High energy consumption industrial user load classification method and system, electronic equipment, medium

Abstract

The present invention provides a method and system for classifying loads of high-energy-consuming industrial users, electronic equipment, and media, which are applied to the field of intelligent power consumption technology. The method includes: considering the load variation characteristics, load utilization rate, and user's sensitivity to electricity prices, determining multi-dimensional load characteristic indicators based on the daily load data of each industrial user; extracting time characteristics of the daily load data of each industrial user based on a convolutional autoencoder, and determining the power consumption characteristics based on the extracted load time characteristics and multi-dimensional load characteristic indicators of each industrial user; inputting the power consumption characteristics into a self-organizing map neural network to determine the output node to which each input data is mapped; screening out the target output node based on the number of input data corresponding to each output node; and clustering the power consumption characteristics of each industrial user using the target output node as the initial clustering center. The present invention solves the problems of poor clustering effect, large computational complexity, and low processing efficiency in the industrial load clustering process.

Description

Technical Field

[0001] The present invention relates to the field of intelligent electricity consumption technology, and in particular to a method and system for classifying loads of high-energy-consuming industrial users, electronic equipment, and a medium. Background Art

[0002] With the increasing reliance on renewable energy, the problem of balancing power grid supply and demand is becoming increasingly serious. To address this issue, leveraging demand-side resources to enhance grid stability is crucial. High-energy-consuming industrial users, due to their remarkably stable and manageable electricity consumption, play a key role in demand response. Understanding and exploring their load characteristics is crucial for optimizing demand response capabilities, ensuring secure grid operation, and achieving lean management.

[0003] Currently, clustering methods are often used to group industrial users with similar consumption behaviors, thereby implementing targeted load management strategies. However, the high-dimensional and nonlinear nature of industrial power consumption data generated by existing advanced metering infrastructure increases the complexity and difficulty of clustering. Furthermore, the clustering process often ignores the temporal characteristics of load data, resulting in poor clustering results, high computational complexity, and low processing efficiency. Summary of the Invention

[0004] In order to overcome the problems of poor clustering effect, large computational complexity and low processing efficiency in the above industrial load clustering process, the present invention provides a method and system for classifying loads of high-energy-consuming industrial users.

[0005] In one aspect, the present invention provides a method for classifying loads of high-energy-consuming industrial users, the method comprising:

[0006] Taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices, the multi-dimensional load characteristic indicators of each industrial user are determined based on their daily load data.

[0007] The convolutional autoencoder is used to extract the time characteristics of the daily load data of each industrial user, and the power consumption characteristics of each industrial user are determined based on the extracted load time characteristics and multi-dimensional load characteristic indicators;

[0008] Input the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped; and screen the target output node based on the number of input data corresponding to each output node;

[0009] The target output node is used as an initial clustering center to cluster the power consumption characteristics of each industrial user to obtain a load classification result of each industrial user.

[0010] Optionally, determining the power consumption characteristics of each industrial user based on the extracted load time characteristics and multi-dimensional load characteristic indicators of each industrial user includes:

[0011] For each industrial user, the index weight of the load characteristic index of each industrial user in each dimension is determined based on the entropy weight method;

[0012] Based on the weighted fusion of the load characteristic indicators of each dimension and the corresponding indicator weights, the key load characteristics of the corresponding industrial users are obtained;

[0013] The power consumption characteristics of each industrial user are determined based on the extracted load time characteristics of each industrial user and the corresponding key load characteristics.

[0014] Optionally, determining the index weight of the load characteristic index of each industrial user in each dimension based on the entropy weight method includes:

[0015] Calculate the entropy of the load characteristic index of each dimension based on the daily load data of each industrial user;

[0016] Based on the entropy of the load characteristic index of each dimension, the index weight of the load characteristic index of each dimension is calculated using an exponential function.

[0017] Optionally, the entropy of the load characteristic index of each dimension is calculated as follows:

[0018] ;

[0019] in, is the entropy of the load characteristic index c, a represents the total number of load characteristic indicators, d represents the index of daily load data, is the value of the load characteristic index c corresponding to the d-th daily load data sample, and m0 represents the total number of daily load data;

[0020] The calculation formula for the indicator weight of the load characteristic indicator of each dimension is:

[0021] ;

[0022] in, is the index weight of load characteristic index c, Indicates the The entropy of a load characteristic index, e is the base of the natural logarithm.

[0023] Optionally, the multi-dimensional load characteristic indicators of each industrial user are determined based on the daily load data of each industrial user, taking into account the load variation characteristics, load utilization rate and user's sensitivity to electricity prices, including:

[0024] For each industrial user, the corresponding daily average load, daily total load, daily maximum load, daily minimum load, daily peak period average load and daily off-peak period average load are calculated based on the daily load data of the industrial user;

[0025] Determine the daily load rate based on the daily average load and the daily maximum load, and determine the daily load peak-to-valley difference rate based on the daily maximum load and the daily minimum load; determine the daily maximum load utilization time based on the daily total load and the daily maximum load; determine the peak period load rate based on the daily peak period average load and the daily maximum load, and determine the valley period load rate based on the daily valley period average load and the daily maximum load;

[0026] The daily load rate and the daily load peak-to-valley difference rate are used as load characteristic indicators of the dimension corresponding to the load change characteristics, the daily maximum load utilization duration is used as the load characteristic indicator of the dimension corresponding to the load utilization rate, and the peak period load rate and the valley period load rate are used as load characteristic indicators of the dimension corresponding to the user's electricity price sensitivity.

[0027] Optionally, the convolutional autoencoder includes an encoder and a decoder, and the training process of the convolutional autoencoder includes:

[0028] Obtain historical load data of each industrial user as training data;

[0029] Inputting the training data into the convolutional autoencoder, and performing convolution and pooling processing on the training data through the encoder to obtain a corresponding hidden layer feature vector;

[0030] Performing deconvolution and upsampling processing on the hidden layer feature vector by the decoder to reconstruct output data with the same dimension as the input data;

[0031] A loss function is calculated based on the output data and the corresponding input data, and model parameters of the encoder and the decoder are updated based on the loss function.

[0032] Optionally, the extracting time features of the daily load data of each industrial user based on the convolutional autoencoder includes:

[0033] The daily load data of each industrial user is input into the trained convolutional autoencoder, and the low-dimensional feature vector of the hidden layer processed by the encoder is extracted as the corresponding load time feature.

[0034] Optionally, the self-organizing map neural network includes input layer and output layer grid nodes, and inputting the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped includes:

[0035] The power consumption characteristics corresponding to the historical load data of each industrial user are used as training samples, and the self-organizing map neural network is trained using the training samples to obtain the weight of each output node in the output layer grid node;

[0036] Based on the weights of each output node in the output layer, the power consumption characteristics of each industrial user are mapped to determine the output node to which each input data is mapped;

[0037] The input data is the electricity consumption characteristics of industrial users, and the output nodes are used to characterize the regional classifications corresponding to the electricity consumption characteristics, so as to achieve a specific response to the input data.

[0038] Optionally, screening out target output nodes based on the amount of input data corresponding to each output node includes:

[0039] Determine the local density of each output node based on the amount of input data corresponding to each output node;

[0040] For each mapped output node, if an output node is determined to be the node with the largest local density in the neighborhood of the current output node, the distance between the current output node and the node with the farthest distance in the neighborhood of the current output node is taken as the attribute distance of the current output node;

[0041] If it is determined that an output node is not the node with the maximum local density in the neighborhood of the current output node, the minimum distance between the current output node and each node in the target node set is used as the attribute distance of the current output node;

[0042] Screening out a plurality of nodes having local density and attribute distance from the output layer grid nodes as the target output nodes;

[0043] The target node set is a node set having a higher local density than the current output node in the domain of the current output node.

[0044] Optionally, clustering the power consumption characteristics of each industrial user is performed with the target output node as the initial clustering center to obtain a load classification result of each industrial user, including:

[0045] Determine the target output nodes corresponding to different numbers of cluster centers, and use the weights corresponding to the target output nodes as the initial cluster centers under the corresponding number of cluster centers;

[0046] For each number of cluster centers, cluster the electricity consumption characteristics of each industrial user based on the corresponding initial cluster center and K-means clustering algorithm, and obtain the clustering results under each cluster center data;

[0047] Based on the clustering performance analysis of the clustering results under different numbers of cluster centers, the target number of cluster centers is determined;

[0048] The clustering results corresponding to the target number of cluster centers are used as the load classification results of each industrial user.

[0049] Optionally, before determining the multi-dimensional load characteristic index of each industrial user based on the daily load data of each industrial user, taking into account the load variation characteristics, load utilization rate, and user's sensitivity to electricity prices, the method further includes:

[0050] Pre-process the daily load data of each industrial user;

[0051] The preprocessing includes at least one of missing data filling, outlier detection and correction, and data normalization.

[0052] Optionally, after obtaining the load classification results of each industrial user, the following steps are further included:

[0053] Carry out load pattern analysis based on the load classification results of each industrial user to determine the load pattern of each industrial user;

[0054] Forecast the electricity demand of each industrial user based on their load patterns;

[0055] Based on the prediction results, load equipment of each industrial user is regulated.

[0056] Optionally, before determining the multi-dimensional load characteristic index of each industrial user based on the daily load data of each industrial user, the method further includes:

[0057] Collect real-time daily operation data of internal load equipment of each industrial user;

[0058] Collect real-time measurement data from internal measurement equipment of each industrial user during the day;

[0059] Collect daily control status data of internal switch equipment of each industrial user;

[0060] Based on the load feature extraction of the daily real-time operation data, the daily real-time measurement data and the daily control status data, the daily load data of each industrial user is obtained.

[0061] On the other hand, the present invention also provides a high-energy-consuming industrial user load classification system, the system comprising:

[0062] The load feature extraction module is used to determine the multi-dimensional load feature indicators of each industrial user based on their daily load data, taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices. The module also extracts temporal features from the daily load data of each industrial user using a convolutional autoencoder, and determines the power consumption characteristics of each industrial user based on the extracted load temporal features and multi-dimensional load feature indicators.

[0063] The node screening module is used to input the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped; and to screen the target output node based on the number of input data corresponding to each output node;

[0064] The clustering module is used to cluster the power consumption characteristics of each industrial user with the target output node as the initial clustering center to obtain the load classification result of each industrial user.

[0065] On the other hand, the present invention also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0066] The memory is used to store one or more programs;

[0067] When the one or more programs are executed by the at least one processor, any one of the above-mentioned methods for classifying loads of high-energy-consuming industrial users is implemented.

[0068] On the other hand, the present invention also provides a readable storage medium having an execution program stored thereon, which, when executed, implements any one of the above-mentioned methods for classifying loads of high-energy-consuming industrial users.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] The present invention provides a load classification method and system for high-energy-consuming industrial users. By considering the load change characteristics, load utilization rate and user's electricity price sensitivity, multi-dimensional load characteristic indicators are extracted, and the load characteristics are reflected from multiple dimensions such as load change, load utilization and user electricity price sensitivity. At the same time, a convolutional autoencoder model is used to extract time features to ensure the comprehensiveness and accuracy of the extracted load features, thereby realizing accurate mining of the comprehensive load characteristics of industrial users. The feature extraction process before clustering can greatly reduce the clustering complexity, reduce the computational complexity of the clustering process, and improve the clustering efficiency.

[0071] The present invention inputs the extracted power consumption characteristics of each industrial user into a self-organizing map neural network, mines the distribution of the input characteristics, and screens out the target output node based on the amount of input data corresponding to each output node. The power consumption characteristics of each industrial user are clustered with the target output node as the initial clustering center. The clustering process can be guided by mining the distribution of the input characteristics, thereby improving the clustering effect and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 This is a flow chart of a method for classifying loads of high-energy-consuming industrial users according to the present invention;

[0073] Figure 2 Schematic diagram of the structure of a convolutional autoencoder according to an example of the present invention;

[0074] Figure 3 The structure and mapping diagram of a self-organizing map neural network of an example of the present invention;

[0075] Figure 4 This is a second flow chart of a method for classifying loads of high-energy-consuming industrial users according to the present invention;

[0076] Figure 5 This is a schematic diagram of the architecture of a high-energy-consuming industrial user load classification system of the present invention;

[0077] Figure 6 Schematic diagram of the structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0078] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0079] Example 1:

[0080] The present invention provides a method for classifying loads of high-energy-consuming industrial users, as shown in the schematic diagram. Figure 1 As shown, including:

[0081] Step S110 , considering load variation characteristics, load utilization, and user sensitivity to electricity prices, determining multi-dimensional load characteristic indicators for each industrial user based on the daily load data of each industrial user;

[0082] Step S120 , extracting time features from the daily load data of each industrial user based on a convolutional autoencoder, and determining the power consumption characteristics of each industrial user based on the extracted load time features and multi-dimensional load characteristic indicators of each industrial user;

[0083] Step S130: Input the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped; and select the target output node based on the number of input data corresponding to each output node;

[0084] Step S140 , clustering the power consumption characteristics of each industrial user with the target output node as the initial clustering center to obtain a load classification result of each industrial user.

[0085] In this example embodiment, characteristic indicators that characterize load variation, load utilization, and user sensitivity to electricity prices are extracted from daily load data. Both the convolutional autoencoder and self-organizing map neural network are trained using historical load data from various industrial users. The extracted load temporal features can be low-dimensional feature vectors of the hidden layer of the convolutional autoencoder. The self-organizing map neural network is used to establish a mapping relationship between the network's input data and output nodes. The self-organizing map neural network includes an input layer and an output layer, with the output layer composed of grid nodes. Clustering can be performed using algorithms such as K-means, fuzzy C-means, and spectral clustering. This method extracts multi-dimensional load characteristic indicators by considering load variation, load utilization, and user sensitivity to electricity prices. Load characteristics are reflected from multiple dimensions, including load variation, load utilization, and user sensitivity to electricity prices. A convolutional autoencoder model is used to extract temporal features, ensuring the comprehensiveness and accuracy of the extracted load features, thereby accurately mining the comprehensive load characteristics of industrial users. The feature extraction process prior to clustering can significantly reduce clustering complexity, reduce the computational complexity of the clustering process, and improve clustering efficiency. By inputting the extracted electricity consumption characteristics of each industrial user into a self-organizing map neural network, the distribution of the input characteristics is mined, and target output nodes are selected based on the amount of input data corresponding to each output node. The target output nodes are then used as initial clustering centers for clustering the electricity consumption characteristics of each industrial user. This can guide the clustering process by mining the distribution of input characteristics, improving clustering effectiveness and efficiency. This invention uses a self-organizing map (SOM) network to determine the initial clustering centers of the K-means algorithm, overcoming the local optimality problem that can occur when the traditional K-means algorithm randomly selects initial clustering centers, thereby improving the stability and effectiveness of clustering.

[0086] During the load data collection process, data may be abnormal or missing due to measurement anomalies, communication interruptions, and human errors. Therefore, it is necessary to preprocess the raw load data. In one exemplary embodiment, before determining the multi-dimensional load characteristic indicators of each industrial user based on the historical load data of each industrial user in S110 by considering the load variation characteristics, load utilization, and user's sensitivity to electricity prices, the following steps are also included:

[0087] Pre-process the daily load data of each industrial user;

[0088] The preprocessing includes at least one of missing data filling, outlier detection and correction, and data normalization.

[0089] In this example embodiment, is the daily load data of industrial users, is the daily load data of the i-th industrial user, is the daily load data of the nth industrial user, n is the total number of industrial users, where ; Represents the Mth sampling point data of the i-th industrial user, where M is the total number of sampling points of the i-th industrial user. In order to maintain the smoothness of the load curve, the missing data can be filled by the mean value based on the same time, specifically: Assume is the m1th sampling point data of the i-th industrial user on date k. For missing data values, the filling value can be calculated by the following formula:

[0090] (1)

[0091] in, It is the date set with measured values ​​at the m1th sampling point, which can be a date set close to the date of the value to be filled. represents the measurement value of the m1th sampling point of the i-th industrial user on date k, N is The number of days in is the missing fill value of the m1-th sampling point of the i-th industrial user on date k. In particular, when the proportion of missing data in the daily load data exceeds 20%, the daily load data is considered invalid and can be deleted directly.

[0092] In addition to the case of missing data, load data may also be abnormal due to failure of the measuring equipment. In this case, the horizontal similarity detection method can be used to identify abnormal data. Specifically, assuming is the load curve of an industrial user on day k, is the load data to be tested, that is, the data of the m1th sampling point in the load curve of the kth day. First, calculate the load change rate of the sampling point m1 according to formula (2): , then specify the normal range of the load change rate of each sampling point, and finally determine whether the daily load change rate of each sampling point is abnormal. If there is an abnormal value, use the value of the adjacent sampling point to perform abnormal correction. For this industrial user, (3) can be used for abnormal correction.

[0093] (2)

[0094] (3)

[0095] in, is the load data of the m1+1th sampling point in the load curve of the kth day, is the load curve of the kth day The load data of the sampling points, is the load curve of the kth day The load data of the sampling points, 、 are the number of sampling points taken forward and backward respectively, and Represents the total number of sampling points taken forward and backward respectively. By taking points before and after the sampling point m1, the average change rate of the sampling point is calculated to eliminate the abnormality of the change rate. In this example, you can set .

[0096] Since the load power values ​​of different industrial users may vary greatly, in order to better quantify the similarity between data and improve the accuracy of subsequent multi-dimensional power feature acquisition, the load data can be normalized, such as minimum-maximum normalization. For a certain industrial user, specifically:

[0097] (4)

[0098] in, for The normalized result of The m1th sampling point data in the load curve of the kth day, is the total number of sampling points for the daily load data of industrial users, and max and min are the maximum and minimum values, respectively.

[0099] This example performs preprocessing on the sampled daily load data, including missing data filling, outlier detection and correction, and data normalization, to improve the accuracy of subsequent feature extraction.

[0100] In an example implementation, determining the power consumption characteristics of each industrial user based on the extracted load time characteristics and multi-dimensional load characteristic indicators of each industrial user includes:

[0101] For each industrial user, the index weight of the load characteristic index of each industrial user in each dimension is determined based on the entropy weight method;

[0102] Based on the weighted fusion of the load characteristic indicators of each dimension and the corresponding indicator weights, the key load characteristics of the corresponding industrial users are obtained;

[0103] The power consumption characteristics of each industrial user are determined based on the extracted load time characteristics of each industrial user and the corresponding key load characteristics.

[0104] In this example implementation, the load characteristic indicators of each dimension describe the electricity consumption characteristics of industrial users to varying degrees, and their importance and importance vary. Therefore, it is necessary to set the weight coefficient of the load characteristic indicator so that it can objectively describe the importance of each load characteristic indicator. This example uses the entropy weight method to determine the indicator weight of each load characteristic indicator, and obtains the key load characteristics by weighted fusion of each load characteristic indicator and the corresponding indicator weight. For example, if there are 5 load characteristic indicators, the corresponding key load characteristics after weight assignment can represent the industrial load characteristic set of industrial user electricity consumption characteristics. , is the index weight of the first load characteristic index, L1 is the first load characteristic index, for example, L1 can be the daily load rate. For example, the index weight of the load characteristic index of each industrial user in each dimension is determined based on the entropy weight method, including:

[0105] Calculate the entropy of the load characteristic index of each dimension based on the daily load data of each industrial user;

[0106] Based on the entropy of the load characteristic index of each dimension, the index weight of the load characteristic index of each dimension is calculated using an exponential function.

[0107] In this example implementation, the entropy of the load characteristic indicator can be calculated based on the definition of entropy. The calculation formula for the entropy of the load characteristic indicator of each dimension is:

[0108] ; (5)

[0109] in, is the entropy of the load characteristic index c, a represents the total number of load characteristic indicators, d represents the index of daily load data, is the value of the load characteristic index c corresponding to the d-th daily load data sample, and m0 represents the total number of daily load data;

[0110] The calculation formula for the indicator weight of the load characteristic indicator of each dimension is:

[0111] ; (6)

[0112] in, is the index weight of load characteristic index c, Indicates the The entropy of each load characteristic indicator, where e is the base of the natural logarithm. It should be noted that the use of an exponential function for weight calculation in this example can more scientifically measure the uncertainty and information content of each indicator's distribution, thereby better reflecting the multiplicative and cumulative properties of probability. Furthermore, the convexity of the exponential function helps capture the uniformity and uncertainty of data distribution, while addressing the zero-value problem. It can also fairly weigh the contributions of low-probability and high-probability events, thereby enabling the final key load signature to more accurately describe the load characteristics.

[0113] In some example embodiments, the consideration of load variation characteristics, load utilization, and user sensitivity to electricity prices in S110 to determine multi-dimensional load characteristic indicators for each industrial user based on daily load data of each industrial user includes:

[0114] For each industrial user, the corresponding daily average load, daily total load, daily maximum load, daily minimum load, daily peak period average load and daily off-peak period average load are calculated based on the daily load data of the industrial user;

[0115] Determine the daily load rate based on the daily average load and the daily maximum load, and determine the daily load peak-to-valley difference rate based on the daily maximum load and the daily minimum load; determine the daily maximum load utilization time based on the daily total load and the daily maximum load; determine the peak period load rate based on the daily peak period average load and the daily maximum load, and determine the valley period load rate based on the daily valley period average load and the daily maximum load;

[0116] The daily load rate and the daily load peak-to-valley difference rate are used as load characteristic indicators of the dimension corresponding to the load change characteristics, the daily maximum load utilization duration is used as the load characteristic indicator of the dimension corresponding to the load utilization rate, and the peak period load rate and the valley period load rate are used as load characteristic indicators of the dimension corresponding to the user's electricity price sensitivity.

[0117] In this example implementation, basic statistics of the data can be first calculated using daily load data, such as daily average load, daily total load, daily maximum load, daily minimum load, daily peak period average load, and daily valley period average load, where the peak period and valley period can adopt existing regulations or standards. Based on the above basic statistics, load characteristic indicators that can reflect load change characteristics, load utilization, and user's sensitivity to electricity prices are determined. For example, the first load characteristic indicator can be selected as the daily load rate L1, the second load characteristic indicator can be selected as the daily peak-valley difference rate L2, and the third load characteristic indicator can be selected as the daily maximum load utilization hours L3 to reflect the load change characteristics and time utilization efficiency of industrial users; the electricity prices of industrial users in different time periods vary significantly, and the fourth load characteristic indicator can be selected as the peak period load rate L4 and the fifth load characteristic indicator can be selected as the valley period load rate L5 to reflect the user's sensitivity to electricity prices. The calculation method of each load characteristic indicator is shown in Table 1, which shows the specific calculation methods of the above five load characteristic indicators.

[0118] Table 1 Calculation method for industrial user load characteristic indicators

[0119]

[0120] In Table 1, represents the load data of industrial users at time t, 、 、 、 、 These represent the daily average load, daily maximum load, daily minimum load, daily peak-hour average load, and daily off-hour average load of industrial users, respectively. This example uses load characteristic indicators to describe the electricity consumption characteristics of industrial users, thereby quantifying the characteristics and attributes of industrial users' electricity usage habits and facilitating subsequent data processing.

[0121] Traditional feature extraction methods, such as principal component analysis (PCA) and independent component analysis (ICA), have difficulty mining nonlinear relationships in data. Therefore, this example uses a convolutional autoencoder for feature extraction. Before feature extraction, the convolutional autoencoder must be trained. The convolutional autoencoder training process includes:

[0122] Obtain historical load data of each industrial user as training data;

[0123] Inputting the training data into the convolutional autoencoder, and performing convolution and pooling processing on the training data through the encoder to obtain a corresponding hidden layer feature vector;

[0124] Performing deconvolution and upsampling processing on the hidden layer feature vector by the decoder to reconstruct output data with the same dimension as the input data;

[0125] A loss function is calculated based on the output data and the corresponding input data, and model parameters of the encoder and the decoder are updated based on the loss function.

[0126] In this example implementation, a convolutional autoencoder (CAE) is an unsupervised learning algorithm consisting of an input layer, a hidden layer, and an output layer, and is a type of neural network. Specifically, a convolutional autoencoder includes an encoder and a decoder. The convolutional autoencoder replaces the fully connected layers in traditional autoencoders with convolutional layers and pooling layers. Its structure is as follows: Figure 2 As shown in Figure 2. Due to the weight sharing in the convolutional layer, the number of training parameters in CAE is less than that in the fully connected layer, which reduces overfitting, speeds up training, and is suitable for processing more complex high-dimensional data. For example, let the input data be is the industrial user load data set, is the load data of the i-th industrial user, , the dimension of the load data at each sampling point is , CAE includes encoder and decoder. During the encoding process, the initial input data First, the feature vector q of the original vector in the hidden layer is obtained through the action of the convolution layer and pooling layer in the encoder. In the decoding process, the q in the hidden layer is reconstructed through the deconvolution layer and upsampling pooling layer in the decoder to obtain the feature vector q of the input data. The output data has the same dimension. The encoding and decoding process is shown in Formula 7 and Formula 8:

[0127] (7)

[0128] (8)

[0129] Among them, formula (7) represents the encoding process of CAE, represents the one-dimensional encoder convolution kernel; Indicates bias in the encoding process; represents the convolution operator; q is the low-dimensional feature vector in the hidden layer after encoding, , is the mth element of q; Represents the activation function of the encoder, which can be ReLU function. Formula (8) represents the decoding process of CAE, represents the one-dimensional decoder convolution kernel; Indicates the deviation in the decoding process; Represents the reconstructed input data; The activation function of the decoder can be Sigmoid function. The loss function in the training process can be calculated by mean square error (MSE). CAE learns the optimal network parameters by minimizing the loss function, so that the reconstructed data As close as possible to the input data .

[0130] Based on the trained convolutional autoencoder, the step S120 extracts time features from the daily load data of each industrial user based on the convolutional autoencoder, including:

[0131] The daily load data of each industrial user is input into the trained convolutional autoencoder, and the low-dimensional feature vector of the hidden layer processed by the encoder is extracted as the corresponding load time feature.

[0132] In this example embodiment, the daily load data of each industrial user to be processed can be input into the trained convolutional autoencoder to extract the hidden layer low-dimensional feature vector, that is, to obtain the low-dimensional feature vector q of the original input, where the dimension of q is less than , that is, the dimension of the extracted load time feature is smaller than the input daily load data, thereby realizing the time feature extraction of industrial user load data and reducing the dimension of the input data. Afterwards, the low-dimensional feature vector H in the CAE hidden layer is extracted as the time feature of the industrial user load data and combined with the industrial user load feature set L to form a multidimensional feature set C, completing the construction of the multidimensional power consumption feature. , Represent the first through fifth load time characteristics, namely the daily load rate, daily peak-to-valley difference ratio, daily maximum load utilization hours, peak-hour load rate, and valley-hour load rate in Table 1 above. q1 represents the first load time characteristic. Convolutional autoencoders can overcome the subjectivity of manually assigned load characteristic indicators and the difficulty in representing complex temporal relationships in load curves. By considering the time series characteristics of load data and extracting the temporal characteristics of industrial user load data, the effectiveness of industrial load clustering can be improved.

[0133] Next, industrial user load clustering is performed using a selective kernel algorithm. Traditional K-means methods encounter difficulty determining the initial cluster centers when handling clustering tasks. In some example embodiments, inputting the power consumption characteristics of each industrial user into a self-organizing map neural network to determine the output nodes to which each input data is mapped in S130 includes:

[0134] The power consumption characteristics corresponding to the historical load data of each industrial user are used as training samples, and the self-organizing map neural network is trained using the training samples to obtain the weight of each output node in the output layer grid node;

[0135] Based on the weight of each output node in the output layer, the power consumption characteristics of each industrial user are mapped, and the output node to which each input data is mapped is determined.

[0136] In this example implementation, the input data is the electricity consumption characteristics of industrial users, and the output nodes are used to characterize the regional classification of the corresponding electricity consumption characteristics to achieve a specific response to the input data. The activation of the output nodes corresponding to the input data reflects the distribution of the input data in the feature space. The distribution of the input data in the feature space is mined by the self-organizing map neural network to guide the clustering process, which can improve the clustering effect. The self-organizing map neural network in this example includes input layer and output layer grid nodes. A SOM network (Self-Organizing Map, SOM) can be established first. The electricity consumption characteristics corresponding to the historical load data of each industrial user can be used to train the self-organizing map neural network to obtain the weights of each output layer node, that is, to obtain the mapping relationship between daily load data and output layer nodes. SOM is an important unsupervised learning method, which is often used in clustering, high-dimensional visualization, data compression, feature extraction, etc. SOM is essentially a two-layer neural network consisting of an input layer and an output layer. The relationship between the output layer nodes and the input data is established through a competitive learning process. Each node competes to become the closest match to a given input data point. The output layer nodes have the same dimension as the input data. Its network structure is as follows: Figure 3 As shown. The SOM training process may include:

[0137] 1) Establish a self-organizing map neural network

[0138] The SOM network is constructed using the historical multidimensional feature set of industrial users (i.e., the multidimensional historical electricity consumption characteristics) as input. Assume that the number of samples is , the sample dimension is f, the number of neurons in the input layer is set to m2, and the number of nodes in the output layer is set to ,For example, .

[0139] Randomly initialize the weights of the output layer neurons ,in , Represents the output node The weight of the f-th dimension.

[0140] 2) Traverse each node of the output layer and find the winning node

[0141] Read the initial feature set from the multidimensional feature set of the industrial user As input, the Euclidean distance between the input data and each output node is calculated, and the output node with the closest distance to the input data is selected as the winning node.

[0142] 3) Calculate the weight update amplitude of each output node

[0143] Assume that the weight update amplitude of the winning node is 1. Based on the Gaussian kernel function, the update amplitude of each node is calculated according to the distance between each output node and the winning node. The update amplitude calculation formula is as follows:

[0144] (9)

[0145] in, Output node The update range, , , Indicates the output node closest to the current input data. is the distance parameter, which indicates the sensitivity of the update amplitude to the distance between nodes. The increase, gradually getting smaller, The update formula is as follows:

[0146] (10)

[0147] in, is the initial distance parameter, which can be taken ; maxstep is the maximum number of iterations, , T is the total time of the time level.

[0148] 4) Update the weight of each output node

[0149] After obtaining the weight update amplitude of each output node, update the weights of the winning node and its neighboring nodes. The node weight update formula is as follows:

[0150] (11)

[0151] in, Represents the output node In the +1 iteration weight, Represents the output node In the The weight at the iteration, Indicates the The learning rate of the iteration is The update formula is as follows:

[0152] (12)

[0153] in, is the initial learning rate, which can be taken .

[0154] Repeat steps 2) to 4) above until the number of iterations reaches maxstep, and the training ends. After training is completed, the weight matrix of each output node in the output layer is obtained. Based on the weights of each output node in the output layer, the power consumption characteristics of each industrial user are mapped, and the output node to which each input data is mapped is determined.

[0155] In some example implementations, the step of selecting a target output node based on the amount of input data corresponding to each output node in step S130 includes:

[0156] Determine the local density of each output node based on the amount of input data corresponding to each output node;

[0157] For each mapped output node, if an output node is determined to be the node with the largest local density in the neighborhood of the current output node, the distance between the current output node and the node with the farthest distance in the neighborhood of the current output node is taken as the attribute distance of the current output node;

[0158] If it is determined that an output node is not the node with the maximum local density in the neighborhood of the current output node, the minimum distance between the current output node and each node in the target node set is used as the attribute distance of the current output node;

[0159] Screening out a plurality of nodes having local density and attribute distance from the output layer grid nodes as the target output nodes;

[0160] The target node set is a node set having a higher local density than the current output node in the domain of the current output node.

[0161] In this example implementation, the mapping relationship between input and output nodes is obtained according to the weight matrix of the output layer nodes, and the number of inputs corresponding to each output node is counted, and the number of input data is used as the local density of the output layer nodes. To facilitate calculation, the different columns of the weight matrix can be concatenated and converted into column vectors. For example, the matrix Convert to column vector For each node in the transformed column vector, if the current node is not the node with the highest local density, its attribute distance is defined as the minimum distance between other nodes in the field with a higher local density than the current node. The distance between two nodes and the node attribute distance are calculated as follows:

[0162] (13)

[0163] (14)

[0164] in, is the distance between node r and node s, is the node r's The weight of the dimension, is the node s's The weight of the dimension, is the attribute distance of node r, represents the local density of node r, represents the local density of node s.

[0165] If the current node is the node with the largest local density, the distance attribute It is defined as the Euclidean distance between the current node and the farthest node in the neighborhood, and its calculation formula is:

[0166] (15)

[0167] in, Indicates the domain scope of node r.

[0168] According to the above process, the local density and attribute distance of each output node in the column vector are calculated, and then K output nodes with local density and attribute distance are randomly selected as target output nodes, where K is the number of clusters.

[0169] This example is based on two basic principles of the Density Peak Clustering (DPC) algorithm: cluster centers typically have a large local density, meaning they are surrounded by a large number of sample points; and cluster centers are often far apart. Based on these two principles, the local density and attribute distance of each output node are calculated. Target output nodes are then selected based on the local density and attribute distance of each node in the output layer. This guides the determination of the initial cluster center and improves the clustering performance of industrial user loads.

[0170] Traditional clustering methods encounter the problem of difficulty in determining the number of clusters when processing clustering tasks, and improper selection of the number of clusters will have an adverse effect on the clustering effect. Therefore, in some embodiments, S140 clusters the power consumption characteristics of each industrial user with the target output node as the initial cluster center to obtain the load classification results of each industrial user, including:

[0171] Determine the target output nodes corresponding to different numbers of cluster centers, and use the weights corresponding to the target output nodes as the initial cluster centers under the corresponding number of cluster centers;

[0172] For each number of cluster centers, cluster the electricity consumption characteristics of each industrial user based on the corresponding initial cluster center and K-means clustering algorithm, and obtain the clustering results under each cluster center data;

[0173] Based on the clustering performance analysis of the clustering results under different numbers of cluster centers, the target number of cluster centers is determined;

[0174] The clustering results corresponding to the target number of cluster centers are used as the load classification results of each industrial user.

[0175] In this example implementation, several different numbers of cluster centers can be pre-configured based on actual conditions. For example, the number of candidate cluster centers can be set to 6, 8, or 9. Then, for each number of cluster centers, a corresponding number of target output nodes is selected, and the weights corresponding to the target output nodes are used as the initial cluster centers. The K-means clustering algorithm is used to cluster the electricity consumption characteristics of each industrial user, and the clustering results for each number of cluster centers are obtained. By analyzing the clustering performance of the clustering results under each number of cluster centers, the target number of cluster centers with the best clustering performance is selected, and the clustering results corresponding to the target number of cluster centers are used as the load classification results for each industrial user. In this example, the clustering performance can be evaluated by observing the Calinski-Harabasz index (CHI), Davies-Bouldin index (DBI), and silhouette coefficient (SC) of the clustering results under different numbers of cluster centers. CHI is the ratio of inter-class dispersion to intra-class dispersion. The larger the CHI value, the greater the intra-class similarity and the smaller the inter-class similarity, that is, the better the clustering performance. The calculation formula for CHI is as follows:

[0176] (16)

[0177] in, is the Calinski-Harabasz index, and denote the between-class variance and within-class variance respectively; represents the number of clusters; Indicates the amount of data; between-class variance and intra-class variance The calculation formula is as follows:

[0178] (17)

[0179] (18)

[0180] in, Indicates the Cluster centers of the classes; Represents the center of all data samples; Indicates the samples With the Class membership, if Belong to Class, then Takes 1, otherwise takes 0.

[0181] DBI describes the compactness of data in the same cluster. The smaller the DBI value, the lower the dispersion and the better the clustering performance. The calculation formula of DBI is as follows:

[0182] (19)

[0183] in, is the Davies-Bouldin index, and Respectively represent The average distance from the samples in class j to the cluster center; It is The distance between the jth cluster center and the

[0184] SC combines the two factors of cohesion and separation to describe the definition of each type of contour. The larger the SC value, the better the clustering performance. The calculation formula of SC is as follows:

[0185] (20)

[0186] in, is the silhouette coefficient, Represents the average distance between data point i1 and other points of the same type; Represents the minimum average distance between data point i1 and samples in other classes.

[0187] Clustering performance evaluation for load clustering requires high intra-cluster similarity and low inter-cluster similarity for each curve. Therefore, the target number of cluster centers (K) is selected based on a large CHI value, a small DBI value, and a large SC value. This example further ensures the final clustering effect by selecting the number of clusters.

[0188] In some example implementations, after obtaining the load classification results of each industrial user, the method further includes:

[0189] Carry out load pattern analysis based on the load classification results of each industrial user to determine the load pattern of each industrial user;

[0190] Forecast the electricity demand of each industrial user based on their load patterns;

[0191] Based on the prediction results, load equipment of each industrial user is regulated.

[0192] In this example implementation, the load pattern corresponding to the daily load data of each industrial user can be analyzed based on the load classification results of each industrial user. For example, several load patterns can be determined based on a large amount of historical load data of each industrial user. Each load pattern has a typical load curve within the corresponding load period. The load period can be set based on the production plan or actual production situation of the industrial user. Based on the typical load curve, the electricity demand of the corresponding industrial user in the future period can be predicted, and the power of the load equipment can be regulated based on the prediction results. Through in-depth mining and analysis of industrial user electricity consumption data, this example can accurately extract and cluster load characteristics, accurately predict electricity demand, regulate equipment power in real time, and achieve intelligent load management. This not only helps to optimize electricity consumption time and power, but also significantly improves the response speed and flexibility of the power grid in application scenarios such as day-ahead and intraday demand response and power auxiliary service markets. Through precise regulation, the efficient absorption of renewable energy is promoted, and the adaptability of the power grid to new energy sources such as distributed photovoltaics is greatly improved.

[0193] In some example embodiments, before determining the multi-dimensional load characteristic index of each industrial user based on the daily load data of each industrial user, the method further includes:

[0194] Collect real-time daily operation data of internal load equipment of each industrial user;

[0195] Collect real-time measurement data from internal measurement equipment of each industrial user during the day;

[0196] Collect daily control status data of internal switch equipment of each industrial user;

[0197] Based on the load feature extraction of the daily real-time operation data, the daily real-time measurement data and the daily control status data, the daily load data of each industrial user is obtained.

[0198] In this example implementation, daily load data from each industrial user must be collected before feature extraction can be performed. Comprehensive load data can be acquired through multi-layered data collection. First, intelligent load devices, such as enterprise-owned generators, distributed power inverters, and electronic control units for arc furnaces and ladle refining furnaces, can be collected through wired or wireless communication to perform real-time operating data. If the devices lack an open access agreement, the equipment manufacturer must collaborate with them for access and commissioning. By monitoring the operating status of these devices and extracting their load signature data, equipment usage and operating efficiency can be analyzed. Second, measurement equipment within each industrial user can include electrical measuring instruments (such as smart meters installed by power grid companies or power demand response implementation agencies) or system measurement and control modules. Real-time measurement data from these devices can be collected through wired or wireless communication. By continuously monitoring these device measurements, load characteristics, including voltage, current, and power factor, can be analyzed. This is particularly important when these devices participate in demand response strategies, allowing for real-time monitoring to determine whether the load on the corresponding equipment is meeting established requirements. Finally, control status data for non-intelligent devices such as load switches or circuit breakers can be collected through wired or wireless communication, enabling direct on / off control of these devices. Feature extraction can be performed on these devices to assess their load characteristics and provide optimization recommendations. This example uses multi-level data collection to obtain load data for various devices, enriching the data source and providing data support for subsequent feature extraction.

[0199] The following is a specific example to illustrate the specific steps of a high energy consumption industrial user load classification method of the present invention. Figure 4 As shown, the following process may be included:

[0200] (1) Input original load data.

[0201] (2) Check whether the original load data is missing or exceeds the threshold. If so, delete the data; otherwise, add the missing load data.

[0202] (3) Check whether there are abnormal data in the original load data. If so, correct the abnormal values ​​in the data. Otherwise, obtain the processed load data (i.e., the supplemented load data).

[0203] (4) Perform maximum-minimum normalization on the processed load data.

[0204] (5) Establish a load feature set based on the standardized load data.

[0205] (6) Based on the convolutional autoencoder, the low-dimensional feature vector of the load data after normalization in (4) is extracted to obtain the time characteristics of the load data.

[0206] (7) Based on the load feature set in (5) and the time feature in (6), a multi-dimensional power consumption feature set is constructed.

[0207] (8) Establish a SOM (self-organizing map) network and train the SOM network to obtain the mapping relationship between input data and output nodes.

[0208] (9) Input the multi-dimensional electricity consumption feature set in (7) into the trained SOM network and determine the output node corresponding to each input data.

[0209] (10) Calculate the local density and attribute distance of each output node based on the number of input data corresponding to each output node.

[0210] (11) Determine the initial cluster centers under different numbers of clusters based on local density and attribute distance.

[0211] (12) Based on the initial cluster center and K-means algorithm, the multi-dimensional electricity consumption feature set in (7) is clustered to obtain the clustering results under different cluster numbers.

[0212] (13) Evaluate the clustering performance of clustering results under different cluster numbers, determine the target cluster number and use its corresponding clustering results as the load classification results.

[0213] (14) Analyze the load pattern of industrial users and conduct electricity demand forecasting and equipment power regulation.

[0214] This method calculates five key load characteristic indicators and weights them using an improved entropy weighting method to form an industrial load feature set. Furthermore, a convolutional autoencoder (CAE) is used to extract the temporal characteristics of the load data and combine them with the load feature set to construct a multidimensional feature set. To overcome the local optimum problem that can occur when the traditional K-means algorithm randomly selects initial cluster centers, this method uses a self-organizing map (SOM) network to determine the initial cluster centers of the K-means algorithm, thereby improving clustering stability and effectiveness. Experimental verification demonstrates that the proposed method has better stability and clustering effectiveness in resolving local optimum issues.

[0215] In existing technologies, load feature extraction for high-energy-consuming industrial users primarily relies on traditional statistical methods and simple empirical models. These methods typically perform static analysis based on historical data and lack the ability to respond to and predict dynamic load changes in real time, making them unable to meet the practical needs of modern smart grids and demand response. Load characteristics for high-energy-consuming industrial users are complex and variable, influenced by multiple factors such as production processes, seasonal variations, and equipment status. Traditional methods struggle to capture the subtle characteristics and underlying patterns of load variations, making it impossible to accurately manage and optimize loads. Existing processing of high-dimensional load data suffers from high complexity, high computational effort, and low efficiency. For example, Chinese patent publication number CN103390117A, entitled "A Method for Extracting Dynamic Characteristics of Power Loads," discloses a method that first preprocesses three-phase transient current data reflecting the dynamic characteristics of power loads, selecting data from the cycle before the fault and two cycles after the onset of the fault as samples. Next, a wavelet basis is selected and its lifting format determined. This wavelet exhibits superior performance in terms of spectral leakage and resolution. After selecting the wavelet basis, the number of decomposition levels is determined, and the three-phase transient current data for each sample is decomposed at each level using the lifting format wavelet packet transform. Next, the original signal is reconstructed, and the feature vector is constructed by extracting the corresponding reconstruction coefficients and calculating the energy value. This method relies on the choice of wavelet basis. If the wavelet basis is not chosen properly, the accuracy of feature extraction may be affected. The wavelet transform and inverse transform processes are relatively complex and require certain computing resources. In addition, this method requires data from one cycle before the fault and two cycles after the fault, which places high demands on data acquisition and storage, increasing the complexity of data processing. For example, the Chinese patent publication number CN116451052A, entitled "An Industrial Electricity Load Feature Extraction System," discloses a system consisting of four main modules: an external analysis module, a confidence configuration module, a caliber grouping module, and a feature extraction module. The external analysis module is responsible for connecting to different external data platforms, scanning and generating sensitive data indexes in real time to obtain industry confidence information. This module ensures the reliability and accuracy of the data through confidence analysis strategies and confidence analysis models. Each industry confidence information includes several industry segments and their corresponding benchmark confidence values. The confidence configuration module uses a dynamic configuration strategy to adjust the benchmark confidence value to generate a dynamic confidence value based on the electricity address and historical electricity usage information of each electricity user. This module selects the industry segment with the highest dynamic confidence value for each user as the industry segmentation feature of the user, thereby improving the precision and accuracy of the analysis. The caliber grouping module uses a caliber grouping strategy to generate a scale grouping benchmark based on feature retrieval requirements, and divides electricity users with the same industry segmentation characteristics into different scale groups. This grouping strategy can improve the efficiency and accuracy of data analysis and ensure that the extracted electricity load characteristics are more in line with actual needs.The feature extraction module is equipped with multiple feature extraction algorithms, which can be called upon based on specific requirements to extract corresponding load characteristics from the target electricity consumption data of different scale groups. This technology is highly dependent on external data platforms and the accuracy of internal electricity consumption data, and has high computational complexity and resource requirements.

[0216] The present invention addresses the problems existing in existing methods for extracting load characteristics of high-energy-consuming industrial users, such as reliance on traditional statistical methods, lack of intelligent analysis means, and inability to accurately reflect load change patterns and response characteristics. A method for extracting load characteristics of high-energy-consuming industrial users based on data and model driving is proposed. By introducing big data analysis and machine learning models, the method aims to accurately extract the load characteristics of high-energy-consuming industrial users and improve the scientificity and effectiveness of load management and optimization. With the rapid development of big data technology and machine learning algorithms, load characteristic extraction methods based on data driving and model driving have gradually become a research hotspot. By introducing big data analysis and machine learning models to conduct in-depth mining and analysis of massive electricity consumption data, the present invention can extract the time series characteristics and nonlinear characteristics of the load of high-energy-consuming industrial users, establish an accurate load forecasting model, and provide real-time response capabilities to load changes; it breaks through the limitations of traditional methods and improves the accuracy and efficiency of load characteristic extraction.

[0217] This method calculates key load characteristic indicators and weights them using an improved entropy weighting method to form an industrial load feature set. A convolutional autoencoder extracts the temporal characteristics of the load data and combines them with the load feature set to construct a multidimensional feature set. A self-organizing map network is used to determine the initial cluster centers of the K-means algorithm, overcoming the local optimality problem that can occur when randomly selecting initial cluster centers in the traditional K-means algorithm, thereby improving the stability and effectiveness of clustering.

[0218] The industrial user load classification method proposed in this invention, which is based on multi-dimensional electricity consumption characteristics, has a significant positive impact on the power system. By accurately mining and classifying the load characteristics of industrial users, this method improves the stability and operating efficiency of the power grid, helps to reduce the imbalance between supply and demand, optimize power dispatch and improve load management. It supports the construction of smart grids, provides data-driven decision support, and enhances the flexibility of demand-side response. The method also improves the security of the power grid, responds to emergencies in a timely manner through anomaly detection and early warning, and promotes the process of electricity marketization, supports market pricing mechanisms and user interaction, promotes the overall optimization and intelligent development of the power system, and improves the safety and stability of the power system.

[0219] This invention enables refined management of equipment within high-energy-consuming industrial users. By analyzing equipment operating data in real time, this method helps users identify energy consumption peaks and optimize equipment operation strategies, significantly improving energy efficiency and reducing energy costs. For market-oriented users, participating in the electricity demand response and ancillary services markets can generate corresponding market incentives through accurate load signature extraction, further reducing production costs and enhancing corporate competitiveness.

[0220] The present invention can provide guarantees for environmental protection and sustainable development for industrial users. By extracting the characteristics of high-energy-consuming industrial users to optimize the allocation of power resources, promote load transfer and peak-shaving and valley-filling, this method improves energy utilization efficiency and reduces carbon emissions and pollution. It supports the access of renewable energy, promotes the development of the green power market, enhances the flexibility and reliability of the power grid, and helps achieve sustainable development goals. By using data provided by advanced metering infrastructure, this method promotes smart grids and digital transformation, optimizes industrial electricity consumption behavior, and enhances users' environmental awareness and social responsibility, thereby promoting the use of green energy and environmental protection as a whole.

[0221] This invention can enhance the market competitiveness of industrial users: by implementing a data- and model-based load signature extraction method, industrial users can not only achieve cost control and improve energy efficiency, but also gain additional benefits through participation in the electricity market. This dual benefit enables enterprises to gain a competitive advantage in the market and promote industrial upgrading and technological progress.

[0222] This method accurately extracts and characterizes the load characteristics of high-energy-consuming industrial users, characterizing their operational characteristics and enabling refined load modeling. This supports the precise assessment and refined management of various industrial load resources, helping to improve the quality and efficiency of on-site inspections of high-energy-consuming industrial users. It also provides scientific, rational, efficient, and precise theoretical support for power load management tasks such as power demand response and orderly power consumption. Furthermore, the application of this method can optimize power consumption plans for high-energy-consuming industrial users, reduce electricity costs, and improve grid stability and reliability. This method can also promote the integration of renewable energy and support green and low-carbon development. This method has broad application prospects in load management, smart grids, demand response, and energy big data. The high-energy-consuming industrial load characteristic analysis system will support the development of new power load management systems, enabling precise load control and regularized, refined energy management for users to ensure secure power supply. It will also facilitate the construction of an adjustable load resource library containing a certain percentage (e.g., 5%) of maximum load, thereby delaying grid construction investment.

[0223] Example 2

[0224] Based on the same inventive concept, the present invention also provides a high-energy-consuming industrial user load classification system, the system comprising:

[0225] The load feature extraction module is used to determine the multi-dimensional load feature indicators of each industrial user based on their daily load data, taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices. The module also extracts temporal features from the daily load data of each industrial user using a convolutional autoencoder, and determines the power consumption characteristics of each industrial user based on the extracted load temporal features and multi-dimensional load feature indicators.

[0226] The node screening module is used to input the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped; and to screen the target output node based on the number of input data corresponding to each output node;

[0227] The clustering module is used to cluster the power consumption characteristics of each industrial user with the target output node as the initial clustering center to obtain the load classification result of each industrial user.

[0228] In one possible implementation, the load feature extraction module includes:

[0229] The weight calculation submodule is used to determine the index weight of the load characteristic index of each industrial user in each dimension based on the entropy weight method for each industrial user;

[0230] The fusion submodule is used to obtain the key load characteristics of the corresponding industrial users based on the weighted fusion of the load characteristic indicators of each dimension and the corresponding indicator weights;

[0231] The feature determination submodule is used to determine the power consumption feature of each industrial user based on the extracted load time feature and the corresponding key load feature of each industrial user.

[0232] In a possible implementation, the weight calculation submodule is specifically configured to:

[0233] Calculate the entropy of the load characteristic index of each dimension based on the daily load data of each industrial user;

[0234] Based on the entropy of the load characteristic index of each dimension, the index weight of the load characteristic index of each dimension is calculated using an exponential function.

[0235] In a possible implementation, the entropy of the load characteristic index of each dimension is calculated as follows:

[0236] ;

[0237] in, is the entropy of the load characteristic index c, a represents the total number of load characteristic indicators, d represents the index of daily load data, and m0 represents the total number of daily load data;

[0238] The calculation formula for the indicator weight of the load characteristic indicator of each dimension is:

[0239] ;

[0240] in, is the index weight of load characteristic index c, Indicates the The entropy of a load characteristic index, e is the base of the natural logarithm.

[0241] In a possible implementation, the load feature extraction module includes an index calculation submodule, which is configured to:

[0242] For each industrial user, the corresponding daily average load, daily total load, daily maximum load, daily minimum load, daily peak period average load and daily off-peak period average load are calculated based on the daily load data of the industrial user;

[0243] Determine the daily load rate based on the daily average load and the daily maximum load, and determine the daily load peak-to-valley difference rate based on the daily maximum load and the daily minimum load; determine the daily maximum load utilization time based on the daily total load and the daily maximum load; determine the peak period load rate based on the daily peak period average load and the daily maximum load, and determine the valley period load rate based on the daily valley period average load and the daily maximum load;

[0244] The daily load rate and the daily load peak-to-valley difference rate are used as load characteristic indicators of the dimension corresponding to the load change characteristics, the daily maximum load utilization duration is used as the load characteristic indicator of the dimension corresponding to the load utilization rate, and the peak period load rate and the valley period load rate are used as load characteristic indicators of the dimension corresponding to the user's electricity price sensitivity.

[0245] In one possible implementation, the convolutional autoencoder includes an encoder and a decoder, and the system further includes an encoder training module, where the encoder training module is configured to:

[0246] Obtain historical load data of each industrial user as training data;

[0247] Inputting the training data into the convolutional autoencoder, and performing convolution and pooling processing on the training data through the encoder to obtain a corresponding hidden layer feature vector;

[0248] Performing deconvolution and upsampling processing on the hidden layer feature vector by the decoder to reconstruct output data with the same dimension as the input data;

[0249] A loss function is calculated based on the output data and the corresponding input data, and model parameters of the encoder and the decoder are updated based on the loss function.

[0250] In a possible implementation, the load feature extraction module includes a time feature extraction submodule, which is configured to:

[0251] The daily load data of each industrial user is input into the trained convolutional autoencoder, and the low-dimensional feature vector of the hidden layer processed by the encoder is extracted as the corresponding load time feature.

[0252] In a possible implementation, the self-organizing map neural network includes input layer and output layer grid nodes, and the node screening module includes a mapping submodule, which is used to:

[0253] The power consumption characteristics corresponding to the historical load data of each industrial user are used as training samples, and the self-organizing map neural network is trained using the training samples to obtain the weight of each output node in the output layer grid node;

[0254] Based on the weights of each output node in the output layer, the power consumption characteristics of each industrial user are mapped to determine the output node to which each input data is mapped;

[0255] The input data is the electricity consumption characteristics of industrial users, and the output nodes are used to characterize the regional classifications corresponding to the electricity consumption characteristics, so as to achieve a specific response to the input data.

[0256] In a possible implementation, the node screening module includes a target node screening submodule, which is configured to:

[0257] Determine the local density of each output node based on the amount of input data corresponding to each output node;

[0258] For each mapped output node, if an output node is determined to be the node with the largest local density in the neighborhood of the current output node, the distance between the current output node and the node with the farthest distance in the neighborhood of the current output node is taken as the attribute distance of the current output node;

[0259] If it is determined that an output node is not the node with the maximum local density in the neighborhood of the current output node, the minimum distance between the current output node and each node in the target node set is used as the attribute distance of the current output node;

[0260] Screening out a plurality of nodes having local density and attribute distance from the output layer grid nodes as the target output nodes;

[0261] The target node set is a node set having a higher local density than the current output node in the domain of the current output node.

[0262] In a possible implementation, the clustering module includes:

[0263] The initial cluster center determination submodule is used to determine the target output nodes corresponding to different numbers of cluster centers, and use the weights corresponding to the target output nodes as the initial cluster centers under the corresponding number of cluster centers;

[0264] The clustering submodule is used to cluster the electricity consumption characteristics of each industrial user based on the corresponding initial cluster center and the K-means clustering algorithm for each cluster center number, and obtain the clustering results under each cluster center data;

[0265] The performance analysis submodule is used to determine the target number of cluster centers based on the clustering performance analysis of the clustering results under different numbers of cluster centers;

[0266] The classification submodule is used to use the clustering results corresponding to the target number of cluster centers as the load classification results of each industrial user.

[0267] In a possible implementation, the load feature extraction module further includes a preprocessing submodule, which is configured to:

[0268] Pre-process the daily load data of each industrial user;

[0269] The preprocessing includes at least one of missing data filling, outlier detection and correction, and data normalization.

[0270] In a possible implementation, the system further includes a prediction and control module, which is configured to:

[0271] Carry out load pattern analysis based on the load classification results of each industrial user to determine the load pattern of each industrial user;

[0272] Forecast the electricity demand of each industrial user based on their load patterns;

[0273] Based on the prediction results, load equipment of each industrial user is regulated.

[0274] In a possible implementation, the system further includes a data acquisition module, which is configured to:

[0275] Collect real-time daily operation data of internal load equipment of each industrial user;

[0276] Collect real-time measurement data from internal measurement equipment of each industrial user during the day;

[0277] Collect daily control status data of internal switch equipment of each industrial user;

[0278] Based on the load feature extraction of the daily real-time operation data, the daily real-time measurement data and the daily control status data, the daily load data of each industrial user is obtained.

[0279] The high-energy-consuming industrial user load classification system of the present invention extracts the load characteristics of high-energy-consuming industrial users through the above modules, and realizes the collection of electric quantity and non-electric quantity parameters through the integration of the intensive sensing platform and the existing platform or direct technical transformation. It combines the key parameters of the main production equipment of the enterprise to analyze the load characteristics of high-energy-consuming industrial users, supports the formulation of load participation in the power grid interaction strategy, and realizes the operation requirements of the power grid such as peak shaving and valley filling and new energy consumption.

[0280] The high energy consumption industrial user load classification system architecture of the present invention is as follows Figure 5 As shown, the system can be deployed in factories and enterprises within industrial parks, including power plants and distribution rooms, areas with good public wireless network coverage and low electromagnetic interference, as well as areas within the factory where operators perform routine duties. The system connects to the upper-level master station via a vertical security authentication platform, ensuring system security and efficient data transmission. For downlink connectivity, the system utilizes a hybrid wired and wireless networking approach to connect various devices, such as load edge terminals and load monitoring terminals, to obtain electrical, status, and environmental parameters. Through multi-layered data collection and analysis, the system extracts load signature data from various devices and conducts detailed analysis to optimize device utilization and capacity regulation. When executing demand response strategies, the system monitors device responses in real time to ensure compliance with the planned schedule. The entire system achieves efficient communication through a hybrid wired and wireless network, while overall security is ensured by the vertical security authentication platform. This deployment and connection approach for the load signature extraction system for high-energy-consuming industrial users not only enhances the intelligent management of the power system and improves energy efficiency, but also ensures system security and reliability.

[0281] Example 3

[0282] like Figure 6 As shown, the present invention also provides an electronic device, which may be a computer, a single-chip microcomputer, a smart mobile device, or the like. The electronic device in this embodiment may include a processor, a memory, a transceiver component, and the like. The memory, processor, and transceiver component are connected via a bus; the memory may be used to store an execution program, which may include instructions; and the processor may be used to execute the instructions stored in the memory. The memory may also be used to store data, which may be accessed and / or modified during the execution of the instructions.

[0283] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the storage medium to implement the corresponding method flow or corresponding function, so as to implement the steps of a high-energy-consuming industrial user load classification method in the above embodiment.

[0284] Example 4

[0285] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device-readable storage medium (Memory). This electronic device-readable storage medium is a memory device within the electronic device, used to store programs and data. It is understood that the storage medium herein may include both built-in storage media within the electronic device and, of course, extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for being loaded and executed by a processor. These instructions may be one or more executable programs (including program code). It should be noted that the storage medium herein may be high-speed RAM memory or non-volatile memory, such as at least one disk storage device. The processor loading and executing one or more instructions stored in the storage medium can implement the steps of a method for classifying high-energy-consuming industrial user loads in the above-described embodiment.

[0286] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0287] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0288] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0289] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0290] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that after reading the present invention, those skilled in the art may still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims.

Claims

1. A method for classifying loads of high-energy-consuming industrial users, characterized in that: include: Taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices, the multi-dimensional load characteristic indicators of each industrial user are determined based on their daily load data. Based on the daily load data of each industrial user, the entropy of the corresponding load characteristic index in each dimension is calculated; based on the entropy of the load characteristic index in each dimension, the index weight of the load characteristic index in each dimension is calculated using an exponential function; based on the weighted fusion of the load characteristic index in each dimension and the corresponding index weight, the key load characteristics of the corresponding industrial user are obtained; The time feature extraction of daily load data of each industrial user is performed based on the convolutional autoencoder. The extracted load time feature of each industrial user is combined with the corresponding key load feature as the power consumption feature of each industrial user. Inputting the power consumption characteristics of each industrial user into a self-organizing map neural network to determine the output node to which each input data is mapped; screening the target output node based on the local density and attribute distance of each output node; clustering the power consumption characteristics of each industrial user using the target output node as the initial cluster center to obtain the load classification results of each industrial user; The industrial users are high-energy-consuming industrial users; the multi-dimensional load characteristic indicators include the daily maximum load utilization duration used to reflect the time-rate efficiency of each industrial user, and the peak load rate and valley load rate used to reflect the sensitivity of each industrial user to electricity prices; the local density of each output node is determined based on the number of input data corresponding to the output node; the attribute distance of each output node is determined based on the node with the maximum local density in the neighborhood of the corresponding output node; The calculation formula of the entropy of the load characteristic index of each dimension is: ； in, is the entropy of the load characteristic index c, a represents the total number of load characteristic indicators, d represents the index of daily load data, For the The value of the load characteristic index c corresponding to the daily load data sample, m0 represents the total number of daily load data; The calculation formula for the indicator weight of the load characteristic indicator of each dimension is: ； in, is the index weight of load characteristic index c, Indicates the The entropy of the load characteristic index, is the base of natural logarithms.

2. The method according to claim 1, characterized in that The multi-dimensional load characteristic indicators of each industrial user are determined based on the daily load data of each industrial user, taking into account the load variation characteristics, load utilization rate and user's sensitivity to electricity prices, including: For each industrial user, the corresponding daily average load, daily total load, daily maximum load, daily minimum load, daily peak period average load and daily off-peak period average load are calculated based on the daily load data of the industrial user; Determine the daily load rate based on the daily average load and the daily maximum load, and determine the daily load peak-to-valley difference rate based on the daily maximum load and the daily minimum load; determine the daily maximum load utilization time based on the daily total load and the daily maximum load; determine the peak period load rate based on the daily peak period average load and the daily maximum load, and determine the valley period load rate based on the daily valley period average load and the daily maximum load; The daily load rate and the daily load peak-to-valley difference rate are used as load characteristic indicators of the dimension corresponding to the load change characteristics, the daily maximum load utilization duration is used as the load characteristic indicator of the dimension corresponding to the load utilization rate, and the peak period load rate and the valley period load rate are used as load characteristic indicators of the dimension corresponding to the user's electricity price sensitivity.

3. The method according to claim 1, characterized in that The convolutional autoencoder includes an encoder and a decoder, and the training process of the convolutional autoencoder includes: Obtain historical load data of each industrial user as training data; Inputting the training data into the convolutional autoencoder, and performing convolution and pooling processing on the training data through the encoder to obtain a corresponding hidden layer feature vector; Performing deconvolution and upsampling processing on the hidden layer feature vector by the decoder to reconstruct output data with the same dimension as the input data; A loss function is calculated based on the output data and the corresponding input data, and model parameters of the encoder and the decoder are updated based on the loss function.

4. The method according to claim 3, characterized in that The time feature extraction of the daily load data of each industrial user based on the convolutional autoencoder includes: The daily load data of each industrial user is input into the trained convolutional autoencoder, and the low-dimensional feature vector of the hidden layer processed by the encoder is extracted as the corresponding load time feature.

5. The method according to claim 1, wherein The self-organizing map neural network includes input layer and output layer grid nodes. Inputting the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped includes: The power consumption characteristics corresponding to the historical load data of each industrial user are used as training samples, and the self-organizing map neural network is trained using the training samples to obtain the weight of each output node in the output layer grid node; Based on the weights of each output node in the output layer, the power consumption characteristics of each industrial user are mapped to determine the output node to which each input data is mapped; The input data is the electricity consumption characteristics of industrial users, and the output nodes are used to characterize the regional classifications corresponding to the electricity consumption characteristics, so as to achieve a specific response to the input data.

6. The method according to claim 5, characterized in that The step of selecting a target output node based on the local density and attribute distance of each output node includes: For each mapped output node, if an output node is determined to be the node with the largest local density in the neighborhood of the current output node, the distance between the current output node and the node with the farthest distance in the neighborhood of the current output node is taken as the attribute distance of the current output node; If it is determined that an output node is not the node with the maximum local density in the neighborhood of the current output node, the minimum distance between the current output node and each node in the target node set is used as the attribute distance of the current output node; Screening out a plurality of nodes having local density and attribute distance from the output layer grid nodes as the target output nodes; The target node set is a node set having a higher local density than the current output node in the domain of the current output node.

7. The method according to claim 1, characterized in that Clustering the power consumption characteristics of each industrial user with the target output node as the initial clustering center to obtain the load classification results of each industrial user, including: Determine the target output nodes corresponding to different numbers of cluster centers, and use the weights corresponding to the target output nodes as the initial cluster centers under the corresponding number of cluster centers; For each number of cluster centers, cluster the electricity consumption characteristics of each industrial user based on the corresponding initial cluster center and K-means clustering algorithm, and obtain the clustering results under each cluster center data; Based on the clustering performance analysis of the clustering results under different numbers of cluster centers, the target number of cluster centers is determined; The clustering results corresponding to the target number of cluster centers are used as the load classification results of each industrial user.

8. The method according to any one of claims 1 to 7, characterized in that Before determining the multi-dimensional load characteristic indicators of each industrial user based on their daily load data, taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices, the following also needs to be included: Pre-process the daily load data of each industrial user; The preprocessing includes at least one of missing data filling, outlier detection and correction, and data normalization.

9. The method according to claim 1, characterized in that After obtaining the load classification results of each industrial user, it also includes: Carry out load pattern analysis based on the load classification results of each industrial user to determine the load pattern of each industrial user; Forecast the electricity demand of each industrial user based on their load patterns; Based on the prediction results, load equipment of each industrial user is regulated.

10. The method according to claim 1, characterized in that Before determining the multi-dimensional load characteristic indicators of each industrial user based on the daily load data of each industrial user, it also includes: Collect real-time daily operation data of internal load equipment of each industrial user; Collect real-time measurement data from internal measurement equipment of each industrial user during the day; Collect daily control status data of internal switch equipment of each industrial user; Based on the load feature extraction of the daily real-time operation data, the daily real-time measurement data and the daily control status data, the daily load data of each industrial user is obtained.

11. A load classification system for high energy consumption industrial users, characterized in that: include: The load feature extraction module is used to determine the multi-dimensional load feature indicators of each industrial user based on their daily load data, taking into account load variation characteristics, load utilization, and user sensitivity to electricity prices. The module also calculates the entropy of the corresponding load feature indicator for each dimension based on the daily load data of each industrial user. Based on the entropy of the load feature indicator for each dimension, an exponential function is used to calculate the indicator weight of the load feature indicator for each dimension. The module obtains the key load characteristics of the corresponding industrial user by weighted fusion of the load feature indicator for each dimension and the corresponding indicator weight. The module extracts the time characteristics of the daily load data of each industrial user based on a convolutional autoencoder, and combines the extracted load time characteristics of each industrial user with the corresponding key load characteristics as the power consumption characteristics of each industrial user. The node screening module is used to input the power consumption characteristics of each industrial user into the self-organizing map neural network to determine the output node to which each input data is mapped; the target output node is screened based on the local density and attribute distance of each output node; A clustering module, configured to cluster the power consumption characteristics of each industrial user using the target output node as an initial clustering center to obtain a load classification result for each industrial user; The industrial users are high-energy-consuming industrial users; the multi-dimensional load characteristic indicators include the daily maximum load utilization duration used to reflect the time-rate efficiency of each industrial user, and the peak load rate and valley load rate used to reflect the sensitivity of each industrial user to electricity prices; the local density of each output node is determined based on the number of input data corresponding to the output node; the attribute distance of each output node is determined based on the node with the maximum local density in the neighborhood of the corresponding output node; The calculation formula of the entropy of the load characteristic index of each dimension is: ； in, is the entropy of the load characteristic index c, a represents the total number of load characteristic indicators, d represents the index of daily load data, For the The value of the load characteristic index c corresponding to the daily load data sample, m0 represents the total number of daily load data; The calculation formula for the indicator weight of the load characteristic indicator of each dimension is: ； in, is the index weight of load characteristic index c, Indicates the The entropy of the load characteristic index, is the base of natural logarithms.

12. An electronic device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the method according to any one of claims 1 to 10 is implemented.

13. A readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, the method according to any one of claims 1 to 10 is implemented.

Images