Demand side response capability evaluation method for industrial user and related device

Abstract

The invention provides a demand side response capability evaluation method for an industrial user and a related device. The method comprises the following steps: acquiring basic electricity consumption information, historical response information and demand response event information; performing data acquisition on a target industrial user to obtain a load curve data set; calculating the load curve data set to obtain a typical load curve; constructing a reference data set according to the basic power utilization information, the historical response information, the demand response event information and the typical load curve; preprocessing the reference data set to obtain a target data set; dividing the target data set into a training set and a test set; obtaining a reference model, initializing model network parameters to obtain initial model network parameters, and optimizing the initial model network parameters to obtain target model network parameters; adjusting the reference model according to the target model network parameters to obtain a target model; and evaluating the target industrial user according to the target model to obtain a target evaluation score.

Description

Technical Field

[0001] The present application relates to the technical field of power planning, and in particular to a method and related device for evaluating demand-side response capability for industrial users. Background Art

[0002] When the power grid faces peak load, difficulty in absorbing renewable energy or sudden failures, by assessing the potential of users to respond to demand, power grid operators can quickly identify and mobilize user groups with high response potential and adjust their electricity consumption patterns to reduce grid pressure and avoid overload and power outages.

[0003] At present, the user's response capability can be evaluated by analyzing the typical load curve, power consumption mode, response history and other information of industrial users. It is also possible to focus on equipment characteristics and analyze the adjustable load, response time, response speed and other parameters of industrial users' production equipment, energy storage systems and other equipment to evaluate the industrial user's ability to respond to grid demand. However, when the industrial user's production process changes or the characteristic information of its equipment cannot be fully obtained, it is impossible to accurately evaluate the user's ability to respond to grid demand.

[0004] Therefore, how to improve the stability and accuracy of the evaluation results of industrial users' responsiveness on the demand side needs to be urgently addressed. Summary of the invention

[0005] The embodiments of the present application provide a method and related devices for evaluating the demand-side responsiveness of industrial users. The method improves the stability and accuracy of the evaluation results by comprehensively considering the power consumption information and historical information of the industrial users and continuously optimizing the model network parameters in order to evaluate the responsiveness of the industrial users on the demand side.

[0006] In a first aspect, an embodiment of the present application provides a method for evaluating demand-side response capability of industrial users, the method comprising:

[0007] Obtain basic electricity consumption information, historical response information, and demand response event information of target industrial users;

[0008] Collect data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set;

[0009] Calculating the load curve data set according to a preset clustering algorithm to obtain a typical load curve;

[0010] Constructing a data set according to the basic electricity consumption information, the historical response information, the demand response event information and the typical load curve to obtain a reference data set;

[0011] Preprocessing the reference data set to obtain a target data set;

[0012] Dividing the target data set into a training set and a test set according to a preset division ratio;

[0013] Acquire a preset reference model, and randomly initialize model network parameters in the reference model to obtain initial model network parameters;

[0014] Optimizing the initial model network parameters according to the training set to obtain target model network parameters;

[0015] Adjusting the reference model according to the target model network parameters to obtain a target model;

[0016] Testing the target model according to the test set to obtain a test result;

[0017] If the error between the test result and the actual result is less than a preset target threshold, the target industrial user is evaluated according to the target model to obtain a target evaluation score; the target evaluation score represents the demand-side response capability of the target industrial user.

[0018] In a second aspect, an embodiment of the present application provides a demand-side response capability evaluation device for industrial users, the device comprising an acquisition module, a collection module, a calculation module, a construction module, a preprocessing module, a division module, a control center module, a test module, and an evaluation module, wherein:

[0019] The acquisition module is used to acquire basic electricity consumption information, historical response information, and demand response event information of target industrial users;

[0020] The acquisition module is used to collect data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set;

[0021] The calculation module is used to calculate the load curve data set according to a preset clustering algorithm to obtain a typical load curve;

[0022] The construction module is used to construct a data set according to the basic power consumption information, the historical response information, the demand response event information and the typical load curve to obtain a reference data set;

[0023] The preprocessing module is used to preprocess the reference data set to obtain a target data set;

[0024] The partitioning module is used to partition the target data set into a training set and a test set according to a preset partitioning ratio;

[0025] The control center module is used to obtain a preset reference model, and randomly initialize the model network parameters in the reference model to obtain initial model network parameters; optimize the initial model network parameters according to the training set to obtain target model network parameters; adjust the reference model according to the target model network parameters to obtain the target model;

[0026] The testing module is used to test the target model according to the test set to obtain a test result;

[0027] The evaluation module is used to evaluate the target industrial user according to the target model to obtain a target evaluation score if the error between the test result and the actual result is less than a preset target threshold; the target evaluation score represents the demand-side response capability of the target industrial user.

[0028] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the program includes instructions for executing the steps of any method in the first aspect of the embodiment of the present application.

[0029] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute part or all of the steps described in any method of the first aspect of the embodiment of the present application.

[0030] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps described in any method of the first aspect of the embodiment of the present application. The computer program product may be a software installation package.

[0031] By implementing the embodiments of the present application, in the process of evaluating the responsiveness of industrial users on the demand side, the electricity consumption information and historical information of industrial users are comprehensively considered, and the model network parameters are continuously optimized to improve the stability and accuracy of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0033] Figure 1 It is a flow chart of a method for evaluating demand-side response capability of industrial users provided in an embodiment of the present application;

[0034] Figure 2 It is a flow chart of an FT-Transformer model provided in an embodiment of the present application;

[0035] Figure 3 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application;

[0036] Figure 4 This is a functional module composition block diagram of a demand-side response capability assessment device for industrial users provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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.

[0038] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0039] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article indicates that the associated objects before and after are in an "or" relationship. The "plurality" appearing in the embodiments of the present application refers to two or more.

[0040] In the embodiments of the present application, "at least one item" or similar expressions refer to any combination of these items, including any combination of single items or plural items, and refer to one or more, and multiple refers to two or more. For example, at least one item of a, b, or c can represent the following seven situations: a, b, c, a and b, a and c, b and c, a, b, and c. Among them, each of a, b, and c can be an element or a set containing one or more elements.

[0041] The "connection" that appears in the embodiments of the present application refers to various connection methods such as direct connection or indirect connection to achieve communication between devices, and the embodiments of the present application do not impose any limitations on this.

[0042] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0043] The following is an explanation of the relevant terms involved in this application, as follows:

[0044] FT-Transformer (Feature-wise Transformation Transformer) model: is a deep learning model designed to process tabular data. It combines the advantages of the Transformer architecture and innovations for the characteristics of tabular data. It can process input data through feature transformation to enhance the model's expressiveness and performance.

[0045] Multilayer Perceptron (MLP): It is a common artificial neural network model. It consists of an input layer, several hidden layers and an output layer. Each layer contains several neurons, and the neurons are connected by weights.

[0046] Fully connected layer: It is the basic component of MLP and is used to implement complex transformations of deep neural networks. It can be represented as a type of layer in a neural network, connecting every neuron in the previous layer to every neuron in the current layer.

[0047] When the power grid faces peak load, difficulty in absorbing renewable energy or sudden failure, by evaluating the potential of users to respond to demand, power grid operators can quickly identify and mobilize user groups with high response potential, and adjust their electricity consumption patterns to reduce power grid pressure and avoid overload and power outages. At present, the user's response ability can be evaluated by analyzing the typical load curve, electricity consumption pattern, response history and other information of industrial users. It can also focus on equipment characteristics and analyze the adjustable load, response time, response speed and other parameters of industrial users' production equipment, energy storage systems and other equipment to evaluate the ability of industrial users to respond to power grid demand. However, when the production process of industrial users changes or the characteristic information of their equipment cannot be fully obtained, it is impossible to accurately evaluate the ability of users to respond to power grid demand. Therefore, how to improve the stability and accuracy of the evaluation results of the response ability of industrial users on the demand side is an urgent problem to be solved.

[0048] To solve the above problems, the embodiments of the present application provide a method and related devices for evaluating the demand-side response capability of industrial users, which obtain basic electricity consumption information, historical response information, and demand response event information of target industrial users; collect data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set; calculate the load curve data set according to a preset clustering algorithm to obtain a typical load curve; construct a data set according to the basic electricity consumption information, the historical response information, the demand response event information, and the typical load curve to obtain a reference data set; pre-process the reference data set to obtain a target data set; and divide the target industrial users into the target data set according to a preset division ratio. The target data set is divided into a training set and a test set; a preset reference model is obtained, and the model network parameters in the reference model are randomly initialized to obtain the initial model network parameters; the initial model network parameters are optimized according to the training set to obtain the target model network parameters; the reference model is adjusted according to the target model network parameters to obtain the target model; the target model is tested according to the test set to obtain the test result; if the error between the test result and the true result is less than the preset target threshold, the target industrial user is evaluated according to the target model to obtain the target evaluation score; the target evaluation score represents the demand-side response capability of the target industrial user. In view of the responsiveness of industrial users on the demand side, the stability and accuracy of the evaluation results are improved by comprehensively considering the power consumption information and historical information of industrial users and continuously optimizing the model network parameters.

[0049] Combine the following Figure 1 A demand-side response capability evaluation method for industrial users in an embodiment of the present application is described. Figure 1A flow chart of a method for evaluating the demand-side response capability of industrial users provided in an embodiment of the present application specifically includes the following steps:

[0050] Step S101, obtaining basic electricity consumption information, historical response information, and demand response event information of a target industrial user.

[0051] Specifically, the basic electricity consumption information, historical response information, and demand response event information of the target industrial users can be obtained from the preset power system. Among them, the target industrial users include but are not limited to users in the automobile manufacturing industry, food processing industry, and logistics and transportation industry. The basic electricity consumption information includes but is not limited to the total electricity consumption, electricity load, and electricity consumption period. The historical response information includes but is not limited to response events, response effects, and user participation. The demand response event information includes but is not limited to the event and type of the event, event target, and event feedback, which are not specifically limited here.

[0052] Step S102: collecting data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set.

[0053] Specifically, the preset sampling frequency may be once every 15 minutes, that is, 96 times a day, which is not specifically limited here. At the same time, in order to ensure that the extracted typical load curve can well reflect the electricity consumption behavior pattern of the target industrial user, the preset sampling duration may be one year or two years, which is not specifically limited here. Data is collected from the target industrial user according to the preset sampling frequency and the preset sampling duration to obtain a load curve data set.

[0054] Step S103, calculating the load curve data set according to a preset clustering algorithm to obtain a typical load curve.

[0055] The load curve data set includes m load curve data; m is a positive integer; the load curve data set is calculated according to a preset clustering algorithm to obtain a typical load curve, and the specific steps include:

[0056] A1. Randomly select n load curve data from the m load curve data, and use the n load curve data as cluster centers respectively to obtain n cluster centers; n is a positive integer;

[0057] Specifically, n load curve data are randomly selected from m load curve data, and each load curve data in the n load curve data is used as a cluster center, and n cluster centers can be obtained. Among them, the number of cluster centers can be evaluated according to the silhouette coefficient of the load curve data set, that is, the silhouette coefficient corresponding to each load curve data is first calculated, and then all the silhouette coefficients are averaged to obtain the comprehensive silhouette coefficient, wherein the closer the comprehensive silhouette coefficient is to 1, the more ideal the clustering result is. Therefore, in order to ensure the accuracy of the clustering effect, the comprehensive silhouette coefficients corresponding to different numbers of cluster centers can be calculated, and the number of cluster centers with the comprehensive silhouette coefficient closest to 1 can be selected as the optimal number of cluster centers, that is, n cluster centers.

[0058] The calculation formula of the silhouette coefficient is as follows:

[0059]

[0060] Among them, d(i) is the silhouette coefficient corresponding to the load curve data i, p(i) is the average distance between the load curve data i and other load curve data of its corresponding cluster center, and q(i) is the minimum value of the average distance between the load curve data i and all load curves corresponding to other cluster centers.

[0061] A2. Calculate the distance between each load curve data in the m load curve data and the n cluster centers, and assign it to the cluster center with the closest distance;

[0062] Specifically, the calculation can be performed according to the Euclidean distance formula, that is, the distance between each load curve data in the m load curve data and the n cluster centers is calculated, and it is allocated to the nearest cluster center according to the principle of proximity to improve the clustering effect.

[0063] A3. After all the m load curve data are distributed, the average of all the load curve data corresponding to each of the n cluster centers is taken to obtain the optimized n cluster centers;

[0064] Specifically, after the m load curve data are distributed, the average of all load curve data corresponding to each cluster center in the n cluster centers is taken, and each average is used as the optimized cluster center to obtain the optimized n cluster centers.

[0065] A4, repeating the steps A2 and A3 until each cluster center no longer changes, obtaining n reference cluster centers, and then going to step A5;

[0066] Specifically, step A2 and step A3 are repeatedly performed to continuously optimize the cluster centers and improve the accuracy and reliability of clustering. When each cluster center no longer changes, the optimization is determined to be complete, and the final optimized cluster center is used as the reference cluster center, that is, n reference cluster centers are obtained, and then the process goes to step A5.

[0067] A5. Determine the target cluster center as the typical load curve; the target cluster center is the reference cluster center with the most corresponding load curve data among the n reference cluster centers.

[0068] Specifically, among the n reference cluster centers, a reference cluster center with the most load curve data is selected as the target cluster center, and the target cluster center is determined to be a typical load curve.

[0069] It can be seen that by evaluating the silhouette coefficient, reasonably selecting and optimizing the clustering centers, and implementing efficient data allocation and update strategies, the clustering effect can be improved while maintaining the efficiency of data processing.

[0070] Step S104: construct a data set according to the basic electricity consumption information, the historical response information, the demand response event information, and the typical load curve to obtain a reference data set.

[0071] The field information and data types of the reference dataset are as follows:

[0072]

[0073]

[0074] It should be noted that categorical data refers to data used to represent different categories or groups. It can be string data, and is often represented by labels or names rather than specific numerical values. For example, the field information corresponding to "Nature of the enterprise" includes state-owned enterprises, joint-stock companies, etc.; the field information corresponding to "Industry" includes industry, wholesale and retail, real estate, etc. Numerical data refers to data that can be represented by numbers and has quantitative meaning. It can be discrete or continuous data. For example, the field information corresponding to "Whether there is self-contained energy storage" includes 0 or 1, that is, 0 and 1 can be used to indicate whether the target industrial user has self-contained energy storage, 0 for no, and 1 for yes.

[0075] Step S105: preprocess the reference data set to obtain a target data set.

[0076] The reference data set includes a plurality of numerical data and a plurality of categorical data, and the preprocessing of the reference data set to obtain the target data set specifically includes:

[0077] Calculate the mean and standard deviation of all numerical items in the first numerical data to obtain a first mean and a first standard deviation; the first numerical data is any one of the multiple numerical data; determine a normal numerical range according to the first mean and the first standard deviation; if the first numerical data is not in the normal numerical range, determine that the first numerical data is abnormal data; determine multiple preset category items corresponding to the first category data; the first category data is any one of the multiple category data; if the category item of the first category data does not belong to the multiple preset category items, determine that the first category data is the abnormal data; determine an abnormal processing method for the abnormal data according to a preset business logic and / or data type; the data type includes a numerical type and a category type; the abnormal processing method includes at least one of the following: deletion, replacement, and interpolation; after processing the abnormal data according to the abnormal processing method, perform a preset standardization operation on all numerical data in the reference data set to obtain multiple target numerical data; perform a preset encoding operation on all category data in the reference data set to obtain multiple target category data; determine the target data set according to the multiple target numerical data and the multiple target category data.

[0078] In a possible embodiment, first, the mean and standard deviation of all numerical items in the first numerical data are calculated to obtain a first mean and a first standard deviation, wherein the first numerical data is any one of the multiple numerical data. Then, a normal numerical range is determined according to the first mean and the first standard deviation, for example, the sum of the first mean and three times the first standard deviation is the upper limit of the normal numerical range, and the difference between the first mean and three times the first standard deviation is the upper limit of the normal numerical range; if the first numerical data is not in the normal numerical range, for example, greater than the upper limit or less than the lower limit, the first numerical data is determined to be abnormal data.

[0079] In one possible embodiment, the preset category item corresponding to the first category data is determined. For example, the preset category item corresponding to the "electricity consumption type" may be "electricity consumption for automobile manufacturing", "electricity consumption for food production", or "electricity consumption for logistics and transportation", which is not specifically limited here. The first category data is any one of the multiple category data. If the category item of the first category data does not belong to the preset category item, the first category data is determined to be abnormal data. For example, the category item of the first category data contains "residential electricity consumption", and the "residential electricity consumption" does not belong to the preset category item, which proves that the first category data is abnormal data.

[0080] In one possible embodiment, the exception handling method of abnormal data is determined according to the preset business logic and / or data type, the data type includes a numerical type and a categorical type, and the exception handling method includes at least one of the following: deletion, replacement, and interpolation. Among them, the business logic is used to select a suitable processing method for abnormal data, which can be obtained based on expert experience or communication and discussion with relevant business personnel. For example, assuming that there are abnormal values ​​far from the normal range in the numerical data, and the abnormal value is a negative income, then the numerical data can be deleted. When selecting the corresponding exception handling method according to the data type, if the abnormal data is ordinary numerical data, it can be replaced with the median or mean corresponding to the numerical data. If the abnormal data is numerical data with time series, for example, typical load curve data, it can be interpolated using an interpolation method.

[0081] The step of performing a preset standardization operation on all the numerical data in the reference data set to obtain a plurality of target numerical data specifically includes:

[0082] All the numerical data are converted according to a preset standardization formula to obtain the multiple target numerical data; wherein the standardization formula is as follows:

[0083]

[0084] in, is the i-th numerical item in the standardized reference numerical data; the reference numerical data is any numerical data among all the numerical data; x i is the i-th numerical item of the reference numerical data; μ is the mean of all numerical items in the reference numerical data; σ is the standard deviation of all numerical items in the reference numerical data.

[0085] It should be noted that the numerical data X can also be converted by normalization, which is not specifically limited here. The numerical data X is scaled to the range of 0-1 according to a preset normalization formula, wherein the normalization formula is as follows:

[0086]

[0087] in, is the i-th numerical item in the normalized numerical data X; x i is the i-th numerical item of the numerical data X; max(X) is the maximum value of all numerical items in the numerical data X; min(X) is the minimum value of all numerical items in the numerical data X.

[0088] The step of performing a preset encoding operation on all the category data in the reference data set to obtain a plurality of target category data specifically includes:

[0089] Acquire multiple category items in reference category data; the reference category data is any category data among all the category data; assign a binary vector to each category item in the multiple category items according to a preset one-hot encoding method to obtain multiple binary vectors; each category item corresponds to a binary vector; determine the target category data corresponding to the reference category data according to the multiple binary vectors.

[0090] In a possible embodiment, assuming that the categorical data corresponding to "industry" includes the first industry, the second industry, and the third industry, a binary vector is assigned to each industry according to the preset one-hot encoding method, and the value in the binary vector can only be 0 or 1, that is, the binary vector corresponding to the first industry is [1,0,0], the binary vector corresponding to the second industry is [0,1,0], and the binary vector corresponding to the third industry is [0,0,1]. It should be noted that it is also possible to assign a unique integer value to each category item according to the preset label encoding method, for example, the label code corresponding to the first industry is 0, the label code corresponding to the second industry is 1, and the label code corresponding to the third industry is 2.

[0091] Step S106, dividing the target data set into a training set and a test set according to a preset division ratio.

[0092] Specifically, the preset division ratio can be 7:3 or 8:2, which is not specifically limited here. If it is necessary to divide the validation set and optimize the hyperparameters of the model according to the validation set, the target data set can be divided into a training set, a test set, and a validation set according to a division ratio of 7:2:1, which is not specifically limited here.

[0093] Step S107, obtaining a preset reference model, and randomly initializing model network parameters in the reference model to obtain initial model network parameters.

[0094] Specifically, after obtaining the preset reference model, the model network parameters in the reference model are randomly initialized to obtain the initial model network parameters. For example, the reference model may include an input layer, a hidden layer, and an output layer. The connection weights from the input layer to the hidden layer may be randomly initialized in a normal distribution manner, and the bias may be initialized to 0 or a small random value.

[0095] It should be noted that the preset reference model can be an FT-Transformer model, which is not specifically limited here. The model is used to capture the complex relationship between data and explore the inherent key factors that affect the user's demand response ability, thereby evaluating the comprehensive response ability of industrial users to demand events. Among them, the model network parameters include but are not limited to weights, biases, and hyperparameters. The weights determine the intensity of information transmission. The bias is a constant added to the output of each neuron and is used to adjust the output of the model. Hyperparameters such as learning rate and batch size have a certain impact on the training process of the model.

[0096] Step S108, optimizing the initial model network parameters according to the training set to obtain target model network parameters.

[0097] The step of optimizing the initial model network parameters according to the training set to obtain the target model network parameters specifically includes:

[0098] B1. Inputting the sample data in the training set into the reference model according to the initial model network parameters, and outputting the predicted comprehensive response capability score;

[0099] Specifically, the reference model is configured according to the initial model network parameters, for example, the reference model is configured according to the initial weights and biases, and the sample data in the training set is input into the reference model, and the reference model outputs a predicted comprehensive response capability score.

[0100] B2. Calculate the predicted comprehensive response capability score according to a preset root mean square error calculation formula to obtain a reference loss value;

[0101] The root mean square error calculation formula is as follows:

[0102]

[0103] Wherein, R is the reference loss value; n is the number of sample data in the training set; Score the predicted comprehensive responsiveness; i Score the true overall responsiveness;

[0104] B3, calculating the gradient value of the reference loss value to the initial model network parameter to obtain a reference gradient value;

[0105] Specifically, the gradient value of the reference loss value to the initial model network parameter is calculated, that is, the partial derivative of the loss function with respect to each initial model network parameter is calculated, and the loss function is the root mean square error calculation formula.

[0106] B4. Calculate the reference gradient value and the preset learning rate according to a preset optimization calculation formula to obtain reference model network parameters;

[0107] The optimization calculation formula is as follows:

[0108]

[0109] Wherein, θ′ is the reference model network parameter; θ is the initial model network parameter; ε is the preset learning rate; is the reference gradient value;

[0110] It should be noted that the preset learning rate can be 0.001, and the learning rate can be adjusted according to the validation set to optimize the performance of the model, which is not specifically limited here.

[0111] B5. Repeat steps B1-B4 until the change in the reference loss value is less than a preset reference threshold, thereby obtaining the target model network parameters.

[0112] Specifically, after each execution of steps B1-B4, the reference loss value is recorded. If the change between the recorded reference loss values ​​is greater than or equal to the preset reference threshold, the training cycle is continued and steps B1-B4 are repeated. If the change between the recorded reference loss values ​​is less than the preset reference threshold, the training cycle is stopped, and it is determined that the model has converged to obtain the target model network parameters.

[0113] It can be seen that by stopping training when the loss change is less than the threshold, the risk of overfitting can be reduced, ensuring that the model has better generalization ability and saving computing cost and time.

[0114] Step S109: adjusting the reference model according to the target model network parameters to obtain a target model.

[0115] Specifically, the target model network parameters may include a target weight and a target bias. The target model may be obtained by adjusting the reference model according to the target weight and the target bias.

[0116] Step S110, testing the target model according to the test set to obtain a test result.

[0117] Specifically, the data of the test set is input into the target model for testing, and the target model outputs the test result, which is the predicted comprehensive responsiveness score.

[0118] Step S111, if the error between the test result and the actual result is less than a preset target threshold, the target industrial user is evaluated according to the target model to obtain a target evaluation score; the target evaluation score represents the demand-side response capability of the target industrial user.

[0119] Specifically, if the error between the test result and the true result is less than the preset target threshold, it means that the target model is available, and the target industrial user is evaluated according to the target model to obtain the target evaluation score. To a certain extent, the smaller the preset target threshold, the higher the evaluation accuracy of the target model.

[0120] In one possible embodiment, if the error between the test result and the true result is greater than or equal to the preset target threshold, it means that the target model is unavailable. When the target model is unavailable, the model network parameters of the target model can be retrained and optimized through the training set to obtain the optimized model network parameters, and the target model corresponding to the optimized model network parameters can be tested through the test set. If the error between the test result and the true result is still greater than or equal to the preset target threshold, the model network parameters of the target model are trained and optimized again through the training set until the error of the target model corresponding to the optimized model network parameters to the test set is less than the preset target threshold, then the iterative process is terminated, and the latest model network parameters are determined to be the optimal model network parameters, and the corresponding optimal target model is saved to facilitate the subsequent loading of the optimal target model and the corresponding prediction score for the target industrial user.

[0121] The step of evaluating the target industrial user according to the target model to obtain a target evaluation score includes:

[0122] Creating a target interface according to the target model; the target interface is used to receive an evaluation request and call the target model for evaluation;

[0123] Predicting the target industrial user through the target interface to obtain multiple evaluation label values;

[0124] Calculate the multiple evaluation tag values ​​according to a preset comprehensive scoring calculation formula to obtain a target evaluation score;

[0125] The comprehensive score calculation formula is as follows:

[0126] s=k1*c1+k2*c2+k3*c3+k4*c4+k5*c5+k6*c6

[0127] Among them, s represents the target evaluation score; c1-c6 are the multiple evaluation label values; k1-k6 are the weight parameters corresponding to each evaluation label value in the multiple evaluation label values.

[0128] Among them, the target evaluation score ranges from 0 to 1, and each evaluation label corresponds to an evaluation label value. The corresponding weight coefficient can be determined by analysis based on expert experience. For example, k1 is 0.4, k2 is 0.2, k3 is 0.15, k4 is 0.1, k5 is 0.1, and k6 is 0.05.

[0129] It should be noted that the evaluation tags include the maximum adjustable load, response speed, response duration, response timeliness, response frequency, and response percentage. Among them, the maximum adjustable load is the maximum load that the user can reduce during the demand response event; the response speed is the time interval from the user receiving the demand response event notification to the start of the response; the response duration is the actual duration of the user's response event; the response timeliness is the deviation between the time when the user responds to the event and the response time specified by the event, that is, the difference between the user's response time and the time when the event specifies the start of the response; the response frequency is the frequency of the user's participation in the demand response within the preset time period, that is, the number of events that the user responds to within the preset time period divided by the total number of events within the time period; the response percentage is the percentage that the user can reduce relative to the normal power consumption during the demand response event. In order to avoid the calculation of the comprehensive response capability score being affected by the size of the dimension, the evaluation tag values ​​corresponding to all evaluation tags can be normalized using the normalization method.

[0130] It can be seen that evaluating multiple key factors can comprehensively reflect the user's response ability in demand response events and help improve the model's prediction accuracy of user response behavior.

[0131] Combine the following Figure 2 The specific process of an FT-Transformer model in an embodiment of the present application is described. Figure 2 This is a flow chart of an FT-Transformer model provided in an embodiment of the present application. The FT-Transformer model converts structured data into an embedding vector and uses a self-attention mechanism to capture the complex relationship between features. Finally, all feature data information is integrated through a fully connected layer to output a comprehensive responsiveness score for industrial users.

[0132] Among them, first, in the numerical embedding layer, the numerical data can be linearly transformed and biased to map it to an embedding vector. Among them, the weight corresponding to each numerical data is independent, so each numerical data has its own separate transformation parameters. Use the category embedding layer to directly map the category data to an embedding vector. Among them, each category corresponds to a unique vector, which is independent and can capture the relationship between different categories. Then, the generated embedding vectors are concatenated row by row to form an input matrix, and the matrix is ​​input into the encoder part of the model. Among them, the encoder part can use the multi-head self-attention mechanism to process the embedding vector and output complex data vector information. The multi-head self-attention mechanism allows the model to focus on different features in different subspaces, thereby capturing more key factors related to the target variable and being able to learn the complex relationship between data in parallel. Finally, the data vector information is passed to the fully connected layer, where the function of the fully connected layer is to integrate the complex data vector information into the final score, that is, the user's comprehensive responsiveness score can be output.

[0133] It can be seen that by leveraging the powerful characteristics of the FT-Transformer model to mine and integrate complex relationships in the data, it is beneficial to obtain a more accurate and comprehensive comprehensive responsiveness score.

[0134] Combine the following Figure 3 The electronic device in the embodiment of the present application is described. Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the electronic device includes one or more processors, a memory, a communication interface and one or more programs, and the processor is communicatively connected with the memory and the communication interface via an internal communication bus.

[0135] Among them, the processor is mainly used for:

[0136] Obtain basic electricity consumption information, historical response information, and demand response event information of target industrial users;

[0137] Collect data from target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set;

[0138] The load curve data set is calculated according to a preset clustering algorithm to obtain a typical load curve;

[0139] Construct a data set based on basic electricity consumption information, historical response information, demand response event information, and typical load curves to obtain a reference data set;

[0140] Preprocess the reference data set to obtain the target data set;

[0141] Divide the target data set into a training set and a test set according to a preset division ratio;

[0142] Obtain a preset reference model, and randomly initialize the model network parameters in the reference model to obtain initial model network parameters;

[0143] Optimize the initial model network parameters according to the training set to obtain the target model network parameters;

[0144] The reference model is adjusted according to the network parameters of the target model to obtain the target model;

[0145] Test the target model according to the test set and obtain the test results;

[0146] If the error between the test result and the actual result is less than the preset target threshold, the target industrial user is evaluated according to the target model to obtain a target evaluation score; the target evaluation score represents the demand-side response capability of the target industrial user.

[0147] The one or more programs are stored in the above-mentioned memory and are configured to be executed by the above-mentioned processor, and the one or more programs include instructions for executing any step in the above-mentioned method embodiment.

[0148] Among them, the processor can be, for example, a central processing unit (CPU), a general processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logic blocks, units and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements a computing function, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like. The communication unit can be a communication interface, a transceiver, a transceiver circuit, etc., and the storage unit can be a memory.

[0149] The memory may be a volatile memory or a nonvolatile memory, or may include both volatile and nonvolatile memories. Among them, the nonvolatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DR RAM).

[0150] It is understandable that the electronic device may include more or fewer structural elements than those in the above structural block diagram, for example, including a power module, physical buttons, a Wi-Fi module, a speaker, a Bluetooth module, a sensor, a display module, etc., which are not limited here.

[0151] The above mainly introduces the scheme of the embodiment of the present application from the perspective of the execution process on the method side. It is understandable that, in order to realize the above functions, the electronic device includes a hardware structure and / or software module corresponding to the execution of each function. It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments provided herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present application.

[0152] The embodiment of the present application can divide the electronic device into functional units according to the above method example. For example, each functional unit can be divided according to each function, or two or more functions can be integrated into one processing unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units. It should be noted that the division of units in the embodiment of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0153] In the case of dividing each functional module into corresponding functional modules, Figure 4 A functional module composition block diagram of a demand-side response capability evaluation device for industrial users provided in an embodiment of the present application, wherein the demand-side response capability evaluation device 400 for industrial users includes an acquisition module 410, a collection module 420, a calculation module 430, a construction module 440, a preprocessing module 450, a division module 460, a control center module 470, a test module 480 and an evaluation module 490, wherein:

[0154] The acquisition module 410 is used to acquire basic electricity consumption information, historical response information, and demand response event information of target industrial users;

[0155] The acquisition module 420 is used to collect data from the target industrial user according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set;

[0156] The calculation module 430 is used to calculate the load curve data set according to a preset clustering algorithm to obtain a typical load curve;

[0157] The construction module 440 is used to construct a data set according to the basic power consumption information, the historical response information, the demand response event information and the typical load curve to obtain a reference data set;

[0158] The preprocessing module 450 is used to preprocess the reference data set to obtain a target data set;

[0159] The partitioning module 460 is used to partition the target data set into a training set and a test set according to a preset partitioning ratio;

[0160] The control center module 470 is used to obtain a preset reference model, and randomly initialize the model network parameters in the reference model to obtain initial model network parameters; optimize the initial model network parameters according to the training set to obtain target model network parameters; adjust the reference model according to the target model network parameters to obtain the target model;

[0161] The testing module 480 is used to test the target model according to the test set to obtain a test result;

[0162] The evaluation module 490 is used to evaluate the target industrial user according to the target model to obtain a target evaluation score if the error between the test result and the actual result is less than a preset target threshold; the target evaluation score represents the demand-side response capability of the target industrial user.

[0163] Optionally, the load curve data set includes m load curve data; m is a positive integer; in calculating the load curve data set according to a preset clustering algorithm to obtain a typical load curve, the calculation module 430 is specifically used to:

[0164] A1. Randomly select n load curve data from the m load curve data, and use the n load curve data as cluster centers respectively to obtain n cluster centers; n is a positive integer;

[0165] A2. Calculate the distance between each load curve data in the m load curve data and the n cluster centers, and assign it to the cluster center with the closest distance;

[0166] A3. After all the m load curve data are distributed, the average of all the load curve data corresponding to each of the n cluster centers is taken to obtain the optimized n cluster centers;

[0167] A4, repeating the steps A2 and A3 until each cluster center no longer changes, obtaining n reference cluster centers, and then going to step A5;

[0168] A5. Determine the target cluster center as the typical load curve; the target cluster center is the reference cluster center with the most corresponding load curve data among the n reference cluster centers.

[0169] Optionally, the reference data set includes a plurality of numerical data and a plurality of categorical data. In preprocessing the reference data set to obtain the target data set, the preprocessing module 450 is specifically used to:

[0170] Calculating the mean and standard deviation of all numerical items in the first numerical data to obtain a first mean and a first standard deviation; the first numerical data is any one of the multiple numerical data;

[0171] Determine a normal value range according to the first mean and the first standard deviation;

[0172] If the first numerical data is not within the normal numerical range, determining that the first numerical data is abnormal data;

[0173] Determine a plurality of preset category items corresponding to the first category data; the first category data is any one of the plurality of category data;

[0174] If the category item of the first category type data does not belong to the plurality of preset category items, determining that the first category type data is the abnormal data;

[0175] Determine the abnormal data exception handling method according to the preset business logic and / or data type; the data type includes a numerical type and a categorical type; the abnormal data handling method includes at least one of the following: deletion, replacement, and interpolation;

[0176] After processing the abnormal data according to the abnormal processing method, performing a preset standardization operation on all the numerical data in the reference data set to obtain a plurality of target numerical data;

[0177] Performing a preset encoding operation on all the category data in the reference data set to obtain a plurality of target category data;

[0178] The target data set is determined according to the plurality of target numerical data and the plurality of target categorical data.

[0179] Optionally, in the aspect of performing a preset standardization operation on all the numerical data in the reference data set to obtain a plurality of target numerical data, the preprocessing module 450 is further specifically used for:

[0180] All the numerical data are converted according to a preset standardization formula to obtain the multiple target numerical data; wherein the standardization formula is as follows:

[0181]

[0182] in, is the i-th numerical item in the standardized reference numerical data; the reference numerical data is any numerical data among all the numerical data; x i is the i-th numerical item of the reference numerical data; μ is the mean of all numerical items in the reference numerical data; σ is the standard deviation of all numerical items in the reference numerical data.

[0183] Optionally, in the aspect of performing a preset encoding operation on all the category data in the reference data set to obtain a plurality of target category data, the preprocessing module 450 is further specifically used to:

[0184] Acquire multiple category items in reference category data; the reference category data is any one category data among all the category data;

[0185] Assigning a binary vector to each of the multiple category items according to a preset one-hot encoding method to obtain multiple binary vectors; each category item corresponds to a binary vector;

[0186] The target categorical data corresponding to the reference categorical data is determined according to the multiple binary vectors.

[0187] Optionally, in optimizing the initial model network parameters according to the training set to obtain target model network parameters, the control center module 470 is specifically used to:

[0188] B1. Inputting the sample data in the training set into the reference model according to the initial model network parameters, and outputting the predicted comprehensive response capability score;

[0189] B2. Calculate the predicted comprehensive response capability score according to a preset root mean square error calculation formula to obtain a reference loss value;

[0190] The root mean square error calculation formula is as follows:

[0191]

[0192] Wherein, R is the reference loss value; n is the number of sample data in the training set; Score the predicted comprehensive responsiveness; i Score the true overall responsiveness;

[0193] B3, calculating the gradient value of the reference loss value to the initial model network parameter to obtain a reference gradient value;

[0194] B4. Calculate the reference gradient value and the preset learning rate according to a preset optimization calculation formula to obtain reference model network parameters;

[0195] The optimization calculation formula is as follows:

[0196]

[0197] Wherein, θ′ is the reference model network parameter; θ is the initial model network parameter; ε is the preset learning rate; is the reference gradient value;

[0198] B5. Repeat steps B1-B4 until the change in the reference loss value is less than a preset reference threshold, thereby obtaining the target model network parameters.

[0199] Optionally, in evaluating the target industrial user according to the target model to obtain a target evaluation score, the evaluation module 490 is specifically used to:

[0200] Creating a target interface according to the target model; the target interface is used to receive an evaluation request and call the target model for evaluation;

[0201] Predicting the target industrial user through the target interface to obtain multiple evaluation label values;

[0202] Calculate the multiple evaluation tag values ​​according to a preset comprehensive scoring calculation formula to obtain a target evaluation score;

[0203] The comprehensive score calculation formula is as follows:

[0204] s=k1*c1+k2*c2+k3*c3+k4*c4+k5*c5+k6*c6

[0205] Among them, s represents the target evaluation score; c1-c6 are the multiple evaluation label values; k1-k6 are the weight parameters corresponding to each evaluation label value in the multiple evaluation label values.

[0206] It can be seen that in the process of evaluating the responsiveness of industrial users on the demand side, the electricity consumption information and historical information of industrial users are comprehensively considered, and the model network parameters are continuously optimized to improve the stability and accuracy of the evaluation results.

[0207] It should be noted that the specific implementation of each operation can adopt the corresponding description of the method embodiment shown above. The demand-side response capability evaluation device 400 for industrial users can be used to execute the above method embodiment of the present application, which will not be repeated here.

[0208] An embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute part or all of the steps of any method recorded in the above method embodiments, and the above computer includes an electronic device.

[0209] The embodiment of the present application also provides a computer program product, the computer program product includes a non-transitory computer-readable storage medium storing a computer program, the computer program is operable to cause a computer to execute some or all of the steps of any method described in the method embodiment. The computer program product may be a software installation package, and the computer includes an electronic device.

[0210] It should be noted that, for the above-mentioned various embodiments, for the sake of simple description, they are all expressed as a series of action combinations. Those skilled in the art should be aware that the present application is not limited by the described order of actions, because some steps in the embodiments of the present application can be performed in other orders or simultaneously. In addition, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions, steps, modules or units involved are not necessarily required by the embodiments of the present application.

[0211] In the above embodiments, the embodiments of the present application have different focuses on the description of each embodiment. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0212] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

[0213] The steps of the method or algorithm described in the embodiments of the present application can be implemented in hardware or by executing software instructions by a processor. The software instructions can be composed of corresponding software modules, and the software modules can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (electrically EPROM, EEPROM), registers, hard disks, mobile hard disks, read-only compact disks (CD-ROMs) or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC. In addition, the ASIC can be located in a terminal device or a management device. Of course, the processor and the storage medium can also be present in a terminal device or a management device as discrete components.

[0214] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiments of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server, or data center to another website site, computer, server, or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0215] The modules / units included in the devices and products described in the above embodiments may be software modules / units or hardware modules / units, or may be partially software modules / units and partially hardware modules / units. For example, for the devices and products applied to or integrated in the chip, the modules / units included therein may all be implemented in the form of hardware such as circuits, or at least some of the modules / units may be implemented in the form of software programs, which run on the processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits; for the devices and products applied to or integrated in the chip module, the modules / units included therein may all be implemented in the form of hardware such as circuits, and different modules / units may be located in the same component (such as a chip, circuit module, etc.) or in different components of the chip module, or at least some of the modules / units may be implemented in the form of software programs. The software programs run on the processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits. It is implemented in the form of a software program, which runs on a processor integrated inside the chip module, and the remaining (if any) modules / units can be implemented in hardware such as circuits; for various devices and products applied to or integrated in the terminal equipment, the various modules / units contained therein can be implemented in hardware such as circuits, and different modules / units can be located in the same component (for example, chip, circuit module, etc.) or in different components in the terminal equipment, or, at least some modules / units can be implemented in the form of a software program, which runs on a processor integrated inside the terminal equipment, and the remaining (if any) modules / units can be implemented in hardware such as circuits.

[0216] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above description is only the specific implementation method of the embodiments of the present application and is not intended to limit the protection scope of the embodiments of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A method for evaluating the demand-side response capability of industrial users, characterized in that: The method comprises: Obtain basic electricity consumption information, historical response information, and demand response event information of target industrial users; Collect data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set; Calculating the load curve data set according to a preset clustering algorithm to obtain a typical load curve; Constructing a data set according to the basic electricity consumption information, the historical response information, the demand response event information and the typical load curve to obtain a reference data set; Preprocessing the reference data set to obtain a target data set; Dividing the target data set into a training set and a test set according to a preset division ratio; Acquire a preset reference model, and randomly initialize model network parameters in the reference model to obtain initial model network parameters; Optimizing the initial model network parameters according to the training set to obtain target model network parameters; Adjusting the reference model according to the target model network parameters to obtain a target model; Testing the target model according to the test set to obtain a test result; If the error between the test result and the actual result is less than a preset target threshold, the target industrial user is evaluated according to the target model to obtain a target evaluation score; the target evaluation score represents the demand-side response capability of the target industrial user.

2. The method according to claim 1, characterized in that The load curve data set includes m load curve data; m is a positive integer; The step of calculating the load curve data set according to a preset clustering algorithm to obtain a typical load curve includes: A1. Randomly select n load curve data from the m load curve data, and use the n load curve data as cluster centers respectively to obtain n cluster centers; n is a positive integer; A2. Calculate the distance between each load curve data in the m load curve data and the n cluster centers, and assign it to the cluster center with the closest distance; A3. After all the m load curve data are distributed, the average of all the load curve data corresponding to each of the n cluster centers is taken to obtain the optimized n cluster centers; A4, repeating the steps A2 and A3 until each cluster center no longer changes, obtaining n reference cluster centers, and then going to step A5; A5. Determine the target cluster center as the typical load curve; the target cluster center is the reference cluster center with the most corresponding load curve data among the n reference cluster centers.

3. The method according to claim 1, characterized in that The reference data set includes a plurality of numerical data and a plurality of categorical data, and the preprocessing of the reference data set to obtain the target data set includes: Calculating the mean and standard deviation of all numerical items in the first numerical data to obtain a first mean and a first standard deviation; the first numerical data is any one of the multiple numerical data; Determine a normal value range according to the first mean and the first standard deviation; If the first numerical data is not within the normal numerical range, determining that the first numerical data is abnormal data; Determine a plurality of preset category items corresponding to the first category data; the first category data is any one of the plurality of category data; If the category item of the first category type data does not belong to the plurality of preset category items, determining that the first category type data is the abnormal data; Determine the abnormal data exception handling method according to the preset business logic and / or data type; the data type includes a numerical type and a categorical type; the abnormal data handling method includes at least one of the following: deletion, replacement, and interpolation; After processing the abnormal data according to the abnormal processing method, performing a preset standardization operation on all the numerical data in the reference data set to obtain a plurality of target numerical data; Performing a preset encoding operation on all the category data in the reference data set to obtain a plurality of target category data; The target data set is determined according to the plurality of target numerical data and the plurality of target categorical data.

4. The method according to claim 3, characterized in that The performing of a preset standardization operation on all the numerical data in the reference data set to obtain a plurality of target numerical data includes: Converting all the numerical data according to a preset standardization formula to obtain the plurality of target numerical data; The standardized formula is as follows: in, is the i-th numerical item in the standardized reference numerical data; the reference numerical data is any numerical data among all the numerical data; x i is the i-th numerical item of the reference numerical data; μ is the mean of all numerical items in the reference numerical data; σ is the standard deviation of all numerical items in the reference numerical data.

5. The method according to claim 3, characterized in that The performing of a preset encoding operation on all the category data in the reference data set to obtain a plurality of target category data includes: Acquire multiple category items in reference category data; the reference category data is any one category data among all the category data; Assigning a binary vector to each of the multiple category items according to a preset one-hot encoding method to obtain multiple binary vectors; each category item corresponds to a binary vector; The target categorical data corresponding to the reference categorical data is determined according to the multiple binary vectors.

6. The method according to any one of claims 1 to 5, characterized in that: The step of optimizing the initial model network parameters according to the training set to obtain target model network parameters includes: B1. Inputting the sample data in the training set into the reference model according to the initial model network parameters, and outputting the predicted comprehensive response capability score; B2. Calculate the predicted comprehensive response capability score according to a preset root mean square error calculation formula to obtain a reference loss value; The root mean square error calculation formula is as follows: Wherein, R is the reference loss value; n is the number of sample data in the training set; Score the predicted comprehensive responsiveness; i Score the true overall responsiveness; B3, calculating the gradient value of the reference loss value to the initial model network parameter to obtain a reference gradient value; B4. Calculate the reference gradient value and the preset learning rate according to a preset optimization calculation formula to obtain reference model network parameters; The optimization calculation formula is as follows: Wherein, θ′ is the reference model network parameter; θ is the initial model network parameter; ε is the preset learning rate; is the reference gradient value; B5. Repeat steps B1-B4 until the change in the reference loss value is less than a preset reference threshold, thereby obtaining the target model network parameters.

7. The method according to claim 1, characterized in that The step of evaluating the target industrial user according to the target model to obtain a target evaluation score includes: Creating a target interface according to the target model; the target interface is used to receive an evaluation request and call the target model for evaluation; Predicting the target industrial user through the target interface to obtain multiple evaluation label values; Calculate the multiple evaluation tag values ​​according to a preset comprehensive scoring calculation formula to obtain a target evaluation score; The comprehensive score calculation formula is as follows: s=k1*c1+k2*c2+k3*c3+k4*c4+k5*c5+k6*c6 Among them, s represents the target evaluation score; c1-c6 are the multiple evaluation label values; k1-k6 are the weight parameters corresponding to each evaluation label value in the multiple evaluation label values.

8. A demand-side response capability evaluation device for industrial users, characterized in that: The device includes an acquisition module, a collection module, a calculation module, a construction module, a preprocessing module, a division module, a control center module, a test module and an evaluation module, wherein: The acquisition module is used to acquire basic electricity consumption information, historical response information, and demand response event information of target industrial users; The acquisition module is used to collect data from the target industrial users according to a preset sampling frequency and a preset sampling duration to obtain a load curve data set; The calculation module is used to calculate the load curve data set according to a preset clustering algorithm to obtain a typical load curve; The construction module is used to construct a data set according to the basic power consumption information, the historical response information, the demand response event information and the typical load curve to obtain a reference data set; The preprocessing module is used to preprocess the reference data set to obtain a target data set; The partitioning module is used to partition the target data set into a training set and a test set according to a preset partitioning ratio; The control center module is used to obtain a preset reference model, and randomly initialize the model network parameters in the reference model to obtain initial model network parameters; optimize the initial model network parameters according to the training set to obtain target model network parameters; adjust the reference model according to the target model network parameters to obtain the target model; The testing module is used to test the target model according to the test set to obtain a test result; The evaluation module is used to evaluate the target industrial user according to the target model to obtain a target evaluation score if the error between the test result and the actual result is less than a preset target threshold; the target evaluation score represents the demand-side response capability of the target industrial user.

9. An electronic device, characterized in that: include: A processor, a memory, a communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, the programs comprising instructions for executing the steps in the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 7.

Images