Typical adjustable resource optimization method and system for power grid adjustment scene

Abstract

The invention discloses a typical adjustable resource optimization method and system for a power grid adjustment scene, and the method comprises the steps: carrying out the step extraction of operation data for the historical operation data of different adjustable resources; setting typical scene response characteristics by considering response characteristic differences of the adjustable resources in different power grid adjustment scenes, constructing an adjustable resource response potential evaluation index system oriented to the power grid adjustment scenes, and calculating various evaluation indexes; the matching degree difference between the multi-type adjustable resource response characteristics and the power grid adjustment demand is considered, an adjustable resource comprehensive evaluation model is constructed, and adjustable resource response potential comprehensive evaluation in a specific scene is achieved; in combination with an adjustable resource response feature combination and a comprehensive evaluation model, rapid grouping and optimization are carried out on adjustable resources by adopting a combination evaluation and affinity propagation algorithm, and classification and optimization of adjustable resource groups with different response potentials are realized; and in combination with classification and optimization results, analyzing user demand response potential, and optimizing energy management and demand response strategies.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system resource regulation, and more specifically, relates to a typical adjustable resource optimization method and system for power grid regulation scenarios. Background Art

[0002] Under the goals of carbon peak and carbon neutrality, the proportion of clean energy has gradually increased, and thermal power units have gradually transformed into regulating units. The strong uncertainty of clean energy and the gradual increase in the load of the entire network have brought huge challenges to the regulation of the power grid. It is urgent to explore and screen adjustable resources with different regulation preferences and regulation performance, formulate reasonable resource combinations and scheduling plans, and gradually guide them to participate in different power grid regulation scenarios, so as to effectively deal with the regulation gaps such as peak regulation, frequency regulation, voltage regulation and standby in the power system, which is of great significance to the construction and development of new power systems.

[0003] In order to solve the problem of insufficient peak load regulation and consumption flexibility in power grids with a high proportion of renewable energy, on the one hand, we should further tap the potential of existing resource regulation and give full play to its regulation capacity through flexible optimization of conventional power sources; on the other hand, we should adopt new regulation means such as energy storage to enrich and expand grid regulation measures to meet the needs of grid flexibility. In terms of grid flexibility, traditional grid flexibility evaluation methods are mainly divided into two types: indicator calculation evaluation methods based on static parameters and evaluation methods based on dynamic simulation data. Among them, the static flexibility evaluation method refers to the study of the contribution of static parameters such as unit static parameters, unit adjustment range or line transmission capacity to system flexibility from the perspective of a certain variable. The dynamic flexibility evaluation method is based on dynamic simulation of the set application scenario, and from the perspective of flexibility supply and demand balance, the system flexibility is judged according to whether there is a system load loss or insufficient flexibility.

[0004] Static flexibility assessment methods only determine indicators and perform calculations and analyses based on the static parameter data of the system. The assessment perspective is single and conservative, and the assessment conclusions are not dynamic and real-time. At the same time, dynamic flexibility assessment methods mostly start from the perspective of flexibility resources and demands of traditional power systems, and rarely consider the output characteristics of renewable energy such as photovoltaics and their impact on grid flexibility. As a result, existing methods make it difficult to achieve quantitative flexibility assessment in power grids with a high proportion of renewable energy.

[0005] Solving the regulation problem of power systems with a high proportion of new energy requires fully tapping the regulation capabilities of multiple links of source, grid, load and storage. The types of adjustable resources currently studied are relatively single, and the regulation potential of adjustable resources in multiple links of power system source, grid, load and storage has not been fully tapped. The research on the regulation preferences of different types of adjustable resources and the adaptability of regulation scenarios is relatively vague. With the promotion of energy Internet technology and related policies, various types of adjustable and controllable resources have grown rapidly. By establishing a reasonable and effective selection method, various types of adjustable resources can be promoted to participate in various types of grid regulation scenarios such as frequency regulation and voltage regulation in an orderly manner. It can guide grid operators to formulate resource scheduling plans for various grid regulation scenarios, which will help to fully tap the various grid regulation capabilities of adjustable resources and promote the safe and stable operation of distribution networks and the absorption of new energy. Summary of the invention

[0006] In order to achieve reasonable optimization of adjustable resources in various power grid regulation scenarios, the present invention provides a typical adjustable resource optimization method for power grid regulation scenarios, including: first, taking into account the differences in response characteristics of adjustable resources participating in different power grid regulation scenarios, setting four typical scenario response characteristics of response cost, response speed, response capacity, and response time, and then constructing an adjustable resource response potential evaluation index system for power grid regulation scenarios; second, for different adjustable resource historical operation data, quantitative calculation of response potential evaluation indicators is achieved by step extraction of operation data; then, taking into account the differences in matching degree between multi-type adjustable resource response characteristics and power grid regulation requirements, a comprehensive evaluation model for adjustable resources is constructed to achieve comprehensive evaluation of the response potential of adjustable resources in specific scenarios; finally, combining the adjustable resource response feature combination and the comprehensive evaluation model, the subjective and objective combined evaluation and the nearest neighbor propagation algorithm are used to quickly group and optimize the adjustable resources, so as to achieve classification and optimization of adjustable resource groups with different response potentials.

[0007] The present invention adopts the following technical solution. A first aspect of the present invention provides a typical adjustable resource optimization method for a power grid regulation scenario, comprising the following steps:

[0008] Perform step extraction on historical operation data of different adjustable resources to obtain step information;

[0009] Based on the operation data and the extracted step information, taking into account the differences in the response characteristics of adjustable resources in different grid regulation scenarios, typical scenario response characteristics are set, and then an adjustable resource response potential evaluation index system for grid regulation scenarios is constructed, and various adjustable resource response potential evaluation indicators are calculated based on the extracted step information;

[0010] Based on the evaluation indicators, taking into account the differences in matching between the response characteristics of multiple types of adjustable resources and the grid regulation requirements, a comprehensive evaluation model for adjustable resources is constructed through the response sequence corresponding to the response characteristics, and a comprehensive evaluation of the adjustable resource response potential in specific scenarios is conducted;

[0011] Combining the combination of adjustable resource response characteristics and the comprehensive evaluation model, the combined evaluation and neighbor propagation algorithms are used to quickly group and optimize the adjustable resources, and to classify and optimize the adjustable resource groups with different response potentials;

[0012] Based on the classification and optimization results, analyze the user demand response potential and optimize energy management and demand response strategies.

[0013] Obtaining step information includes:

[0014] The step is set as the local average value of the user's change rate η that is lower than a certain threshold for more than 1 hour;

[0015]

[0016] In the formula,

[0017] Δp is the first-order difference value of the running data sequence in a certain time period selected,

[0018] Δp max , Δp mean and Δp std are the maximum, mean and variance of the running data difference series respectively.

[0019] Obtaining step information includes:

[0020] For the daily operation curve, the initial 4 data points of 1h time scale are selected as candidate step arrays, and the load step judgment conditions are defined;

[0021] Z=P[1,2,3,4] (2)

[0022] According to the step judgment condition, start traversing from the running data point corresponding to the next time point and update the candidate step array;

[0023]

[0024] In the formula, t represents the latest moment of traversal starting from the next moment point;

[0025] If the data point P(t) at the next moment belongs to the previous step, it is stored in the previous step array Z, and then the traversal continues; if P(t) does not belong to the previous step, t=t+3;

[0026] Calculate all step information W;

[0027]

[0028] Where W i is the information matrix of the i-th step divided, including the step height H i,ave , start time T i,start and end time T i,end , n represents the total number of steps divided.

[0029] Among them, include:

[0030] According to the differentiated power equipment composition, electricity consumption habits and willingness to respond to demand of different types of adjustable resources, 15 specific response potential indicators are designed from four aspects: response cost, response speed, response capacity and response time to describe the response potential of adjustable resources in different response characteristic dimensions. Based on the step information extracted from the operation data, an adjustable resource response potential evaluation index system is constructed. The specific indicators include:

[0031] Response characteristic 1: Response cost

[0032] Indicator 1.1: User peak-to-average electricity price ratio γ1

[0033]

[0034] In the formula, c i,peak is the peak electricity price for user i;

[0035] c i,flat is the average electricity price for user i;

[0036] Indicator 1.2: User valley-to-average electricity price ratioγ2

[0037]

[0038] In the formula, c i,valley is the off-peak electricity price for user i;

[0039] Response characteristic 2: Response speed

[0040] Indicator 2.1: Maximum level-up speed γ3

[0041]

[0042] In the formula, H valley,valley-1 and H valley,valley+1 is the height difference between the smallest step and the two adjacent steps;

[0043] t valley,valley-1 and t valley,valley+1 is the transition time between the minimum step and the adjacent step;

[0044] Indicator 2.2: Maximum reduction speed γ4

[0045]

[0046] In the formula, H peak,peak-1 and H peak,peak+1 is the height difference between the largest step and the two adjacent steps;

[0047] t peak,peak-1 and t peak,peak+1 is the transition time between the maximum step and the adjacent step;

[0048] Response characteristic 3: Response capacity

[0049] Indicator 3.1: The peak-to-valley difference of the average load step during the day γ5

[0050] γ5=max(μ i )-min(μ i ) (9)

[0051] In the formula, μ i is the mean value of the steps on the i-th day of history;

[0052] Indicator 3.2: Day-ahead load factor γ6

[0053]

[0054] Indicator 3.3: Day-ahead step divergence γ7

[0055]

[0056] In the formula, Set the average of the number of days steps for history;

[0057] Indicator 3.4: Intraday transferable load γ8

[0058]

[0059] In the formula, p peak (i) and p valley (i) is the maximum and minimum value of the step on the i-th day;

[0060] Indicator 3.5: Intraday peak-to-valley difference rate γ9

[0061]

[0062] Indicator 3.6: Intraday step volatility γ 10

[0063]

[0064] In the formula, σ i , μ iare the variance and mean of the weighted step height on the i-th day respectively;

[0065] Indicator 3.7: Maximum real-time curtailable capacityγ 11

[0066]

[0067] In the formula, p peak-1 (i) and p peak+1 (i) is the height of the two steps to the left and right of the highest step;

[0068] Indicator 3.8: Real-time reduction flexibility γ 12

[0069]

[0070] In the formula, Is the historical setting number of days p peak (i) the average value;

[0071] Metric 3.9: Real-time step transition time γ 13

[0072]

[0073] Where, t peak,start (i) t peak,end (i) are the start and end time of the maximum step on day i, respectively;

[0074] t peak-1,end (i) t peak+1,start (i) are the end time of the left side of the maximum step and the start time of the right side of the maximum step, respectively;

[0075] Response feature 4: Response time

[0076] Indicator 4.1: Average minimum durationγ 14

[0077]

[0078] Where, t end (i) and t start (i) are the end time series and the start time series of the step on the i-th day;

[0079] Indicator 4.2: Average maximum durationγ 15

[0080]

[0081] Where, t end (i) and t start(i) are the end time series and start time series of the step on the i-th day.

[0082] Among them, include:

[0083] According to the established adjustable resource potential index, there are four types of response characteristics, which are respectively recorded as response speed V x , Response cost C x , response capacity P x and response time T x Each response feature has its corresponding response sequence, namely C x =[γ x,1 ,γ x,2 ]、V x =[γ x,3 ,γ x,4 ]、P x =[γ x,5 ,γ x,6 ,···,γ x,13 ]、T x =[γ x,14 ,γ x,15 ], the comprehensive evaluation model of adjustable resources can be expressed as follows:

[0084]

[0085] Where: Val x is the comprehensive index value of the response characteristics of the adjustable resource x;

[0086] C max , C min It represents the maximum and minimum values ​​of the response cost sequence in the adjustable resource group;

[0087] V max 、V min Indicates the maximum and minimum values ​​of the response speed sequence in the adjustable resource group;

[0088] P max , P min Indicates the maximum and minimum values ​​of the response capacity sequence in the adjustable resource group;

[0089] T max , T min Indicates the maximum and minimum values ​​of the response time sequence in the adjustable resource group;

[0090] K is the weight of each response characteristic index, and the scoring is performed by the subjective and objective weighting method combining the entropy weight method and the hierarchical analysis method, where α and β are the characteristic weight combinations of the entropy weight method and the hierarchical analysis method, respectively, and the combination coefficients of the two methods are set to 0.5 respectively.

[0091] Among them, include:

[0092] The entropy weight method is used to perform the first evaluation weighting on the response feature combination of adjustable resources, and the weight distribution is determined by calculating the relationship between the entropy value and weight of each response feature indicator;

[0093] The hierarchical analysis method is used to carry out the second evaluation weighting of the response characteristic combination of adjustable resources. The weight of each factor is determined by constructing a judgment matrix and calculating the eigenvalue, and finally the relative weight of the decision-making scheme is obtained.

[0094] Based on the results of the combined evaluation and combined with the comprehensive evaluation model of adjustable resources, the comprehensive response evaluation value of the corresponding adjustable resources is calculated. By arranging in descending order, the adjustable resources with stronger comprehensive potential can be selected to obtain the optimal selection of adjustable resources.

[0095] Based on the response feature combination sequence of various adjustable resources, the neighbor propagation algorithm is used to cluster different adjustable resource users, so as to identify the same type of adjustable resource users and distinguish adjustable resource user groups with different response characteristics.

[0096] Preprocess the raw data;

[0097] The initial potential indicator matrix X is constructed for 15 evaluation indicators and n evaluation users, which is expressed as follows:

[0098]

[0099] By performing Euclidean normalization, x i,j Normalized into;

[0100]

[0101] Where, i = 1, 2, ..., 15; j = 1, 2, ..., n;

[0102] Based on the normalized potential index, the entropy value is calculated;

[0103]

[0104] In the formula, h i is the entropy value of the i-th indicator;

[0105] Calculate the entropy weight based on the entropy value;

[0106]

[0107] In the formula, ω i is the entropy weight of the i-th indicator;

[0108] Based on the entropy weight, the feature vector weight is calculated;

[0109] α=[ω1,ω2,…,ω15 ] (25)

[0110] In the formula, α is the objectively weighted feature weight.

[0111] Among them, include:

[0112] Establish a hierarchical structure, clarify decision-making goals and split the levels, including the goal level, criterion level and plan level;

[0113] Construct a judgment matrix. For each level, establish a judgment matrix A;

[0114]

[0115] In the formula, a i,j Factor a i with a j and meet the following conditions:

[0116]

[0117] Where, i, j = 1, 2, ... n;

[0118] Based on the obtained judgment matrix, the importance is sorted and the maximum characteristic root λ is calculated according to A. max The corresponding eigenvectors are used to obtain the comprehensive calculation results, which are expressed as follows:

[0119]

[0120] Perform consistency check by checking the consistency ratio CR. The consistency ratio calculation formula is as follows:

[0121]

[0122] In the formula, CI is the consistency index,

[0123] RI is the average consistency index,

[0124] If CR is less than 0.1, the judgment result is reliable.

[0125] Among them, include:

[0126] Calculate the Euclidean distance between samples,

[0127]

[0128] In the formula, d(x i ,x j ) is the Euclidean distance between samples;

[0129] i=1,2,…,n; j=1,2,…,n;

[0130] l is the sample feature dimension;

[0131]

[0132]

[0133] x i , x j Represents a user sample data object containing l-dimensional attributes;

[0134] Calculate the similarity matrix,

[0135]

[0136] Where p(k) is the diagonal value of the similarity matrix;

[0137] Initialize the attraction matrix and the belonging matrix to zero matrices with the same structure as the similarity matrix;

[0138] Update the attraction matrix r(i,k),

[0139] r(i,k)=s(i,k)-max k′≠k {a(i,k′)+s(i,k′)} (32)

[0140] Update the attribution matrix a(i,k),

[0141]

[0142] The above two matrices are updated cyclically until the algorithm converges or reaches the maximum number of iterations, and the corresponding cluster centers are output;

[0143] According to the cluster center and the attraction matrix between samples, each sample is assigned to the corresponding cluster, and finally the clustering result is output.

[0144] 13. The second aspect of the present invention provides a typical adjustable resource optimization system for power grid regulation scenarios, and runs a typical adjustable resource optimization method for power grid regulation scenarios, including: a data collection and processing module, an evaluation index module, a comprehensive evaluation module, and a classification optimization module;

[0145] The data collection and processing module extracts the historical operation data of adjustable resources in steps according to the historical operation data of different adjustable resources;

[0146] The evaluation index module is connected to the data collection and processing module. Based on the historical operation data obtained by the data collection and processing module and the step information extracted, it is used to take into account the differences in response characteristics of adjustable resources in different power grid regulation scenarios, set four typical scenario response characteristics of response cost, response speed, response capacity, and response time, and then construct an adjustable resource response potential evaluation index system for power grid regulation scenarios and calculate various adjustable resource response potential evaluation indicators;

[0147] The comprehensive evaluation module is connected with the evaluation index calculation module. Based on the evaluation index calculated by the evaluation index calculation module, the matching degree difference between the response characteristics of multiple types of adjustable resources and the grid regulation demand is taken into account. Through the response sequence corresponding to the response characteristics, an adjustable resource comprehensive evaluation model is constructed to comprehensively evaluate the adjustable resource response potential under specific scenarios.

[0148] The classification and optimization module is connected with the evaluation index module and the comprehensive evaluation module. It combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module, and uses the combined evaluation and neighbor propagation algorithm to quickly group and optimize the adjustable resources, and classify and optimize the adjustable resource groups with different response potentials.

[0149] The strategy optimization module is connected with the classification and optimization module. Based on the classification and optimization results obtained by the classification and optimization module, it analyzes the user demand response potential and optimizes energy management and demand response strategies.

[0150] The classification optimization module includes: optimization unit and classification unit;

[0151] The optimization unit combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module to select the adjustable resource groups with different response potentials through subjective and objective combined evaluation;

[0152] The classification unit is connected to the comprehensive evaluation module. Based on the response feature combination sequence of various adjustable resources obtained by the comprehensive evaluation module, the neighbor propagation algorithm is used to cluster different adjustable resource users, thereby effectively identifying the same type of adjustable resource users and effectively distinguishing adjustable resource user groups with different response characteristics.

[0153] The optimization unit includes: the first evaluation weighting unit, the second evaluation weighting unit, and the combined optimization unit;

[0154] The first evaluation weighting subunit is connected to the comprehensive evaluation module and is used to objectively weight the response feature combination of the adjustable resources obtained by the comprehensive evaluation module using the entropy weight method, and determine the weight distribution by calculating the relationship between the entropy value and weight of each response feature indicator;

[0155] The second evaluation weighting unit is connected to the comprehensive evaluation module. It uses the hierarchical analysis method to subjectively weight the response characteristic combination of adjustable resources obtained by the comprehensive evaluation module. The weight of each factor is determined by constructing a judgment matrix and calculating the eigenvalue. Finally, the relative weight of the decision-making plan is obtained for comprehensive evaluation.

[0156] Combined with the optimization small unit, it is connected with the first evaluation weighting small unit, the second evaluation weighting small unit and the comprehensive evaluation module. Based on the results of subjective and objective evaluation and combined with the comprehensive evaluation model of adjustable resources, the comprehensive response evaluation value of the corresponding adjustable resources is calculated. By arranging in descending order, the adjustable resources with stronger comprehensive potential can be selected, thereby realizing the optimization of adjustable resources.

[0157] The beneficial effect of the present invention is that, compared with the prior art, the present invention designs a typical resource optimization method for power grid regulation scenarios; taking into account the differences in response characteristics of adjustable resources participating in different power grid regulation scenarios, four typical scenario response characteristics of response cost, response speed, response capacity, and response time are set, and then an adjustable resource response potential evaluation index system for power grid regulation scenarios is constructed; for different adjustable resource historical operation data, the quantitative calculation of the response potential evaluation index is realized by step extraction of the operation data; taking into account the difference in matching degree between the response characteristics of multiple types of adjustable resources and the power grid regulation requirements, an adjustable resource comprehensive evaluation model is constructed to realize the comprehensive evaluation of the adjustable resource response potential in a specific scenario; combined with the adjustable resource response feature combination and the comprehensive evaluation model, the subjective and objective combined evaluation and the nearest neighbor propagation algorithm are used to quickly group and optimize the adjustable resources, so as to realize the classification and optimization of adjustable resource groups with different response potentials. The method is conducive to realizing the reasonable classification of adjustable resources with different regulation preferences and regulation performance, improving the flexibility of each link of the system, and promoting the construction and development of new power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0158] Figure 1 A flowchart of the steps of a typical adjustable resource optimization method for power grid regulation scenarios;

[0159] Figure 2 Step information calculation flow chart;

[0160] Figure 3 Adjustable resource potential comprehensive assessment indicator map;

[0161] Figure 4 Flow chart of quantitative ranking of potential indicators;

[0162] Figure 5 Ranking results of different response characteristics potential of various industries

[0163] Figure 6 User clustering result diagram

[0164] Figure 7 User distribution histogram

[0165] Figure 8 The top 20 user potential ranking results chart

[0166] Fig. 9 Historical 15-day load curve of high-potential typical users DETAILED DESCRIPTION

[0167] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0168] like Figure 1 As shown, in order to achieve reasonable optimization of adjustable resources in various power grid regulation scenarios, embodiment 1 of the present invention provides a typical adjustable resource optimization method for power grid regulation scenarios, including: first, taking into account the differences in response characteristics of adjustable resources participating in different power grid regulation scenarios, setting four typical scenario response characteristics of response cost, response speed, response capacity, and response time, and then constructing an adjustable resource response potential evaluation index system for power grid regulation scenarios; secondly, for different adjustable resource historical operation data, quantitative calculation of response potential evaluation indicators is achieved by step extraction of operation data; then, taking into account the differences in matching degree between multi-type adjustable resource response characteristics and power grid regulation requirements, a comprehensive evaluation model for adjustable resources is constructed to achieve comprehensive evaluation of the response potential of adjustable resources in specific scenarios; finally, combining the adjustable resource response feature combination and the comprehensive evaluation model, the subjective and objective combined evaluation and the nearest neighbor propagation algorithm are used to quickly group and optimize the adjustable resources, so as to achieve classification and optimization of adjustable resource groups with different response potentials.

[0169] Embodiment 1 specifically comprises the following steps:

[0170] Step 1: for different adjustable resource historical operation data, perform step extraction on the adjustable resource historical operation data;

[0171] In a further preferred but non-limiting embodiment of the present invention, step 1 specifically comprises:

[0172] The operating curve of the adjustable resource is composed of the load of its own power equipment. Based on the premise that a single device has a limited load state, it can be observed that the operating curve has an approximate "step" phenomenon. Then, the step extraction of the adjustable resource operating curve can be performed, and the above-mentioned potential evaluation indicators can be calculated based on the extracted step information.

[0173] The step is the steady-state information of the operation curve under various operation modes, so it can more accurately describe the differentiated energy usage habits and response potential of different users. Based on the three-fold variance principle, the step is defined as the local average value of the user's change rate η for more than one hour below a certain threshold.

[0174]

[0175] In the formula, Δp is the first-order difference value of the running data sequence in a certain time period selected, Δp max , Δp mean and Δp std are the maximum, mean and variance of the running data difference series respectively.

[0176] Based on the daily operation data of the adjustable resource for the past 15 days, the step information is extracted. Figure 2 , the specific steps are as follows:

[0177] Step S1: For the daily operation curve, select the initial 4 data points of the 1h time scale as the candidate step array, and define the load step judgment conditions.

[0178] Z=P[1,2,3,4] (2)

[0179] Step S2: Based on the step judgment condition obtained in step S1, traverse from the operation data point corresponding to the next time point and update the candidate step array.

[0180]

[0181] In the formula, t represents the latest time of the traversal starting from the next time point. If the data point P(t) at the next time point belongs to the previous step, it is stored in the previous step array Z, and then the traversal continues; if P(t) does not belong to the previous step, t=t+3.

[0182] Step S3: Calculate all step information W;

[0183]

[0184] Where W i is the information matrix of the i-th step divided, including the step height H i,ave , start time T i,start and end time T i,end , n represents the total number of steps divided.

[0185] This Example 1 adopts a typical adjustable resource optimization method for power grid regulation scenarios proposed by the present invention. Taking the power grid voltage regulation scenario as an example, considering that most of the users participating in demand response are adjustable load users, the actual load data of users in 10 industries including mineral resource processing industry, steel manufacturing industry, building materials manufacturing industry, information technology service industry, biomedicine industry, science and technology park, construction development industry, science and technology service industry, banking industry and insurance industry in a certain area of ​​Nanjing are selected for user screening, as shown in Table 1 below. In this embodiment, 105 daily load data of users for 15 consecutive days are extracted as sample data, and the data sampling is set to 15min / point, that is, the selected user load contains 1440 data points.

[0186] Table 2 User selection composition table

[0187]

[0188]

[0189] Step 2: Based on the step information extracted from the operating data in step 1, taking into account the differences in response characteristics of adjustable resources in different grid regulation scenarios, four typical scenario response characteristics of response cost, response speed, response capacity, and response time are set to form an adjustable resource response characteristic combination, and then an adjustable resource response potential evaluation index system for grid regulation scenarios is constructed, and various adjustable resource response potential evaluation indicators are calculated based on the extracted step information;

[0190] In a further preferred but non-limiting embodiment of the present invention, the construction of the evaluation index system in step 2 specifically includes:

[0191] Differentiate the composition of power equipment, electricity usage habits and willingness to respond to demand based on different types of adjustable resources, such as Figure 3 , design 15 specific response potential indicators from four aspects: response cost, response speed, response capacity and response time to describe the response potential of adjustable resources in different response characteristic dimensions, and construct an adjustable resource response potential evaluation indicator system based on the step information extracted from the operation data;

[0192] Specific indicators include:

[0193] Response characteristic 1: Response cost

[0194] The response cost of adjustable resources reflects, to a certain extent, the "willingness" of adjustable resources to participate in demand response. Lower response costs will attract more users. Due to the difficulty in obtaining response cost data, the response cost is simplified to represent the peak-to-valley price ratio of the response electricity price of different types of adjustable resource users, which can be expressed as follows:

[0195] Indicator 1.1: User peak-to-average electricity price ratio γ1

[0196]

[0197] In the formula, c i,peak is the peak electricity price for user i, c i,flat is the average electricity price of user i.

[0198] Indicator 1.2: User valley-to-average electricity price ratioγ2

[0199]

[0200] In the formula, c i,valley is the off-peak electricity price for user i.

[0201] Response characteristic 2: Response speed

[0202] The response speed of adjustable resources reflects how quickly resources adjust their energy consumption patterns and embodies the flexibility of resources in participating in regulation. A faster response speed means that resources can increase or decrease loads more quickly to meet the real-time regulation needs of the power grid. Taking into account the difficulty of extracting discrete indicators from the collected continuous load data, this patent defines the maximum step-up speed and maximum step-down speed indicators based on the load step idea to evaluate the response speed of resources, which can be specifically expressed as follows:

[0203] Indicator 2.1: Maximum level-up speed γ3

[0204]

[0205] In the formula,

[0206] H valley,valley-1 and H valley,valley+1 is the height difference between the smallest step and the two adjacent steps,

[0207] t valley,valley-1 and t valley,valley+1 is the transition time between the minimum step and the adjacent step.

[0208] Indicator 2.2: Maximum reduction speed γ4

[0209]

[0210] In the formula,

[0211] H peak,peak-1 and H peak,peak+1 is the height difference between the largest step and the two adjacent steps,

[0212] t peak,peak-1 and t peak,peak+1 is the transition time between the maximum step and the adjacent step.

[0213] Response characteristic 3: Response capacity

[0214] The response capacity of adjustable resources refers to the degree of change that resources can adjust within a certain period of time, reflecting the degree of flexibility they can provide when participating in grid regulation, also known as the dispatchability of resources. The size of the response capacity directly affects the response potential of users participating in grid regulation. Taking into account the differences in the response capacity of resources at different time scales, the response capacity was evaluated from three time scales: day-ahead, intraday, and real-time.

[0215] Indicator 3.1: The peak-to-valley difference of the average load step during the day γ5

[0216] γ5=max(μ i )-min(μ i ) (9)

[0217] In the formula, μ i is the mean of the steps on the i-th day in history.

[0218] Indicator 3.2: Day-ahead load factor γ6

[0219]

[0220] Among them, the greater the day-ahead load factor, the greater the day-ahead response potential of the adjustable resources.

[0221] Indicator 3.3: Day-ahead step divergence γ7

[0222]

[0223] In the formula, It is the average value of the historical 15-day steps. This value reflects the dispersion of the step distribution. The greater the change, the greater the response potential.

[0224] Indicator 3.4: Intraday transferable load γ8

[0225]

[0226] In the formula, p peak (i) and p valley (i) is the maximum and minimum value of the step on the i-th day.

[0227] Indicator 3.5: Intraday peak-to-valley difference rate γ9

[0228]

[0229] In the formula, For p valley (i) The historical 15-day average.

[0230] Indicator 3.6: Intraday step volatility γ 10

[0231]

[0232] In the formula, σ i , μ i are the variance and mean of the weighted step height on the i-th day. 10 It refers to the variance of the step sequence of resources within the historical 15 days divided by the mean. The greater the fluctuation, the greater the intraday response potential.

[0233] Indicator 3.7: Maximum real-time curtailable capacityγ 11

[0234]

[0235] In the formula, p peak-1 (i) and p peak+1 (i) is the height of the two steps to the left and right of the highest step.

[0236] Indicator 3.8: Real-time reduction flexibility γ 12

[0237]

[0238] In the formula, It is the first time in 15 days in history peak The average value of (i).

[0239] Metric 3.9: Real-time step transition time γ 13

[0240]

[0241] In the formula,

[0242] t peak,start (i) t peak,end (i) are the start time and end time of the maximum step on the i-th day, respectively.

[0243] t peak-1,end (i) t peak+1,start (i) are the end time of the left side of the maximum step and the start time of the right side of the maximum step, respectively.

[0244] Response feature 4: Response time

[0245] The response time of adjustable resources refers to the length of time that the resources can maintain a new operating state after adjusting the energy consumption mode, reflecting the stability of the resources in responding to the grid regulation needs.

[0246] Indicator 4.1: Average minimum durationγ 14

[0247]

[0248] Where, t end (i) and t start (i) are the end time series and start time series of the step on the i-th day.

[0249] Indicator 4.2: Average maximum durationγ 15

[0250]

[0251] Where, t end (i) and t start (i) are the end time series and start time series of the step on the i-th day.

[0252] Step 3: Based on the evaluation index of adjustable resource response potential obtained in step 2, taking into account the difference in matching degree between the response characteristics of multiple types of adjustable resources and the grid regulation demand, a comprehensive evaluation model of adjustable resources is constructed through the response sequence corresponding to the response characteristics to comprehensively evaluate the adjustable resource response potential in specific scenarios;

[0253] In a further preferred but non-limiting embodiment of the present invention, step 3 specifically comprises:

[0254] According to the established adjustable resource potential index, there are four types of response characteristics, which are respectively recorded as response speed V x , Response cost C x , response capacity P x and response time T x Each response feature has its corresponding response sequence, namely C x =[γ x,1 ,γ x,2 ]、V x =[γ x,3 ,γ x,4 ]、P x =[γ x,5 ,γ x,6 ,···,γ x,13 ]、T x =[γ x,14 ,γ x,15 ], therefore, the comprehensive evaluation model of adjustable resources can be expressed as follows:

[0255]

[0256] Where: Val x is the comprehensive index value of the response characteristics of the adjustable resource x, C max , C min represents the maximum and minimum values ​​of the response cost sequence in the adjustable resource group; V max 、Vmin represents the maximum and minimum values ​​of the response speed sequence in the adjustable resource group; P max , P min represents the maximum and minimum values ​​of the response capacity sequence in the adjustable resource group; T max , T min It represents the maximum and minimum values ​​of the response time sequence in the adjustable resource group; K is the weight of each response characteristic index, which is scored by the subjective and objective weighting method combining the entropy weight method and the hierarchical analysis method, where α and β are the characteristic weight combinations of the entropy weight method and the hierarchical analysis method, respectively, and the combination coefficients of the two methods are set to 0.5 respectively.

[0257] Step 4: Combine the adjustable resource response feature combination and comprehensive evaluation model obtained in step 3, use the combined evaluation and neighbor propagation algorithm to quickly group and optimize the adjustable resources, and classify and optimize the adjustable resource groups with different response potentials;

[0258] In a further preferred but non-limiting embodiment of the present invention, Figure 4 , step 4 specifically includes:

[0259] Step 4.1, using the entropy weight method to perform the first evaluation weighting on the response feature combination of adjustable resources, and determine the weight distribution by calculating the relationship between the entropy value and weight of each response feature indicator, specifically including:

[0260] Step A1: preprocessing the original data;

[0261] The initial potential indicator matrix X is constructed for 15 evaluation indicators and n evaluation users, which can be expressed as follows:

[0262]

[0263] In the formula, x i,j is the i-th evaluation index of the j-th evaluation user,

[0264] i=1,2,…,15; j=1,2,…,n.

[0265] For the problem of inconsistent dimensions of different evaluation indicators, this section needs to perform Euclidean normalization. i,j Normalized to y i,j , thereby eliminating the error caused by dimensional differences;

[0266]

[0267] Where, i = 1, 2, ..., 15; j = 1, 2, ..., n; y i,j is the i-th evaluation index of the j-th evaluation user after Euclidean normalization;

[0268] The purpose of dimensional unification is achieved by scaling each indicator to unit length;

[0269] Step A2: Calculating the entropy value based on the normalized potential index obtained in step A1;

[0270]

[0271] In the formula, h i is the entropy value of the i-th indicator;

[0272] Step A3: Calculate the entropy weight based on the entropy value obtained in step A2;

[0273]

[0274] In the formula, ω i is the entropy weight of the i-th indicator;

[0275] Step A4: Calculate the feature vector weight based on the entropy weight obtained in step A3

[0276] α=[ω1,ω2,…,ω 15 ] (25)

[0277] In the formula, α is the objectively weighted feature weight.

[0278] Step 4.2, use the hierarchical analysis method to perform the second evaluation weighting on the response feature combination of adjustable resources, determine the weight of each factor by constructing a judgment matrix and calculating the eigenvalue, and finally obtain the relative weight of the decision-making plan for comprehensive evaluation, which includes:

[0279] Step B1: Establish a hierarchical structure, clarify the decision-making goals and divide them into several levels, including the goal level, the criterion level and the plan level. The goal is the overall goal of the decision, the criterion is the factors that need to be considered to achieve the goal, and the plan is the actual decision-making plan that can be selected;

[0280] Step B2: Construct a judgment matrix. For each level established in step B1, construct a judgment matrix A to represent the relative importance of different factors. This is obtained through expert judgment or empirical investigation;

[0281]

[0282] In the formula, a i,j Factor a i with a j and meet the following conditions:

[0283]

[0284] Where, i, j = 1, 2, ... n;

[0285] Step B3: sorting the importance based on the judgment matrix obtained in step B2;

[0286] Calculate the maximum characteristic root λ based on A max The corresponding eigenvector is then used to obtain the comprehensive calculation result. It can be expressed as follows:

[0287]

[0288] In the formula, β is the subjectively weighted feature weight;

[0289] Step B4: Perform consistency check;

[0290] The consistency check is performed by checking the consistency ratio CR, which can be specifically expressed as follows:

[0291]

[0292] In the formula, CI is the consistency index, RI is the average consistency index,

[0293] If CR is less than 0.1, the result is considered reliable. The RI table can be seen in the table below.

[0294] Table 1 Consistency Index Table

[0295]

[0296] Step 4.3, based on the subjective and objective evaluation results obtained in Step 4.1 and Step 4.2 and combined with the adjustable resource comprehensive evaluation model obtained in Step 3, the comprehensive response evaluation value of the corresponding adjustable resource is calculated, and the adjustable resource with stronger comprehensive potential can be selected by arranging in descending order, thereby realizing the optimization of the adjustable resource;

[0297] Step 4.4: Based on the response feature combination sequence of various adjustable resources obtained in step 3, the neighbor propagation algorithm is used to cluster different adjustable resource users, so as to effectively identify the same type of adjustable resource users and the adjustable resource user groups with different response features. The clustering algorithm specifically includes:

[0298] Step C1: Calculate the Euclidean distance between adjustable resource user samples;

[0299]

[0300] In the formula, d(x i ,x j ) is the Euclidean distance between samples;

[0301] i=1,2,…,n; j=1,2,…,n;

[0302] l is the sample feature dimension;

[0303]

[0304] x i , x j Represents a user sample data object containing l-dimensional attributes;

[0305] Euclidean distance is used to determine the distance of a data point from the cluster center so that the data point can be assigned to the nearest cluster.

[0306] Step C2: Calculate the similarity matrix;

[0307]

[0308] Where s(i,k) is the similarity matrix, which represents the similarity between node i and node k;

[0309] p(k) is the diagonal value of the similarity matrix.

[0310] Step C3: Initialize the attraction matrix and the attribution matrix to zero matrices with the same structure as the similarity matrix.

[0311] Step C4: Update the attraction matrix r(i,k),

[0312] r(i,k)=s(i,k)-max k′≠k {a(i,k′)+s(i,k′)} (32)

[0313] In the formula, r(i,k) is the attraction matrix, which represents the attraction of node k to node i;

[0314] Step C5: Update the attribution matrix a(i,k),

[0315]

[0316] In the formula, a(i,k) is the attribution matrix, which indicates the degree of recognition of node i to node k;

[0317] Step C6: cyclically update the two matrices obtained in step C4 and step C5 until the algorithm converges or reaches the maximum number of iterations, and output the corresponding cluster center.

[0318] Step C7 assigns each sample to a corresponding cluster according to the cluster center and the attraction matrix between samples, and finally outputs the clustering result.

[0319] Based on the classification and optimization results obtained in step 4, analyze the user demand response potential and optimize energy management and demand response strategies.

[0320] Figure 5This is the potential ranking result of different response characteristics of various industries in this embodiment 1. From the perspective of industry classification, the potential of industries with different response characteristics is ranked. It can be seen that in terms of response price, except for the mineral resource processing industry, the other 9 major industries all maintain a high potential level of about 0.1. The reason is that these industries usually have greater flexibility or are more sensitive to electricity prices; in terms of response capacity, the banking industry ranks first with a potential level of 0.11, and the potential of other industries is basically around 0.06. The reason is that the business model of the banking industry is relatively flexible and has a large adjustment space; in terms of response speed, the industry with the highest potential is the steel manufacturing industry, and the industry with the lowest potential is the building materials manufacturing industry. The reason is that the steel manufacturing industry usually has more complex and diversified production processes and equipment, which enables it to adjust the production rhythm more quickly to respond to changes in demand. The building materials manufacturing industry may be restricted by specific production processes and equipment, resulting in a slower adjustment speed; in terms of response time, the mineral resource processing industry and the steel manufacturing industry show a higher potential level. The reason is that the production process of these industries usually takes a long time to adjust. Therefore, once the adjustment is completed, they can maintain a relatively stable production state for a long time. The science and technology service industry may be affected by its more flexible business model, resulting in a relatively short response time and lower potential.

[0321] Figure 6 The user clustering results obtained in Example 1 are as follows based on different clustering category labels and different industry type labels: Figure 7 The user distribution bar chart shown in the figure shows that category 2 only has users in the biopharmaceutical industry, category 5 only has users in the banking industry, and the industry types of the other user groups are quite diverse. Figure 6 It can be seen that users of different categories show great differences in load characteristics, while the response potential characteristics of user groups within the same cluster are similar. This also illustrates the effectiveness of the user-side adjustable resource indicator system established in this paper in identifying user groups. For the demand for grid voltage regulation, accurate identification of user groups with different response characteristics and user optimization can provide a more economical and effective demand response solution for grid voltage regulation.

[0322] Figure 8The distribution results of the top 20 users in descending order are shown in the figure. It can be seen from the figure that the top 20 are composed of the banking industry, biomedicine industry, construction development industry and steel manufacturing industry, with the number of users being 10, 5, 3 and 2 respectively. This also shows that the banking industry is an industry with a comparative advantage in terms of comprehensive potential. According to the classification of general industrial and commercial users and large industrial users, it can be analyzed that general industrial and commercial users are high-potential adjustable resources on the user side, which is consistent with reality. At present, the number of general industrial and commercial users is increasing year by year and they are usually equipped with high-maturity energy management equipment and cost control strategies. They can not only provide feasible demand response solutions but also have a high willingness to respond to demand, and are the main force of adjustable resources on the user side. At the same time, the data centers and branches of the banking industry usually require a lot of electricity to support their complex computing and trading systems, so they have great potential in participating in demand response.

[0323] Take user No. 102 with high potential as an example. Fig. 9 The figure is the daily load curve of the user for the past 15 days. At the same time, the historical load curve group of the user is sorted according to the daily average load size. The black bold curve in the figure is the load curve with the largest daily average load of the user, and the red bold curve is the load curve with the smallest daily average load of the user. It can be seen from the historical load curve of the user that its load volatility is large. The load level between 09:00 and 17:00 will be higher than that in other time periods, and there is a large potential for intraday peak shift. The user belongs to the banking industry. During working hours on weekdays, he may experience large transaction activities, including customer deposits and withdrawals, loan approval, and transaction settlement. This involves significant differences in the usage patterns and load conditions of related equipment such as computers, ATMs, and air conditioners. This also makes the overall load level of such users with low average load days and high average load days have a large gap, and the intraday peak and valley differences are large and the volatility is obvious. In this case, the demand response potential of banking users is relatively large. Through reasonable energy management and demand response strategies, measures can be taken during peak load periods, such as energy storage, adjusting equipment operation modes, and formulating flexible work schedules, to smooth load peaks, reduce energy costs, and improve energy efficiency. At the same time, for days with low average load, measures can be taken to further reduce the base load to reduce the pressure on grid voltage regulation.

[0324] Embodiment 2 of the present invention provides a typical adjustable resource optimization system for power grid regulation scenarios, the system comprising a data collection and processing module, an evaluation index module, a comprehensive evaluation module, a classification optimization module, and a strategy optimization module;

[0325] The data collection and processing module extracts the historical operation data of adjustable resources in steps according to the historical operation data of different adjustable resources;

[0326] Data collection and processing module, specifically including:

[0327] For the daily operation curve, select the initial 4 data points of 1h time scale as the candidate step array, and define the load step judgment conditions; start traversing from the operation data point corresponding to the next time point, and update the candidate step array; calculate all step information W.

[0328] The evaluation index module is connected to the data collection and processing module. Based on the historical operation data obtained by the data collection and processing module and the step information extracted, it is used to take into account the differences in response characteristics of adjustable resources in different power grid regulation scenarios, set four typical scenario response characteristics of response cost, response speed, response capacity, and response time, and then construct an adjustable resource response potential evaluation index system for power grid regulation scenarios and calculate various adjustable resource response potential evaluation indicators;

[0329] The construction of the evaluation indicator system includes:

[0330] According to the differentiated power equipment composition, electricity consumption habits and willingness to respond to demand of different types of adjustable resources, 15 specific response potential indicators are designed from four aspects: response cost, response speed, response capacity and response time to describe the response potential of adjustable resources in different response characteristic dimensions. Based on the step information extracted from the operation data, an adjustable resource response potential evaluation index system is constructed. The specific indicators include:

[0331] The user peak-to-average electricity price ratio γ1 and the user valley-to-average electricity price ratio γ2 in the response cost;

[0332] The maximum step-up speed γ3 and the maximum step-down speed γ4 in the response speed;

[0333] The day-ahead average load step peak-to-valley difference γ5, day-ahead load rate γ6, day-ahead step divergence γ7, intraday transferable load γ8, intraday peak-to-valley difference rate γ9, intraday step volatility γ 10 , Real-time maximum reducible capacityγ 11 , real-time reduction flexibility γ 12 , real-time step conversion time γ 13 ;

[0334] The average shortest duration of the response time γ 14 , the average maximum duration γ 15 ;

[0335] A comprehensive evaluation module is connected to the evaluation index calculation module, based on the evaluation index calculated by the evaluation index calculation module, taking into account the difference in matching degree between the response characteristics of multiple types of adjustable resources and the grid regulation demand, and constructing an adjustable resource comprehensive evaluation model through the response sequence corresponding to the response characteristics, so as to comprehensively evaluate the adjustable resource response potential under specific scenarios;

[0336] The comprehensive assessment module specifically includes:

[0337] According to the established adjustable resource potential index, there are four types of response characteristics, which are respectively recorded as response speed V x , Response cost C x , response capacity P x and response time T x Each response feature has its corresponding response sequence, and a comprehensive evaluation model of adjustable resources is established based on the corresponding response sequence;

[0338] The classification and optimization module is connected with the evaluation index module and the comprehensive evaluation module. It combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module, and uses the combined evaluation and neighbor propagation algorithm to quickly group and optimize the adjustable resources, and classify and optimize the adjustable resource groups with different response potentials.

[0339] The classification optimization module specifically includes: optimization unit and classification unit;

[0340] The optimization unit combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module to select the adjustable resource groups with different response potentials through subjective and objective combined evaluation;

[0341] The optimization unit specifically includes: a first evaluation weighting subunit, a second evaluation weighting subunit, and a combined optimization subunit;

[0342] The first evaluation weighting subunit is connected to the comprehensive evaluation module and is used to objectively weight the response feature combination of the adjustable resources obtained by the comprehensive evaluation module using the entropy weight method, and determine the weight distribution by calculating the relationship between the entropy value and weight of each response feature indicator;

[0343] The second evaluation weighting unit is connected to the comprehensive evaluation module. It uses the hierarchical analysis method to subjectively weight the response characteristic combination of adjustable resources obtained by the comprehensive evaluation module. The weight of each factor is determined by constructing a judgment matrix and calculating the eigenvalue. Finally, the relative weight of the decision-making plan is obtained for comprehensive evaluation.

[0344] Combined with the optimization subunit, connected with the first evaluation weighting subunit, the second evaluation weighting subunit and the comprehensive evaluation module, the comprehensive response evaluation value of the corresponding adjustable resource is calculated based on the results of subjective and objective evaluation and combined with the adjustable resource comprehensive evaluation model. The adjustable resource with strong comprehensive potential can be selected by arranging in descending order, thereby realizing the optimization of the adjustable resource;

[0345] The classification unit is connected to the comprehensive evaluation module. Based on the response feature combination sequence of various adjustable resources obtained by the comprehensive evaluation module, the neighbor propagation algorithm is used to cluster different adjustable resource users, thereby effectively identifying the same type of adjustable resource users and the adjustable resource user groups with different response features.

[0346] The strategy optimization module is connected with the classification and optimization module. Based on the classification and optimization results obtained by the classification and optimization module, it analyzes the user demand response potential and optimizes energy management and demand response strategies.

[0347] The beneficial effect of the present invention is that, compared with the prior art, the types of adjustable resources currently studied are relatively single, and the regulation potential of adjustable resources under multiple links of power system source, grid, load and storage has not been fully explored. The research on the regulation preference and adaptability of different types of adjustable resources to regulation scenarios is relatively vague; therefore, the present invention designs a typical resource optimization method for power grid regulation scenarios; taking into account the differences in response characteristics of adjustable resources participating in different power grid regulation scenarios, four typical scenario response characteristics of response cost, response speed, response capacity and response time are set, and then an adjustable resource response potential evaluation index system for power grid regulation scenarios is constructed; for different historical operation data of adjustable resources, quantitative calculation of response potential evaluation indicators is realized by step extraction of operation data; taking into account the differences in matching degree between response characteristics of multiple types of adjustable resources and power grid regulation requirements, a comprehensive evaluation model for adjustable resources is constructed to realize comprehensive evaluation of the response potential of adjustable resources under specific scenarios; combining the combination of adjustable resource response characteristics and the comprehensive evaluation model, the combined evaluation and neighbor propagation algorithms are used to quickly group and optimize the adjustable resources, so as to realize the classification and optimization of adjustable resource groups with different response potentials. This method is conducive to the reasonable classification of adjustable resources with different regulation preferences and regulation performance, improving the flexibility of each link of the system, and promoting the construction and development of new power systems.

[0348] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A typical adjustable resource optimization method for power grid regulation scenarios, characterized in that: The specific steps include: Perform step extraction on historical operation data of different adjustable resources to obtain step information; Based on the historical operation data and step information of adjustable resources, taking into account the differences in the response characteristics of adjustable resources in different power grid regulation scenarios, typical scenario response characteristics are set, and then an adjustable resource response potential evaluation index system for power grid regulation scenarios is constructed, and the response potential evaluation indicators of various adjustable resources are calculated based on the extracted step information; Based on the evaluation indicators, taking into account the differences in matching between the response characteristics of multiple types of adjustable resources and the grid regulation requirements, a comprehensive evaluation model for adjustable resources is constructed through the response sequence corresponding to the response characteristics, and a comprehensive evaluation of the adjustable resource response potential in specific scenarios is conducted; Based on the obtained adjustable resource response feature combination and comprehensive evaluation model, the adjustable resources are grouped and optimized using the combined evaluation and neighbor propagation algorithm, and the adjustable resource groups with different response potentials are classified and optimized; Based on the classification and optimization results obtained, the user demand response potential is analyzed and energy management and demand response strategies are optimized.

2. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 1, characterized in that: Obtaining step information includes: The step is set as the local average value of the user's change rate η that is lower than a certain threshold for more than 1 hour; In the formula, Δp is the first-order difference value of the running data sequence in a certain time period selected, Δp max , Δp mean and Δp std are the maximum, mean and variance of the running data difference series respectively.

3. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 1, characterized in that: Obtaining step information includes: For the daily operation curve, the initial 4 data points of the 1h time scale are selected as the candidate step array, and the load step judgment conditions are defined; Z=P[1,2,3,4] (2) According to the step judgment condition, start traversing from the running data point corresponding to the next time point and update the candidate step array; In the formula, t represents the latest moment of traversal starting from the next moment point; If the data point P(t) at the next moment belongs to the previous step, it is stored in the previous step array Z, and then the traversal continues; if P(t) does not belong to the previous step, t=t+3; Calculate all step information W; Where W i is the information matrix of the i-th step divided, including the step height H i,ave , start time T i,start and end time T i,end , n represents the total number of steps divided.

4. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 3, characterized in that: include: According to the differentiated power equipment composition, electricity consumption habits and willingness to respond to demand of different types of adjustable resources, response potential indicators are designed according to the response characteristics to describe the response potential of adjustable resources in different response characteristic dimensions, and based on the step information extracted from the operation data, an adjustable resource response potential evaluation index system is constructed. The specific indicators include: Response characteristic 1: Response cost Indicator 1.1: User peak-to-average electricity price ratio γ1 In the formula, c i,peak is the peak electricity price for user i; c i,flat is the average electricity price for user i; Indicator 1.2: User valley-to-average electricity price ratioγ2 In the formula, c i,valley is the off-peak electricity price for user i; Response characteristic 2: Response speed Indicator 2.1: Maximum level-up speed γ3 In the formula, H valley,valley-1 and H valley,valley+1 is the height difference between the smallest step and the two adjacent steps; t valley,valley-1 and t valley,valley+1 is the transition time between the minimum step and the adjacent step; Indicator 2.2: Maximum reduction speed γ4 In the formula, H peak,peak-1 and H peak,peak+1 is the height difference between the largest step and the two adjacent steps; t peak,peak-1 and t peak,peak+1 is the transition time between the maximum step and the adjacent step; Response characteristic 3: Response capacity Indicator 3.1: The peak-to-valley difference of the average load step during the day γ5 γ5=max(μ i )-min(μ i ) (9) In the formula, μ i is the mean value of the steps on the i-th day of history; Indicator 3.2: Day-ahead load factor γ6 Indicator 3.3: Day-ahead step divergence γ7 In the formula, μ is the average value of the historical setting day steps; Indicator 3.4: Intraday transferable load γ8 In the formula, p peak (i) and p valley (i) is the maximum and minimum value of the step on the i-th day; Indicator 3.5: Intraday peak-to-valley difference rate γ9 Indicator 3.6: Intraday step volatility γ 10 In the formula, σ i , μ i are the variance and mean of the weighted step height on the i-th day; Indicator 3.7: Real-time maximum reducible capacity γ 11 In the formula, p peak-1 (i) and p peak+1 (i) is the height of the two steps to the left and right of the highest step; Indicator 3.8: Real-time reduction flexibility γ 12 In the formula, Is the historical setting number of days p peak (i) the average value; Metric 3.9: Real-time step transition time γ 13 In the formula, t peak,start (i) t peak,end (i) are the start and end time of the maximum step on day i, respectively; t peak-1,end (i) t peak+1,start (i) are the end time of the left side of the maximum step and the start time of the right side of the maximum step, respectively; Response feature 4: Response time Indicator 4.1: Average minimum durationγ 14 Where, t end (i) and t start (i) are the end time series and the start time series of the step on the i-th day; Indicator 4.2: Average maximum durationγ 15 Where, t end (i) and t start (i) are the end time series and start time series of the step on the i-th day.

5. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 4, characterized in that: include: According to the established adjustable resource potential index, there are four types of response characteristics, which are respectively recorded as response speed V x , Response cost C x , response capacity P x and response time T x ; Each response feature has its corresponding response sequence, namely C x =[γ x,1 ,γ x,2 ]、V x =[γ x,3 ,γ x,4 ]、P x =[γ x,5 ,γ x,6 ,···,γ x,13 ]、T x =[γ x,14 ,γ x,15 ], the comprehensive evaluation model of adjustable resources can be expressed as follows: Where: Val x is the comprehensive index value of the response characteristics of the adjustable resource x; C max , C min It represents the maximum and minimum values ​​of the response cost sequence in the adjustable resource group; V max 、V min Indicates the maximum and minimum values ​​of the response speed sequence in the adjustable resource group; P max , P min Indicates the maximum and minimum values ​​of the response capacity sequence in the adjustable resource group; T max 、T min Indicates the maximum and minimum values ​​of the response time sequence in the adjustable resource group; K is the weight of each response characteristic index, and the score is scored by the weighting method combining the entropy weight method and the hierarchical analysis method, where α and β are the characteristic weight combinations of the entropy weight method and the hierarchical analysis method, respectively.

6. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 5, characterized in that: include: The entropy weight method is used to perform the first evaluation weighting on the response feature combination of adjustable resources, and the weight distribution is determined by calculating the relationship between the entropy value and weight of each response feature indicator; The hierarchical analysis method is used to carry out the second evaluation weighting of the response characteristic combination of adjustable resources. The weight of each factor is determined by constructing a judgment matrix and calculating the eigenvalue, and finally the relative weight of the decision-making scheme is obtained. Based on the results of the combined evaluation and combined with the comprehensive evaluation model of adjustable resources, the comprehensive response evaluation value of the corresponding adjustable resources is calculated. By arranging in descending order, the adjustable resources with stronger comprehensive potential can be selected to obtain the optimal selection of adjustable resources. Based on the response feature combination sequence of various adjustable resources, the neighbor propagation algorithm is used to cluster different adjustable resource users, so as to identify the same type of adjustable resource users and distinguish adjustable resource user groups with different response characteristics.

7. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 6, characterized in that: include: Preprocess the raw data; The initial potential indicator matrix X is constructed for 15 evaluation indicators and n evaluation users, which is expressed as follows: By performing Euclidean normalization, x i,j Normalized into; Where, i = 1, 2, ..., 15; j = 1, 2, ..., n; Based on the normalized potential index obtained, the entropy value is calculated; In the formula, h i is the entropy value of the i-th indicator; The entropy weight is calculated based on the obtained entropy value; In the formula, ω i is the entropy weight of the i-th indicator; Based on the obtained entropy weight, the feature vector weight is calculated; α=[ω1,ω2,…,ω 15 ] (25) In the formula, α is the objectively weighted feature weight.

8. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 6, characterized in that: include: Establish a hierarchical structure, clarify decision-making goals and split the levels, including the goal level, criterion level and plan level; Construct a judgment matrix. For each level obtained, establish a judgment matrix A; In the formula, a i,j Factor a i with a j and meet the following conditions: Where, i, j = 1, 2, ... n; Based on the obtained judgment matrix, the importance is sorted and the maximum characteristic root λ is calculated according to A. max The corresponding eigenvectors are used to obtain the comprehensive calculation results, which are expressed as follows: Perform consistency check by checking the consistency ratio CR. The consistency ratio calculation formula is as follows: In the formula, CI is the consistency index, RI is the average consistency index, If CR is less than 0.1, the judgment result is reliable.

9. A typical adjustable resource optimization method for power grid regulation scenarios according to claim 6, characterized in that: include: Calculate the Euclidean distance between samples, In the formula, d(x i ,x j ) is the Euclidean distance between samples; i=1,2,…,n; j=1,2,…,n; l is the sample feature dimension; x i , x j Represents a user sample data object containing l-dimensional attributes; Calculate the similarity matrix, Where p(k) is the diagonal value of the similarity matrix; Initialize the attraction matrix and the belongingness matrix to zero matrices with the same structure as the similarity matrix; Update the attraction matrix r(i,k), r(i,k)=s(i,k)-max k′≠k {a(i,k′)+s(i,k′)} (32) Update the attribution matrix a(i,k), The above two matrices are updated cyclically until the algorithm converges or reaches the maximum number of iterations, and the corresponding cluster centers are output; According to the cluster center and the attraction matrix between samples, each sample is assigned to the corresponding cluster, and finally the clustering result is output.

10. A typical adjustable resource optimization system for power grid regulation scenarios, characterized in that: include: The data collection and processing module extracts the historical operation data of adjustable resources in steps according to the historical operation data of different adjustable resources; The evaluation index module is connected to the data collection and processing module. Based on the historical operation data obtained by the data collection and processing module and the step information extracted, it is used to take into account the differences in response characteristics of adjustable resources in different power grid regulation scenarios, set four typical scenario response characteristics of response cost, response speed, response capacity, and response time, and then construct an adjustable resource response potential evaluation index system for power grid regulation scenarios and calculate various adjustable resource response potential evaluation indicators; The comprehensive evaluation module is connected with the evaluation index calculation module. Based on the evaluation index calculated by the evaluation index calculation module, the matching degree difference between the response characteristics of multiple types of adjustable resources and the grid regulation demand is taken into account. Through the response sequence corresponding to the response characteristics, an adjustable resource comprehensive evaluation model is constructed to comprehensively evaluate the adjustable resource response potential under specific scenarios. The classification and optimization module is connected with the evaluation index module and the comprehensive evaluation module. It combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module, and uses the combined evaluation and neighbor propagation algorithm to group and optimize the adjustable resources, and classify and optimize the adjustable resource groups with different response potentials. The strategy optimization module is connected with the classification and optimization module. Based on the classification and optimization results obtained by the classification and optimization module, it analyzes the user demand response potential and optimizes energy management and demand response strategies.

11. A typical adjustable resource optimization system for power grid regulation scenarios according to claim 10, characterized in that: The classification optimization module includes: optimization unit and classification unit; The optimization unit combines the adjustable resource response feature combination and the comprehensive evaluation model obtained by the comprehensive evaluation module to select the adjustable resource groups with different response potentials through subjective and objective combined evaluation; The classification unit is connected to the comprehensive evaluation module. Based on the response feature combination sequence of various adjustable resources obtained by the comprehensive evaluation module, the neighbor propagation algorithm is used to cluster different adjustable resource users, thereby effectively identifying the same type of adjustable resource users and effectively distinguishing adjustable resource user groups with different response characteristics.

12. A typical adjustable resource optimization system for power grid regulation scenarios according to claim 11, characterized in that: The optimization unit includes: the first evaluation weighting unit, the second evaluation weighting unit, and the combined optimization unit; The first evaluation weighting subunit is connected to the comprehensive evaluation module and is used to objectively weight the response feature combination of the adjustable resources obtained by the comprehensive evaluation module using the entropy weight method, and determine the weight distribution by calculating the relationship between the entropy value and weight of each response feature indicator; The second evaluation weighting unit is connected to the comprehensive evaluation module. It uses the hierarchical analysis method to subjectively weight the response characteristic combination of adjustable resources obtained by the comprehensive evaluation module. The weight of each factor is determined by constructing a judgment matrix and calculating the eigenvalue. Finally, the relative weight of the decision-making plan is obtained for comprehensive evaluation. Combined with the optimization small unit, it is connected with the first evaluation weighting small unit, the second evaluation weighting small unit and the comprehensive evaluation module. Based on the results of subjective and objective evaluation and combined with the comprehensive evaluation model of adjustable resources, the comprehensive response evaluation value of the corresponding adjustable resources is calculated. By arranging in descending order, the adjustable resources with stronger comprehensive potential can be selected, thereby realizing the optimization of adjustable resources.

Images