Industrial load regulation response potential prediction method, device, equipment and medium

Abstract

The invention relates to an industry load regulation response potential prediction method and device, equipment and a medium, and belongs to the technical field of big data analysis. According to the method, firstly, historical power consumption data of users in the industry are clustered, typical power consumption scenes of the users are described by clustering centers of clusters, and prediction of future short-term power consumption data of the users is completed based on specific types of the typical power consumption scenes where the users are located currently; and determining the user load regulation quantity based on the prediction result and the upper and lower bounds of the clustering cluster, thereby completing the judgment of the industry load regulation response potential. According to the method, data acquisition is simple, prediction is based on completely real and reliable historical data, prediction is carried out based on the accurately obtained subdivided typical power consumption scene, and the judgment accuracy and judgment efficiency of the industry load regulation response potential can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of big data analysis technology, and in particular to a method, device, equipment and medium for predicting industry load regulation response potential. Background Art

[0002] Under the background of new power system construction and power system reform, the traditional demand response model based on administrative means (orderly power consumption) can no longer adapt to the requirements of instantaneous consumption of new energy and large-scale grid stabilization. Demand-side management urgently needs to transform into an "active response" model driven by market mechanisms, that is, the demand-side load needs to be actively adjusted to participate in the instantaneous consumption and stabilization process of the grid to ensure the quality of power supply. To this end, it is necessary to first accurately grasp the size of the load adjustment response of each industry on the demand side.

[0003] Although the industry has carried out load resource investigation and built an adjustable load resource library, providing a certain data basis for demand-side management, from a practical point of view, there are still problems such as insufficient exploration of demand-side resource potential, and load aggregation and regulation strategies that cannot meet the efficient configuration needs of massive adjustable load resources. At the same time, the current determination of the adjustable load on the demand side still relies to a large extent on numerical weather data and customer load adjustment willingness survey data. This method of determining the adjustable load has the problems of cumbersome data acquisition and inaccurate estimated adjustable load due to a large proportion of human subjective factors.

[0004] That is, the current method of estimating the adjustable load of the industry still has technical problems of low efficiency and inaccuracy, resulting in poor regulation effect on the demand side of the current power grid. How to efficiently and accurately grasp the adjustable load of the industry is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method, device, equipment and medium for predicting the response potential of industry load regulation to solve the technical problem that the current industry adjustable load estimation is not efficient and accurate enough.

[0006] In a first aspect, a method for predicting industry load regulation response potential is provided, the method comprising:

[0007] Acquire daily electricity consumption data of current users in the current industry, take the latest historical electricity consumption data of the current users for a first set period of time as historical data, and cluster the historical data to obtain various clusters;

[0008] Determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the clusters;

[0009] Taking the cluster center of the cluster to which the current user's daily electricity consumption data belongs as a specific cluster center, and predicting the current user's tomorrow's electricity consumption data based on the specific cluster center and the historical data;

[0010] The load regulation response amount of the current industry is determined based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

[0011] In a second aspect, a device for predicting the response potential of industry load regulation is provided, the device comprising:

[0012] A data acquisition clustering module is used to acquire the daily electricity consumption data of the current user in the current industry, take the latest historical electricity consumption data of the current user for a first set time as historical data, and cluster the historical data to obtain various clusters;

[0013] A cluster parameter determination module, used for determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the cluster;

[0014] A user power consumption data prediction module, configured to use the cluster center of the cluster to which the current user's power consumption data for the day belongs as a specific cluster center, and predict the current user's power consumption data for tomorrow based on the specific cluster center and the historical data;

[0015] The industry load regulation potential determination module is used to determine the load regulation response of the current industry based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

[0016] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the industry load regulation response potential prediction method as described in the first aspect when executing the computer program.

[0017] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting the industry load regulation response potential as described in the first aspect is implemented.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The present invention is based on the acquisition of historical data of electricity users in the industry, abandons the consideration of data on users' willingness to participate in load regulation, and completely uses real historical data as the basis for predicting the load regulation response potential. After clustering the historical data, typical electricity consumption curves are obtained to characterize the user's electricity consumption behavior under different electricity consumption influencing factors, without the need to obtain numerical weather data, which is difficult to obtain, to complete the prediction. In the prediction process, the typical electricity consumption scenario of the current user's electricity consumption behavior is considered, and future short-term electricity consumption data is predicted based on the user's current typical electricity consumption scenario combined with historical data, and the user's ability to participate in demand-side load regulation is determined, thereby reflecting the industry's load regulation response potential. The present invention completes the prediction based entirely on real historical data, and the user's electricity consumption behavior is segmented before the prediction, and subsequent predictions are carried out based on the segmentation results, which can significantly improve the prediction accuracy and efficiency, thereby achieving efficient and accurate judgment of the industry's load regulation response potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0021] Figure 1 This is a schematic diagram of an application environment of a method for predicting industry load regulation response potential provided by the first embodiment of the present invention;

[0022] Figure 2 It is a flow chart of a method for predicting the potential of industry load regulation response provided by the first embodiment of the present invention;

[0023] Figure 3 It is a schematic diagram of clusters obtained after clustering historical data of a manufacturing enterprise provided by the first embodiment of the present invention;

[0024] Figure 4 It is a schematic diagram of cluster centers of various clusters obtained after clustering historical data of a manufacturing enterprise provided by the first embodiment of the present invention;

[0025] Figure 5 This is a schematic diagram of comparing the predicted value and the actual value obtained by predicting the electricity consumption data of a manufacturing enterprise tomorrow provided by the first embodiment of the present invention;

[0026] Figure 6 This is a schematic diagram of load prediction values ​​and load upper and lower limits provided in the first embodiment of the present invention;

[0027] Figure 7It is a structural schematic diagram of an industry load regulation response potential prediction device provided by Embodiment 2 of the present invention;

[0028] Figure 8 It is a structural diagram of a computer device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0029] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0030] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0031] It should also be understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0032] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]", depending on the context.

[0033] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0034] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0035] The embodiments of the present invention can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0036] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0037] It should be understood that the order of execution of the steps in the following embodiments does not imply a precedence of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0038] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0039] The first embodiment of the present invention provides a method for predicting the potential response of industry load regulation, which can be applied in the following aspects: Figure 1 In the application environment, the client communicates with the server. The client includes but is not limited to PDA, desktop computer, laptop computer, ultra-mobile personal computer (UMPC), netbook, cloud terminal device, personal digital assistant (PDA) and other terminal devices. The server can be implemented by an independent server or a server cluster composed of multiple servers. The sensing device is used to obtain the user's electricity consumption data, including but not limited to smart meters and mutual inductors.

[0040] See also Figure 2 , is a flow chart of a method for predicting the potential response of industry load regulation provided by the first embodiment of the present invention. The method for predicting the potential response of industry load regulation can be applied to Figure 1 The server in the server, the computer device corresponding to the server connects to the corresponding database, rule base, etc. to obtain the corresponding data in the database. The above computer device can also be connected to the client to collect data and control instructions sent by the client user. Figure 2 As shown, the load regulation response potential prediction method of the industry can include the following steps:

[0041] Step S101 , obtaining daily electricity consumption data of a current user in a current industry, taking the latest historical electricity consumption data of the current user for a first set period of time as historical data, and clustering the historical data to obtain various clusters.

[0042] The electricity-consuming industries involve a wide range of industries, including but not limited to the textile industry, non-metallic mineral products industry, chemical raw materials and chemical products manufacturing industry, coal mining and washing industry, accommodation and catering industry, etc. Each industry can be further divided into multiple specific sub-industries.

[0043] This embodiment selects the household textile finished product manufacturing industry in the textile industry as the current industry to be analyzed, and selects a manufacturing enterprise in a certain place in the household textile finished product manufacturing industry as the current user to start the solution presentation.

[0044] Among them, regarding the selection of the historical electricity consumption data of the manufacturing enterprise, this embodiment preferably uses the daily electricity consumption data of the manufacturing enterprise in the most recent year as the historical data, that is, the first set time period is preferably one year, and the daily electricity consumption data is preferably obtained by the 96-point load power acquisition method commonly used in the power industry, and then the historical data is clustered and analyzed.

[0045] The electricity consumption characteristics of various industries have different periodicity and randomness, so clustering analysis of their load curves may help understand and predict the electricity consumption behavior of these industries. The goal of the clustering algorithm is to classify similar objects into the same group, and different objects into different groups. This embodiment preferably uses the K-medoids clustering algorithm to perform clustering analysis on the daily load data in the historical data of enterprises in the industry, construct its typical electricity consumption scenario, and mine the power load characteristics based on this.

[0046] K-medoids is a sample-based clustering algorithm similar to the classic K-means algorithm, but it differs in the way it selects the center point. The K-medoids clustering algorithm divides the data set into K clusters, each cluster represented by a representative sample (called a "medoid" or "center point"). Its goal is to minimize the total distance between the data point and the center point of the cluster to which it belongs, thereby forming a tight, low-cohesion cluster structure.

[0047] The core idea of ​​the K-medoids algorithm is to select the real sample points from the sample set as the center of the cluster (ie, medoid), rather than using the centroid (the average position of the points in the cluster) to represent the center like the K-means algorithm. This selection method has better stability when dealing with data with noise or outliers. Because the medoid is a real data point, it will not be extremely affected by outliers like the centroid.

[0048] The specific execution steps of the basic K-medoids algorithm are as follows:

[0049] A. Randomly select K data points as initial medoids.

[0050] B. According to the distance criterion, each data point is assigned to the cluster to which the medoid closest to it belongs.

[0051] C. For each cluster, calculate the total distance from all data points in the cluster to its center point (the current medoid), then try to replace the other data points in the cluster with the new medoid and recalculate the total distance. Select the point that can minimize the total distance as the new medoid.

[0052] D. Repeat steps B and C until the medoids no longer change or the predetermined number of iterations is reached. This means that the algorithm has converged and the medoid of each cluster is stable and unchanged.

[0053] E. The algorithm outputs K clusters, each cluster is represented by a medoid and contains the data points closest to the medoid.

[0054] Since the basic K-medoids algorithm randomly selects the initial cluster center, which may lead to inconsistent clustering results each time, in order to improve the stability of the clustering results, it is integrated with the K-means++ algorithm, and it is preferred to use the initial cluster center method in the K-means++ algorithm to determine the initial cluster center. The K-means++ algorithm selects the initial cluster centers in an intelligent way so that they are as far away from each other as possible. The specific implementation steps are as follows:

[0055] a. Randomly select a sample from the data set χ as the first center point.

[0056] b. Calculate the shortest distance between each sample and the current cluster center, represented by D(x); then calculate the probability P(x) of each sample point being selected as the next cluster center, and finally select the sample point corresponding to the maximum probability value (or probability distribution) as the next cluster center.

[0057]

[0058] c. Repeat the previous step until K cluster centers are found.

[0059] In addition, the basic K-medoids algorithm uses Euclidean distance when calculating distance. Euclidean distance is the straight-line distance between two samples, which reflects the "shortest path" or "straight-line distance" between samples. It is suitable for situations where the distribution of data is relatively regular. However, Euclidean distance is more sensitive to outliers, because its calculation process involves square operations, which amplifies the impact of outliers on distance, which can cause outliers to significantly affect the clustering results. And in high-dimensional data, the Euclidean distance between data points tends to increase, which may cause the distance distribution between data points to become uniform, thereby reducing the clustering effect. In this regard, this embodiment preferably uses Manhattan distance (Cityblock Distance) to replace Euclidean distance.

[0060] Manhattan distance is another distance measurement method, which calculates the sum of the absolute values ​​of the coordinate differences between two points in each dimension. The calculation formula is as follows:

[0061]

[0062] Since Manhattan distance only considers the absolute value of coordinate differences, extreme values ​​have relatively little impact on it. Therefore, for power load data with abnormalities or mutations (such as peak power consumption on certain dates), Manhattan distance can better reflect the actual distribution of data. In addition, Manhattan distance still retains good discrimination in high-dimensional space and does not cause significant distortion in distance calculation, which performs better under the high-dimensional characteristics of daily data of corporate power load.

[0063] As for the number of clusters, commonly used methods for determining the number of clusters include the elbow method, the silhouette coefficient method, and the Gap statistic method.

[0064] The elbow method means that when the number of clusters increases, the within-cluster sum of squares (WCSS) of each group will decrease. However, when the number of clusters reaches a certain value, the rate of decrease of WCSS will slow down significantly. This value is the optimal number of clusters. The specific implementation steps are as follows:

[0065] 1. Calculate and plot the WCSS values ​​for different numbers of clusters.

[0066] 2. Observe the changing trend of WCSS from the graph and find the "elbow point", that is, the point where the rate of decline of WCSS slows down significantly.

[0067] 3. Select the number of clusters corresponding to the "elbow point" as the optimal number of clusters.

[0068] The silhouette coefficient method is a method for evaluating clustering effects. It can provide a measure to indicate the closeness and separation of data point clusters. The value of the silhouette coefficient ranges from -1 to 1, where a larger value generally indicates a better clustering effect. The average value of the silhouette coefficients of all data points can be calculated to obtain the silhouette coefficient of the entire clustering result. When determining the number of clusters, the silhouette coefficients under different numbers of clusters can be calculated, and the number of clusters that maximizes the silhouette coefficient can be selected. The specific implementation steps are as follows:

[0069] (1) Calculate the silhouette coefficient of each sample. The silhouette coefficient of each sample is equal to (ba) / max(a,b), where a is the average distance from the sample to other samples in the same cluster, and b is the average distance from the sample to other samples in the nearest cluster.

[0070] (2) Calculate the average of the silhouette coefficients of all samples as the silhouette coefficient of the clustering result.

[0071] (3) By changing the number of clusters, the silhouette coefficients under different numbers of clusters are calculated and compared, and the number of clusters that maximizes the silhouette coefficient is selected as the optimal number of clusters.

[0072] The Gap statistic method determines the optimal number of clusters by comparing the clustering results of real data with the clustering results of random data. The purpose of this method is to find a number of clusters so that the clustering structure of real data shows more obvious structural characteristics than random data. The specific implementation steps are as follows:

[0073] 1) Generate a set of random data for each possible number of clusters and apply the same clustering algorithm to the original data and each set of random data. The purpose of this step is to find the cluster center for each data set and calculate the quality of the clustering.

[0074] 2) Calculate the internal clustering criteria of the original data and random data for each number of clusters, usually using the logarithm of the total intra-class distance.

[0075] 3) For each number of clusters, calculate the logarithmic difference between the clustering results of the original data and the clustering results of the random data, i.e., the Gap statistic. The larger the Gap value, the more obvious the clustering structure.

[0076] 4) Select the number of clusters with the largest Gap statistic, or the first number of clusters that makes the Gap value no longer increase significantly as the optimal number of clusters.

[0077] In this embodiment, preferably, the elbow method, the silhouette coefficient method and the gap statistic method are respectively used to calculate the best number of clusters, and then the calculation results of the three methods are combined to comprehensively consider the final number of clusters.

[0078] Specifically, after respectively determining the first number of clusters, the second number of clusters, and the third number of clusters by the elbow method, the silhouette coefficient method, and the gap statistic method, the first number of clusters, the second number of clusters, and the third number of clusters are averaged and rounded down to obtain the final number of clusters when clustering the historical data.

[0079] After determining the clustering method and the number of clusters, Figure 3 As shown, for the manufacturing enterprise as the current user, the elbow rule, silhouette coefficient method and Gap statistic method are used to determine the final cluster number of the manufacturing enterprise, and then the improved K-medoids clustering algorithm is used to cluster the electricity consumption data of the most recent year as the historical data of the manufacturing enterprise, and 6 categories, namely 6 clusters, are obtained.

[0080] Step S102: determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the clusters.

[0081] The purpose of clustering the historical data of the current user, that is, the manufacturing enterprise, is to clarify the category of the current user's electricity usage habits. Figure 3 After clustering the clusters shown, Figure 4 As shown, the cluster center of each cluster is used as the typical electricity consumption scenario of the enterprise, and the upper and lower bounds of the electricity load corresponding to each typical electricity consumption scenario, that is, each cluster, are determined.

[0082] Combination Figure 3 as well as Figure 4 It can be seen that the daily load curve of the enterprise is basically the same in terms of fluctuation, and the load changes are very small. The main difference is reflected in the scale of electricity consumption, which mainly depends on the factor of season (temperature).

[0083] Electricity consumption in summer is significantly higher than in autumn, reaching a peak in August; in autumn it is higher than in winter, with the lowest point in January; from summer to winter, electricity consumption shows a gradually decreasing trend; electricity consumption fluctuates greatly in spring, and from April to June, it shows a gradually increasing trend.

[0084] In addition, the third and fifth categories of sets contain a small number of curves and present obviously abnormal power consumption, which can be ignored.

[0085] Therefore, the cluster centers of the remaining four sets can be used to represent the typical electricity consumption scenarios of the enterprise in a year, and given the following practical significance: the first type of electricity consumption scenario represents electricity consumption in winter, the second type of electricity consumption scenario represents electricity consumption when the enterprise is on vacation or under maintenance, the fourth type of electricity consumption scenario represents electricity consumption in spring and autumn, and the sixth type of electricity consumption scenario represents electricity consumption in the hot summer months.

[0086] Step S103 : taking the cluster center of the cluster to which the current user's daily electricity consumption data belongs as a specific cluster center, and predicting the current user's tomorrow's electricity consumption data based on the specific cluster center and the historical data.

[0087] From the analysis results of the above step S102, it can be seen that the typical electricity usage scenarios of the current user are diverse, and the typical electricity usage scenarios of the current user are different at different times. Therefore, in order to accurately complete the prediction of the current user's electricity usage data for tomorrow, it is first necessary to determine what the typical electricity usage scenario of the current user is currently in.

[0088] Specifically, this embodiment preferably determines the typical electricity usage scenario of the current user through the cluster to which the current user's daily electricity usage data belongs, and takes the cluster center of the cluster to which the current user's daily electricity usage data belongs, which represents the typical electricity usage scenario, as the specific cluster center; based on the determined specific cluster center, i.e., the typical electricity usage scenario and historical data, predicts the current user's electricity usage data for tomorrow.

[0089] Regarding the prediction method, this embodiment preferably uses the BiGRU-ATTENTION recurrent neural network to predict the power load.

[0090] About Bidirectional Gated Recurrent Unit Neural Network BiGRU:

[0091] GRU (Gated Recurrent Unit) is a type of recurrent neural network (RNN). Its biggest feature is that it contains two gates: update gate and reset gate. The update gate determines to what extent the new input information is combined with the previous memory. The reset gate determines to what extent the previous memory is combined with the current input. GRU has a relatively simple structure, fewer parameters, fast training speed, and can effectively capture long-term dependencies when processing sequence data.

[0092] BiGRU is a variant of GRU, which adds reverse sequence input to GRU. In simple terms, BiGRU contains two GRU layers, one for processing the forward sequence and the other for processing the reverse sequence. Combining the outputs of the two GRU layers can capture the information in the sequence data more comprehensively. In many tasks, especially those involving prediction tasks of time series data, BiGRU has better performance than the unidirectional GRU.

[0093] About the attention mechanism:

[0094] The attention mechanism imitates the way humans process visual information. It is an important technology in deep learning and is widely used in natural language processing, computer vision and other fields. The core idea of ​​the attention mechanism is to enable the model to focus on the most relevant information when processing input data, rather than processing all information equally. The most relevant information is given a larger weight ratio, and vice versa.

[0095] The basic principle of the attention mechanism can be described in the following steps:

[0096] Calculate the attention score S i For each input vector, calculate its similarity score with the query vector. Common scoring functions include additive attention, dot product attention, and scaled dot attention. Take additive attention as an example:

[0097] S(x i ,h j )=v T tanh(W[x i :h j ]+b)

[0098] In the formula, x i is the input vector; h j is the hidden state vector; W and v are the trainable weight matrix and vector; b is the bias term.

[0099] Calculate the attention weight α ij The attention score is converted into a probability distribution through the softmax function to obtain the attention weight α of each input vector ij .

[0100]

[0101] Weighted summation. By taking a weighted average of the input data, the final attention output vector C is calculated. i .

[0102]

[0103] Based on the BiGRU-ATTENTION recurrent neural network, the prediction of tomorrow's electricity consumption data of the current users, namely the above-mentioned manufacturing enterprises, is carried out. Specifically:

[0104] For the convenience of presentation and subsequent verification of the prediction accuracy, this embodiment selects a certain day that has passed as the current day, such as October 8, 2024 as the current day, and the electricity consumption data from October 7, 2023 to October 7, 2024 as the above-mentioned historical data, and then selects the current user's latest historical electricity consumption data of the second set time as the load output. This embodiment preferably sets the second set time to one week, then the electricity consumption data one week before October 8, 2024 is selected as the load output, and the cluster center of the cluster to which the load on October 8, 2024 belongs and the load output one week before that are used as feature variables of the BiGRU-ATTENTION recurrent neural network, and the historical data from October 7, 2023 to October 7, 2024 are used as the input of the BiGRU-ATTENTION recurrent neural network, and the BiGRU-ATTENTION recurrent neural network is trained to predict the load curve on October 8. The ratio of the training set to the test set is 3:1, the sliding window length is 8, and the prediction step is 1, that is, the data of the first 8 time points are used to predict the data of the next time point. Since this embodiment preferably obtains the daily power consumption data by the 96-point load power acquisition method, the data of the first 8 load power sampling points are used to predict the data of the next load power sampling point. Comparison of the load power prediction value and the actual value on October 8, 2024 Figure 5 As shown in the figure, the root mean square error of the load is 2.13%, the fitting degree is 95.72%, and the average absolute error is 1.94%. It can be seen that the predicted value is very close to the true value.

[0105] Step S104, determining the load regulation response of the current industry based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

[0106] Each cluster obtained by clustering the historical data of the current user actually contains a number of daily power consumption data. Based on the preferred method of obtaining daily power consumption data using 96 load power points in this embodiment, each cluster, i.e., a cluster set, is as follows: Figure 3 As shown, there will be a maximum value and a minimum value at each load sampling point, so the maximum value and the minimum value of each sampling point can form the following Figure 6 The upper and lower load bounds of the cluster are shown.

[0107] Then, the upper and lower bounds of the cluster corresponding to the specific cluster center can be determined, and based on the prediction of the current user's tomorrow's electricity consumption data, the load regulation response of the current user tomorrow can be determined. Specifically:

[0108] Based on the daily electricity consumption data obtained by the 96-point load power acquisition method, we first determine the upward adjustable difference between the load prediction value at each sampling point and the upper limit of the load at the sampling point, and then accumulate the upward adjustable differences at each sampling point to get the current user's load upward adjustment. Similarly, we determine the downward adjustable difference between the load prediction value at each sampling point and the lower limit of the load at the sampling point, and then accumulate the downward adjustable differences at each sampling point to get the current user's load downward adjustment.

[0109] In other embodiments, it is easy to understand that the 96-point load power acquisition method can be used instead of directly acquiring the entire day's power consumption. Then, correspondingly, the predicted power consumption value for tomorrow can be directly subtracted from the corresponding upper and lower bounds to determine the current user's load increase and load decrease.

[0110] In a preferred embodiment, taking into account that the electricity consumption data of a single user, i.e., a single manufacturing enterprise, is likely to be insufficiently representative, after obtaining the load increase and load decrease of the current user, the load increase and load decrease of a preset number of other users are determined in the same manner, and the average of the load increase of the current user and all the other users is used as the load increase of the current industry, and the average of the load decreases of the current user and all the other users is used as the load decrease of the current industry.

[0111] The value of the preset number can be determined by the operator based on a trade-off between prediction accuracy requirements and prediction speed requirements.

[0112] In another preferred embodiment, considering that some industries do not have obvious differences in power load with different time periods, seasons and environments, that is, they do not have load regulation response capabilities, in this preferred embodiment, after obtaining the historical data and before clustering the historical data, it is also judged whether the current user has load regulation response capabilities, so as to judge whether the industry where the current user is located has load regulation response possibilities:

[0113] The historical data is divided into interval data at set intervals. When the power consumption difference between any two adjacent interval data is less than the set difference threshold for more than a preset number of times, it is artificially determined that the current industry does not have load regulation response potential, and the load regulation response potential prediction for the current industry is terminated.

[0114] The setting interval may be half a day, a day, a week, a month, or a quarter. The setting difference threshold is set according to the size of the setting interval and the scale of the user's electricity consumption. The preset number of times is set by the operator according to the urgency of load regulation.

[0115] When the number of times that the power consumption difference between any two adjacent interval data is less than the set difference threshold exceeds the preset number of times, it means that the current user's power consumption remains stable over a larger time dimension, and its power consumption is not affected by time period, season and environment. There is no obvious ability to participate in demand-side load regulation, which means that to a considerable extent, the current industry in which the current user is located does not have obvious load regulation response potential.

[0116] Corresponding to the method of the above embodiment, Figure 7 The structural block diagram of the industry load regulation response potential prediction device provided by the second embodiment of the present invention is shown. The above prediction device is applied to a computer device, and the computer device is connected to the target database through a preset application program interface. When the target database is driven to execute a corresponding task, a corresponding task log is generated, and the above task log can be collected through the API. For the convenience of explanation, only the part related to the embodiment of the present invention is shown.

[0117] See also Figure 7 , the load regulation response potential prediction device of this industry includes:

[0118] The data acquisition clustering module 21 is used to acquire the daily electricity consumption data of the current user in the current industry, take the latest historical electricity consumption data of the current user for a first set time as historical data, and cluster the historical data to obtain various clusters;

[0119] A cluster parameter determination module 22, used for determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the cluster;

[0120] A user power consumption data prediction module 23, configured to use the cluster center of the cluster to which the current user's power consumption data for the day belongs as a specific cluster center, and predict the current user's power consumption data for tomorrow based on the specific cluster center and the historical data;

[0121] The industry load regulation potential determination module 24 is used to determine the load regulation response of the current industry based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

[0122] Optionally, the data acquisition clustering module 21 includes:

[0123] A cluster number determination unit is used to determine a first cluster number, a second cluster number and a third cluster number based on the historical data by using an elbow method, a silhouette coefficient method and a gap statistic method, respectively, and to determine a final cluster number when clustering the historical data by using the first cluster number, the second cluster number and the third cluster number.

[0124] Optionally, the data acquisition clustering module 21 further includes:

[0125] A clustering method determination unit is used to cluster the historical data using a K-medoids clustering algorithm, and during the clustering process, an initial clustering center of the K-medoids clustering algorithm is determined using an initial clustering center determination method of the K-means++ clustering algorithm, and the Manhattan distance between any two samples in the historical data is used as the difference between the any two samples.

[0126] Optionally, the user power consumption data prediction module 23 includes:

[0127] An electricity consumption data prediction unit is used to use the latest historical electricity consumption data of the current user for a second set time as the load output, the specific cluster center and the load output as feature variables of the BiGRU-ATTENTION recurrent neural network, and the historical data as the input of the BiGRU-ATTENTION recurrent neural network to train the BiGRU-ATTENTION recurrent neural network and output the tomorrow's electricity consumption data of the current user;

[0128] The second set time length is shorter than the first set time length.

[0129] Optionally, the user power consumption data prediction module 23 further includes:

[0130] The parameter limiting unit is used to limit the first set time length to one year and the second set time length to one week.

[0131] Optionally, the industry load regulation potential determination module 24 includes:

[0132] a load regulation response amount determination unit, for determining the load increase and load decrease of the current user based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center, and obtaining the load increase and load decrease of a preset number of other users in the current industry according to the method for obtaining the load increase and load decrease of the current user, taking the average of the load increase of the current user and all the other users as the load increase of the current industry, and taking the average of the load decreases of the current user and all the other users as the load decrease of the current industry.

[0133] Optionally, the above device further comprises:

[0134] The load regulation potential rough judgment module is used to divide the historical data into interval data at set intervals after acquiring the historical data and before clustering the historical data. When the number of times the power consumption difference between any two adjacent interval data is less than the set difference threshold exceeds a preset number, it is artificially determined that there is no load regulation response potential in the current industry, and the load regulation response potential prediction for the current industry is terminated.

[0135] It should be noted that the information interaction, execution process and other contents between the above modules are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0136] Figure 8 This is a schematic diagram of the structure of a computer device provided in Embodiment 4 of the present invention. Figure 8 As shown, the computer device of this embodiment includes: at least one processor ( Figure 8 Only one is shown), a memory, and a computer program stored in the memory and executable on at least one processor, wherein when the processor executes the computer program, the steps in any of the above-mentioned health prediction method embodiments are implemented.

[0137] The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that Figure 8 This is merely an example of a computer device and does not constitute a limitation on the computer device. The computer device may include more or fewer components than those shown in the figure, or a combination of certain components, or different components. For example, it may also include a network interface, a display screen, and an input device.

[0138] The processor may be a CPU, or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0139] The memory includes a readable storage medium, an internal memory, etc., wherein the internal memory may be the memory of a computer device, and the internal memory provides an environment for the operation of an operating system and computer-readable instructions in the readable storage medium. The readable storage medium may be a hard disk of a computer device, and in other embodiments may also be an external storage device of a computer device, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on a computer device. Further, the memory may also include both an internal storage unit of a computer device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as program codes of computer programs, etc. The memory may also be used to temporarily store data that has been output or is to be output.

[0140] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the above-mentioned method embodiment, which will not be repeated here. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiment can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying computer program code, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0141] The present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by a computer program product. When the computer program product runs on a computer device, the computer device can implement the steps in the above-mentioned method embodiment when executing.

[0142] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0143] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software 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 invention.

[0144] In the embodiments provided by the present invention, it should be understood that the disclosed devices / computer equipment and methods can be implemented in other ways. For example, the device / computer equipment embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0145] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0146] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for predicting the response potential of industry load regulation, characterized in that: The method comprises: Acquire daily electricity consumption data of current users in the current industry, take the latest historical electricity consumption data of the current users for a first set period of time as historical data, and cluster the historical data to obtain various clusters; Determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the clusters; Taking the cluster center of the cluster to which the current user's daily electricity consumption data belongs as a specific cluster center, and predicting the current user's tomorrow's electricity consumption data based on the specific cluster center and the historical data; The load regulation response amount of the current industry is determined based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

2. The method for predicting the industry load regulation response potential according to claim 1 is characterized in that: The clustering of the historical data to obtain various clusters includes: Based on the historical data, the elbow method, silhouette coefficient method and gap statistic method are respectively used to determine the first cluster number, the second cluster number and the third cluster number, and the first cluster number, the second cluster number and the third cluster number are used to determine the final cluster number when clustering the historical data.

3. The method for predicting the industry load regulation response potential according to claim 2 is characterized in that: The historical data are clustered using a K-medoids clustering algorithm, and during the clustering process, the initial clustering center of the K-medoids clustering algorithm is determined using the initial clustering center determination method of the K-means++ clustering algorithm, and the Manhattan distance between any two samples in the historical data is used as the difference between the any two samples.

4. The method for predicting the industry load regulation response potential according to claim 1, characterized in that: The predicting tomorrow's electricity consumption data of the current user based on the specific cluster center and the historical data includes: Taking the latest historical electricity consumption data of the current user for a second set time as the load output, taking the specific cluster center and the load output as the characteristic variables of the BiGRU-ATTENTION recurrent neural network, taking the historical data as the input of the BiGRU-ATTENTION recurrent neural network, training the BiGRU-ATTENTION recurrent neural network and outputting the tomorrow's electricity consumption data of the current user; The second set time length is shorter than the first set time length.

5. The method for predicting the industry load regulation response potential according to claim 4 is characterized in that: The first set time period is one year, and the second set time period is one week.

6. The method for predicting the industry load regulation response potential according to claim 1, characterized in that: Determining the load regulation response amount of the current industry includes: The load increase and load decrease of the current user are determined based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center. The load increase and load decrease of a preset number of other users in the current industry are obtained according to the method for obtaining the load increase and load decrease of the current user. The average of the load increase of the current user and all the other users is used as the load increase of the current industry, and the average of the load decreases of the current user and all the other users is used as the load decrease of the current industry.

7. The method for predicting the industry load regulation response potential according to any one of claims 1 to 6, characterized in that: After acquiring the historical data and before clustering the historical data, the method further includes: The historical data is divided into interval data at set intervals. When the power consumption difference between any two adjacent interval data is less than the set difference threshold for more than a preset number of times, it is artificially determined that the current industry does not have load regulation response potential, and the load regulation response potential prediction for the current industry is terminated.

8. An industry load regulation response potential prediction device, characterized in that: The device comprises: A data acquisition clustering module is used to acquire the daily electricity consumption data of the current user in the current industry, take the latest historical electricity consumption data of the current user for a first set time as historical data, and cluster the historical data to obtain various clusters; A cluster parameter determination module, used for determining the upper and lower bounds of each cluster and the cluster center of each cluster based on the cluster; A user power consumption data prediction module, configured to use the cluster center of the cluster to which the current user's power consumption data for the day belongs as a specific cluster center, and predict the current user's power consumption data for tomorrow based on the specific cluster center and the historical data; The industry load regulation potential determination module is used to determine the load regulation response of the current industry based on the tomorrow's electricity consumption data and the upper and lower bounds of the cluster corresponding to the specific cluster center.

9. A computer device, characterized in that: The computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting the industry load regulation response potential as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for predicting the industry load regulation response potential as described in any one of claims 1 to 7 is implemented.

Images