the typical feature mining and load evaluation method described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the typical feature mining and load evaluation method described in any one of claims 1 to 6 is implemented.

Images