A manufacturing enterprise cluster energy consumption situation prediction method based on heterogeneous energy consumption characteristic analysis

Abstract

The application discloses a manufacturing enterprise cluster energy consumption situation prediction method based on heterogeneous energy consumption characteristic analysis, which comprises the following steps: collecting the electricity consumption data of manufacturing enterprises, and constructing the electricity consumption data set E of the manufacturing enterprises i and the overall electricity consumption data set E of the manufacturing enterprise cluster; constructing the manufacturing enterprise cluster energy consumption situation prediction network F, and training E i , saving the final parameters θ' in the model fitting process, and realizing the analysis of the heterogeneous energy consumption characteristics of different manufacturing enterprises in the manufacturing enterprise cluster; training F by using E, migrating the final parameters θ' as the initial parameters of the training, saving the final parameters θ in the model fitting process, and inputting the real-time collected overall electricity consumption data of the manufacturing enterprise cluster into the trained F, so that the overall energy consumption situation prediction of the manufacturing enterprise cluster based on the heterogeneous energy consumption characteristic analysis is realized. The application is based on the historical energy consumption data of the manufacturing enterprise cluster, analyzes the heterogeneous energy consumption characteristics of the manufacturing enterprise cluster, and realizes the overall energy consumption situation prediction of the manufacturing enterprise cluster.

Description

Technical Field

[0001] This invention relates to the interdisciplinary fields of machine learning, industrial manufacturing, and demand-side management, and in particular to a method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis. Background Technology

[0002] Energy consumption forecasting refers to the pre-analysis of the total energy consumption and intensity of energy consumption in different time periods by energy supply-side entities such as power companies. It is the core of ensuring energy system stability and a key technology for coordinating energy supply and demand balance. A typical manufacturing cluster consists of multiple enterprises performing different types of manufacturing tasks, resulting in a large overall energy consumption and high intensity. Therefore, forecasting the overall energy consumption of this cluster can better promote peak shaving and valley filling and maintain energy system stability. However, due to the diverse product production and manufacturing modes of different enterprises within a manufacturing cluster, their energy consumption characteristics exhibit significant heterogeneity, leading to a complex periodic pattern in the overall energy consumption of the cluster. Therefore, how to analyze the heterogeneous energy consumption characteristics of multiple enterprises within a manufacturing cluster, dissect the complex energy consumption patterns of the cluster under its influence, and thus achieve accurate forecasting of the overall energy consumption of the manufacturing cluster is a critical problem that urgently needs to be solved.

[0003] Research on energy consumption forecasting has always been a hot topic in the industry, primarily used for advance energy planning and management on the energy supply side. Current forecasting methods in the manufacturing sector mainly apply to manufacturing processes, equipment, and workshops. They achieve energy consumption forecasting by mining data correlations between specific process, equipment, and workshop energy consumption characteristics and covariates such as process parameters and manufacturing tasks. However, traditional forecasting methods only predict the energy consumption of individual processes, equipment, and workshops, failing to provide the energy supply side with theoretical and data support for global energy consumption forecasting across manufacturing enterprise clusters.

[0004] Currently, comprehensive and systematic energy consumption forecasting methods mainly focus on areas related to people's livelihoods, such as analyzing the energy consumption patterns and trends of buildings, urban areas, or national regions. Existing methods also achieve trend forecasting by performing time-based modeling of energy consumption trends and mining data correlations between energy consumption and multiple covariates. However, these methods primarily consider the impact of seasonal and locational periodic spatiotemporal characteristics on overall energy consumption trends. The difficulty in forecasting energy consumption trends in manufacturing enterprise clusters lies mainly in the complexity of energy consumption patterns caused by the heterogeneity of manufacturing due to fluctuations in production orders and tasks. The periodic patterns that can be mined are sparse, making them unsuitable for forecasting energy consumption trends in manufacturing enterprise clusters. Furthermore, existing methods rely on covariate correlations to achieve trend forecasting. However, due to the complex types and structures of energy-consuming units within manufacturing enterprise clusters, it is difficult to screen and collect covariates while ensuring the timeliness and accuracy required for demand response. Summary of the Invention

[0005] This invention provides a method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis. Based on the historical energy consumption data of the manufacturing enterprise cluster, the method analyzes the heterogeneous energy consumption characteristics of the manufacturing enterprise cluster and realizes the prediction of the overall energy consumption situation of the manufacturing enterprise cluster.

[0006] This invention provides a method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis, including:

[0007] Collect electricity consumption data from manufacturing enterprises, and construct an electricity consumption dataset E for manufacturing enterprises based on the electricity consumption data. i The data set E represents the overall electricity consumption of the manufacturing enterprise cluster; wherein the manufacturing enterprise cluster includes multiple manufacturing enterprises.

[0008] Construct an energy consumption prediction network F for a cluster of manufacturing enterprises, and utilize the electricity consumption dataset E of the manufacturing enterprises. i Training is performed, and the network parameters are optimized by minimizing the loss function. The final parameters θ′ of the model fitting are saved to enable the analysis of heterogeneous energy consumption characteristics of different manufacturing enterprises within a manufacturing enterprise cluster.

[0009] The energy consumption data set E of the manufacturing enterprise cluster is used to train the energy consumption status prediction network of the manufacturing enterprise cluster. The final parameter θ′ is transferred as the initial parameter for training. The network parameters are optimized by minimizing the loss function, and the final parameter θ is saved when the model is fitted.

[0010] The real-time collected electricity consumption data of the manufacturing enterprise cluster is input into the trained energy consumption situation prediction network F of the manufacturing enterprise cluster to realize the prediction of the overall energy consumption situation of the manufacturing enterprise cluster based on the analysis of heterogeneous energy consumption characteristics.

[0011] Furthermore, the electricity consumption data of the manufacturing enterprises is collected, and an electricity consumption dataset E of the manufacturing enterprises is constructed based on the electricity consumption data. i In the steps of obtaining the overall electricity consumption dataset E of the manufacturing enterprise cluster,

[0012] The granularity of the electricity consumption data is 15 minutes, and the unit is kW·h;

[0013] The electricity consumption dataset E of the manufacturing enterprise i The overall electricity consumption dataset E for the manufacturing enterprise cluster is as follows:

[0014]

[0015] E = {x1, x2, ..., x} n}

[0016] in, Let x represent the electricity consumption data of the nth manufacturing enterprise at the nth time step. tn This represents the electricity consumption data at the nth time step of the manufacturing enterprise cluster.

[0017] Furthermore, the energy consumption prediction network for the manufacturing enterprise cluster includes a characteristic analysis module M. a and situation prediction module M p .

[0018] Furthermore, the characteristic analysis module M a Includes the following steps:

[0019] Energy consumption data frequency selection:

[0020]

[0021] Where x[n] represents the nth input time series energy consumption data, X[k] represents the kth transformed frequency information, and N represents the sample length of the input energy consumption data. Complex root of unity;

[0022] Two-dimensional reconstruction of energy consumption data:

[0023]

[0024] Where A[k] represents the amplitude corresponding to the k-th frequency, x X[k] f represents the electricity consumption data of the k-th input. avg (·) denotes the mean function, g amp (·) represents the amplitude acquisition function;

[0025]

[0026] Where Lx represents the length of the energy data input, and P[k] represents the k-th frequency;

[0027] Characterization and extraction of heterogeneous properties:

[0028]

[0029] in, Let x represent the j-th two-dimensional convolution operation. i This represents the input data segment x′. i This represents the output data after the convolution operation;

[0030] One-dimensional restructuring and importance weighting of energy consumption data:

[0031] S i =Softmax(A i )

[0032] Where Softmax(·) represents the Softmax function, A i S represents the i-th amplitude. i This represents the importance score corresponding to the i-th amplitude.

[0033]

[0034] in, and These represent the energy consumption data after one-dimensional recombination and the energy consumption data after weighting, respectively.

[0035]

[0036] In the formula, x f This indicates the final output.

[0037] Furthermore, the situation prediction module M p It includes a unidirectional LSTM layer and a linear classification layer, used for time modeling and prediction of energy consumption data that has undergone heterogeneous energy consumption characteristic analysis.

[0038] Furthermore, the energy consumption prediction network F for the manufacturing enterprise cluster is constructed, and the electricity consumption dataset E of the manufacturing enterprises is utilized. i The steps involved in training the network, optimizing the network parameters by minimizing the loss function, and saving the final parameters θ′ from the model fitting process, are as follows: This allows for the analysis of heterogeneous energy consumption characteristics among different manufacturing enterprises within a manufacturing cluster.

[0039] Set the number of pre-training rounds e and the batch size B of the input data for the energy consumption prediction network F of the manufacturing enterprise cluster;

[0040] For any batch of data in any training round, calculate the loss between the predicted value and the true value using the loss function L:

[0041]

[0042] in, x represents the predicted consumption value at time step ti. ti This represents the actual consumption value for time step ti;

[0043] After each round of training, the parameters θ′ of the convolutional neural network are optimized:

[0044]

[0045] Where α represents the learning rate, and (θ')' represents the parameters of the optimized convolutional neural network F';

[0046] Save the final parameters θ' of the energy consumption situation prediction network F for manufacturing enterprise clusters after e rounds of pre-training.

[0047] Further, the steps of training the energy consumption prediction network F of the manufacturing enterprise cluster using the overall electricity consumption dataset E of the manufacturing enterprise cluster, transferring the final parameter θ' as the initial parameter for training, optimizing the network parameters by minimizing the loss function, and saving the final parameter θ during model fitting specifically include:

[0048] Set the number of training rounds e and the batch size B of the input data for the energy consumption prediction network F of the manufacturing enterprise cluster;

[0049] For any batch of data in any training round, calculate the loss between the predicted value and the true value using the loss function L:

[0050]

[0051] After each round of training, the parameters θ of the energy consumption situation prediction network F for manufacturing enterprise clusters are optimized:

[0052]

[0053] Where (θ)' represents the parameters of the optimized energy consumption prediction network F for manufacturing enterprise clusters;

[0054] Save the final parameters θ of the energy consumption situation prediction network F for manufacturing enterprise clusters after e rounds of training, and the training ends.

[0055] This invention also provides a device for predicting the energy consumption situation of a manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis, comprising:

[0056] The data acquisition module is used to collect electricity consumption data from manufacturing enterprises and construct an electricity consumption dataset E for the manufacturing enterprises based on the electricity consumption data. i The data set E represents the overall electricity consumption of the manufacturing enterprise cluster; wherein the manufacturing enterprise cluster includes multiple manufacturing enterprises.

[0057] The module is used to construct an energy consumption prediction network F for a cluster of manufacturing enterprises, and utilizes the electricity consumption dataset E of the manufacturing enterprises. i Training is performed, and the network parameters are optimized by minimizing the loss function. The final parameters θ' of the model fitting are saved to enable the analysis of heterogeneous energy consumption characteristics of different manufacturing enterprises within a manufacturing enterprise cluster.

[0058] The training module is used to train the energy consumption status prediction network F of the manufacturing enterprise cluster using the overall electricity consumption dataset E of the manufacturing enterprise cluster, transfer the final parameter θ' as the initial parameter for training, optimize the network parameters by minimizing the loss function, and save the final parameter θ when the model is fitted.

[0059] The prediction module is used to input the real-time collected overall electricity consumption data of the manufacturing enterprise cluster into the trained manufacturing enterprise cluster energy consumption situation prediction network F, so as to realize the overall energy consumption situation prediction of the manufacturing enterprise cluster based on the analysis of heterogeneous energy consumption characteristics.

[0060] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0061] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0062] The beneficial effects of this invention are as follows:

[0063] 1. From the perspective of data science, this invention transforms the problem of predicting the energy consumption of manufacturing enterprises into an autoregressive problem that only requires historical energy consumption data, which greatly reduces the difficulty and demand for data collection.

[0064] 2. This invention utilizes the idea of ​​two-dimensional reconstruction of one-dimensional time series data and extracts cross-period variation characteristics of energy consumption time series data through convolution operations, thereby enabling the analysis of heterogeneous energy consumption characteristics of different enterprises within a cluster.

[0065] 3. This invention utilizes a transfer learning mechanism to assist a data-driven prediction network in constructing a cluster-wide energy consumption time-series correlation based on prior knowledge of energy consumption characteristics, thereby achieving accurate prediction of energy consumption trends. Attached Figure Description

[0066] Figure 1 This is a flowchart illustrating the energy consumption pattern prediction method for manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis, as described in this invention.

[0067] Figure 2 This is a schematic diagram of the energy consumption situation prediction device for manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis according to the present invention.

[0068] Figure 3 This is a schematic diagram of the internal structure of the computer device in this invention.

[0069] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0070] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0071] Predicting the energy consumption patterns of manufacturing enterprise clusters can be further understood as a time series autoregressive problem that relies on the relationship between time steps, i.e., using historical time steps T. n Total energy consumption forecast for future time steps T n+i The total energy consumption. Traditional time series forecasting methods construct deep neural networks, such as recurrent neural networks and convolutional neural networks, to explore the data dependencies between time steps. However, these methods cannot effectively analyze and predict sequences with time variations. Therefore, the main technical problem solved by this invention is to analyze the heterogeneous energy consumption characteristics of manufacturing enterprise clusters based on their historical energy consumption data, thereby achieving overall energy consumption trend prediction for the manufacturing enterprise clusters.

[0072] like Figure 1 As shown, this invention provides a method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis, including:

[0073] S1. Collect electricity consumption data from manufacturing enterprises and construct an electricity consumption dataset E for manufacturing enterprises based on the electricity consumption data. i The data set E represents the overall electricity consumption of the manufacturing enterprise cluster; wherein the manufacturing enterprise cluster includes multiple manufacturing enterprises.

[0074] By assembling electricity meters in each manufacturing enterprise within the manufacturing enterprise cluster, electricity consumption data with a granularity of 15 minutes and a unit of kilowatt-hours (kW·h) is collected. Therefore, the granularity of the electricity consumption data is 15 minutes and the unit is kW·h.

[0075] The electricity consumption dataset E of the single manufacturing enterprise i The overall electricity consumption dataset E for the manufacturing enterprise cluster is as follows:

[0076]

[0077] E = {x1, x2, ..., x} n}

[0078] in, Let x represent the electricity consumption data of the i-th manufacturing enterprise at the n-th time step. tn This represents the electricity consumption data at the nth time step of the manufacturing enterprise cluster.

[0079] S2. Construct a manufacturing enterprise cluster energy consumption prediction network F, and utilize the electricity consumption dataset E of the manufacturing enterprises. i Training is performed, and the network parameters are optimized by minimizing the loss function. The final parameters θ' of the model fitting are saved to enable the analysis of heterogeneous energy consumption characteristics of different manufacturing enterprises within a manufacturing enterprise cluster.

[0080] The energy consumption prediction network F for the manufacturing enterprise cluster includes a characteristic analysis module M. a and situation prediction module M p .

[0081] Step S2 includes:

[0082] S201. To analyze the heterogeneous energy consumption characteristics of different manufacturing enterprises within a manufacturing enterprise cluster, a manufacturing enterprise cluster energy consumption situation prediction network F is constructed, wherein the characteristic analysis module M... a This includes the following 5 operations:

[0083] 1. Selection of energy consumption data frequency:

[0084]

[0085] Where x[n] represents the nth input time series energy consumption data, X[k] represents the kth transformed frequency information, and N represents the sample length of the input energy consumption data. Complex root of unity;

[0086] 2. Two-dimensional reconstruction of energy consumption data:

[0087]

[0088] Where A[k] represents the amplitude corresponding to the k-th frequency, x X[k] f represents the electricity consumption data of the k-th input. avg (·) denotes the mean function, g amp (·) represents the amplitude acquisition function;

[0089]

[0090] Where Lx represents the length of the energy data input, and P[k] represents the k-th frequency;

[0091] 3. Characterization and extraction of heterogeneous properties:

[0092]

[0093] in, Let x represent the j-th two-dimensional convolution operation. i Let x' represent the energy consumption data of the i-th input segment. i This represents the output data after the convolution operation;

[0094] 4 and 5, One-dimensional restructuring and importance weighting of energy consumption data:

[0095] S i =Softmax(A i )

[0096] Where Softmax(·) represents the Softmax function, A i S represents the i-th amplitude. i This represents the importance score corresponding to the i-th amplitude.

[0097]

[0098] in, and These represent the energy consumption data after one-dimensional recombination and the energy consumption data after weighting, respectively.

[0099]

[0100] In the formula, x f This indicates the final output.

[0101] The situation prediction module M p It includes a unidirectional LSTM layer and a linear classification layer, used for time modeling and prediction of energy consumption data that has undergone heterogeneous energy consumption characteristic analysis.

[0102] S202, The energy consumption prediction network F for the manufacturing enterprise cluster is constructed, and the electricity consumption dataset E of the manufacturing enterprises is used. i Training involves setting the number of pre-training rounds e and the batch size B of the input data for the energy consumption prediction network F of the manufacturing enterprise cluster; for any batch of data in any training round, the loss function L is used to calculate the loss between the predicted value and the true value:

[0103]

[0104] in, x represents the predicted consumption value at time step ti. ti This represents the actual consumption value for time step ti;

[0105] S203. After each round of training, optimize the parameters θ' of the convolutional neural network F:

[0106]

[0107] Where α represents the learning rate, and (θ')' represents the parameters of the optimized convolutional neural network F';

[0108] S204. Save the final parameters θ' of the manufacturing enterprise cluster energy consumption prediction network F after e rounds of pre-training.

[0109] S3. Use the overall electricity consumption dataset E of the manufacturing enterprise cluster to train the energy consumption situation prediction network F of the manufacturing enterprise cluster, transfer the final parameter θ' as the initial parameter for training, optimize the network parameters by minimizing the loss function, and save the final parameter θ when the model is fitted.

[0110] S301, The manufacturing enterprise cluster energy consumption situation prediction network F is trained using the overall electricity consumption dataset E of the manufacturing enterprise cluster, and the number of training rounds e and the number of input data batches B are set for the manufacturing enterprise cluster energy consumption situation prediction network F.

[0111] S302. For any batch of data in any training round, calculate the loss between the predicted value and the true value using the loss function L:

[0112]

[0113] S303. After each round of training, optimize the parameters θ of the energy consumption situation prediction network F for manufacturing enterprise clusters:

[0114]

[0115] Where (θ)' represents the parameters of the optimized energy consumption prediction network F for manufacturing enterprise clusters;

[0116] S304. Save the final parameters θ of the energy consumption situation prediction network F of the manufacturing enterprise cluster after e rounds of training, and the training ends.

[0117] S4. Input the real-time collected overall electricity consumption data of the manufacturing enterprise cluster into the trained manufacturing enterprise cluster energy consumption situation prediction network F to realize the overall energy consumption situation prediction of the manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis.

[0118] The key to accurate prediction of energy consumption patterns in manufacturing enterprise clusters lies in enhancing the model's analysis of the energy consumption characteristics of different manufacturing enterprises within the cluster, enabling the prediction model to accurately uncover the dependencies between time steps in historical energy consumption data. This invention fully utilizes the ability of convolutional neural networks to extract two-dimensional representations of data and the ability of recurrent neural networks to model implicit temporal relationships, thereby addressing the challenge of predicting the overall energy consumption patterns of manufacturing enterprise clusters due to the significant differences in energy consumption characteristics caused by the heterogeneity of manufacturing processes among different enterprises.

[0119] This invention discloses a method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis. It involves collecting energy consumption data from manufacturing enterprise clusters to construct enterprise energy consumption datasets and cluster energy consumption datasets; constructing a manufacturing enterprise energy consumption situation prediction network including a characteristic analysis module and a situation prediction module; pre-training the manufacturing enterprise cluster energy consumption situation prediction network using the enterprise energy consumption dataset, optimizing the model parameters using the mean squared error function, and saving the final model parameters to achieve analysis of heterogeneous energy consumption characteristics within the cluster; training the manufacturing enterprise energy consumption situation prediction network using the cluster energy consumption dataset, transferring the pre-trained parameters as initial parameters for training, optimizing the model parameters using the mean squared error function, and saving the final model parameters to achieve accurate prediction of the energy consumption situation of the manufacturing enterprise cluster. This invention can achieve energy consumption situation prediction for manufacturing enterprise clusters containing multiple manufacturing enterprises.

[0120] like Figure 2 As shown, the present invention also provides a device for predicting the energy consumption situation of a manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis, comprising:

[0121] Data acquisition module 1 is used to collect electricity consumption data of manufacturing enterprises and construct an electricity consumption dataset E of the manufacturing enterprises based on the electricity consumption data. i The data set E represents the overall electricity consumption of the manufacturing enterprise cluster; wherein the manufacturing enterprise cluster includes multiple manufacturing enterprises.

[0122] Module 2 is used to construct an energy consumption prediction network F for a cluster of manufacturing enterprises, and utilizes the electricity consumption dataset E of the manufacturing enterprises. i Training is performed, and the network parameters are optimized by minimizing the loss function. The final parameters θ' of the model fitting are saved to enable the analysis of heterogeneous energy consumption characteristics of different manufacturing enterprises within a manufacturing enterprise cluster.

[0123] Training module 3 is used to train the energy consumption status prediction network F of the manufacturing enterprise cluster using the overall electricity consumption dataset E of the manufacturing enterprise cluster, transfer the final parameter θ' as the initial parameter for training, optimize the network parameters by minimizing the loss function, and save the final parameter θ when the model is fitted.

[0124] Prediction module 4 is used to input the real-time collected overall electricity consumption data of the manufacturing enterprise cluster into the trained manufacturing enterprise cluster energy consumption situation prediction network F, so as to realize the overall energy consumption situation prediction of the manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis.

[0125] In one embodiment, in acquisition module 1,

[0126] The granularity of the electricity consumption data is 15 minutes, and the unit is kW·h;

[0127] The electricity consumption dataset E of the manufacturing enterprisei The overall electricity consumption dataset E for the manufacturing enterprise cluster is as follows:

[0128]

[0129] E = {x1, x2, ..., x} n}

[0130] in, Let x represent the electricity consumption data of the i-th manufacturing enterprise at the n-th time step. tn This represents the electricity consumption data at the nth time step of the manufacturing enterprise cluster.

[0131] In one embodiment, the energy consumption prediction network F for the manufacturing enterprise cluster includes a characteristic analysis module M. a and situation prediction module M p .

[0132] In one embodiment, in the manufacturing enterprise cluster energy consumption prediction network F of construction module 2, the characteristic analysis module M a Includes the following steps:

[0133] Energy consumption data frequency selection:

[0134]

[0135] Where x[n] represents the nth input time series energy consumption data, X[k] represents the kth transformed frequency information, and N represents the sample length of the input energy consumption data. Complex root of unity;

[0136] Two-dimensional reconstruction of energy consumption data:

[0137]

[0138] Where A[k] represents the amplitude corresponding to the k-th frequency, x X[k] f represents the electricity consumption data of the k-th input. avg (·) denotes the mean function, g amp (·) represents the amplitude acquisition function;

[0139]

[0140] Where Lx represents the length of the energy data input, and P[k] represents the k-th frequency;

[0141] Characterization and extraction of heterogeneous properties:

[0142]

[0143] in, Let x represent the j-th two-dimensional convolution operation. i Let x' represent the energy consumption data of the i-th input segment. i This represents the output data after the convolution operation;

[0144] One-dimensional restructuring and importance weighting of energy consumption data:

[0145] S i =Softmax(A i )

[0146] Where Softmax(·) represents the Softmax function, A i S represents the i-th amplitude. i This represents the importance score corresponding to the i-th amplitude.

[0147]

[0148] in, and These represent the energy consumption data after one-dimensional recombination and the energy consumption data after weighting, respectively.

[0149]

[0150] In the formula, x f This indicates the final output.

[0151] In one embodiment, in the energy consumption situation prediction network F of the manufacturing enterprise cluster of construction module 2, the situation prediction module M p It includes a unidirectional LSTM layer and a linear classification layer, used for time modeling and prediction of energy consumption data that has undergone heterogeneous energy consumption characteristic analysis.

[0152] In one embodiment, construction module 2 specifically includes:

[0153] The setting unit is used to set the number of pre-training rounds e and the batch size B of the input data for the energy consumption status prediction network F of the manufacturing enterprise cluster;

[0154] For any batch of data in any training round, calculate the loss between the predicted value and the true value using the loss function L:

[0155]

[0156] in, x represents the predicted consumption value at time step ti. ti This represents the actual consumption value for time step ti;

[0157] The first optimization unit is used to optimize the parameters θ' of the convolutional neural network F after each round of training:

[0158]

[0159] Where α represents the learning rate, and (θ')' represents the parameters of the optimized convolutional neural network F';

[0160] The first storage unit is used to store the final parameters θ' of the manufacturing enterprise cluster energy consumption situation prediction network F after e rounds of pre-training.

[0161] In one embodiment, the steps of training the energy consumption prediction network F of the manufacturing enterprise cluster using the overall electricity consumption dataset E of the manufacturing enterprise cluster, transferring the final parameter θ' as the initial parameter for training, optimizing the network parameters by minimizing the loss function, and saving the final parameter θ during model fitting specifically include:

[0162] The setting unit is used to set the number of training rounds e and the batch size B of the input data for the energy consumption status prediction network F of the manufacturing enterprise cluster;

[0163] The computational unit is used to calculate the loss between the predicted value and the true value using the loss function L for any batch of data in any training round.

[0164]

[0165] The second optimization unit is used to optimize the parameters θ of the energy consumption pattern prediction network F of the manufacturing enterprise cluster after each round of training:

[0166]

[0167] Where (θ)' represents the parameters of the optimized energy consumption prediction network F for manufacturing enterprise clusters;

[0168] The second storage unit is used to store the final parameters θ of the energy consumption situation prediction network F of the manufacturing enterprise cluster after e rounds of training, at which point training ends.

[0169] Each of the above modules and units is used to perform the respective steps in the above method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristics analysis. The specific implementation methods are as described in the above method embodiments, and will not be repeated here.

[0170] like Figure 3 As shown, the present invention also provides a computer device, which may be a server, and its internal structure may be as follows: Figure 3As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores all data required for the process of predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis.

[0171] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment on which the present application is applied.

[0172] An embodiment of this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the above-described methods for predicting the energy consumption status of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis.

[0173] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media provided in this application and in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0174] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0175] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for predicting the energy consumption pattern of a manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis, characterized in that, include: Collect electricity consumption data from manufacturing enterprises, and construct electricity consumption datasets for each manufacturing enterprise based on the electricity consumption data. Data set of overall electricity consumption of manufacturing enterprise clusters The manufacturing cluster includes multiple manufacturing enterprises. Constructing a network for predicting energy consumption patterns in manufacturing enterprise clusters The network Includes feature analysis module and situation prediction module The characteristic analysis module Includes the following steps: Energy consumption data frequency selection: (1) in, Indicates the first The input time series data can be used. Represents the transformed th Frequency information, This indicates the sample length of the input energy consumption data. Represents the root of unity of a complex number; Two-dimensional reconstruction of energy consumption data: (2) (3) in, Indicates the length of the input energy data. Indicates the first One frequency; Heterogeneity feature extraction: (4) in, Indicates the first A two-dimensional convolution operation, Indicates the input number of the first... Section energy usage data, This represents the output data after the convolution operation; One-dimensional restructuring and importance weighting of energy consumption data: (5) in, Indicates the first Each amplitude, Indicates the first The importance score corresponding to each amplitude value; (6) in, Represents the first dimension after one-dimensional recombination. Section energy usage data, This represents the weighted energy consumption data; (7) in, This indicates the feature analysis module The final output; Using the electricity consumption datasets of the aforementioned manufacturing enterprises Energy consumption prediction network for the aforementioned manufacturing enterprise cluster Pre-training is performed to optimize the network parameters by minimizing the loss function, and the final parameters from the model fitting process are saved. This enables the analysis of heterogeneous energy consumption characteristics among different manufacturing enterprises within a manufacturing enterprise cluster; Using the overall electricity consumption dataset of the manufacturing enterprise cluster Energy consumption prediction network for the aforementioned manufacturing enterprise cluster Training is performed, and the final parameters are... The initial parameters are used for transfer learning, and the network parameters are optimized by minimizing the loss function. The final parameters used for model fitting are then saved. ; The real-time collected overall electricity consumption data of the manufacturing enterprise cluster is input into the trained energy consumption prediction network of the manufacturing enterprise cluster. This enables the prediction of the overall energy consumption status of manufacturing enterprise clusters based on the analysis of heterogeneous energy consumption characteristics.

2. The method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis according to claim 1, characterized in that, The process involves collecting electricity consumption data from manufacturing enterprises and constructing an electricity consumption dataset for these enterprises based on this data. Data set of overall electricity consumption of manufacturing enterprise clusters In the steps: The granularity of the electricity consumption data is 15 minutes, and the unit is... ; The electricity consumption dataset of the manufacturing enterprise Data set of overall electricity consumption of manufacturing enterprise clusters for: (8) (9) in, Indicates the first The first manufacturing enterprise Electricity consumption data at each time step Indicates the first manufacturing enterprise cluster Electricity consumption data at each time step.

3. The method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis according to claim 1, characterized in that, The situation prediction module It includes a unidirectional LSTM layer and a linear classification layer, used for time modeling and prediction of energy consumption data that has undergone heterogeneous energy consumption characteristic analysis.

4. The method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis according to claim 1, characterized in that, The electricity consumption dataset of each manufacturing enterprise is utilized. Energy consumption prediction network for the aforementioned manufacturing enterprise cluster The pre-training process includes the following steps: Setting up a network for predicting energy consumption patterns in manufacturing enterprise clusters Number of pre-training rounds and the batch size of the input data ; For any batch of data in any training round, using the loss function Calculate the loss between the predicted value and the actual value: (10) in, Indicates the first Predicted energy consumption values ​​at each time step. Indicates the first The actual energy consumption value at each time step Indicates the number of samples; After each round of training, the energy consumption prediction network for the manufacturing enterprise cluster is evaluated. Pre-trained parameters Optimize: (11) in, Indicates the learning rate. This represents the optimized pre-trained parameters; save The manufacturing enterprise cluster energy consumption prediction network after rounds of pre-training final parameters .

5. The method for predicting the energy consumption situation of manufacturing enterprise clusters based on heterogeneous energy consumption characteristic analysis according to claim 1, characterized in that, The use of the overall electricity consumption dataset of the manufacturing enterprise cluster Energy consumption prediction network for the aforementioned manufacturing enterprise cluster The training process includes the following steps: Setting up a network for predicting energy consumption patterns in manufacturing enterprise clusters Number of training rounds and the batch size of the input data ; For any batch of data in any training round, using the loss function Calculate the loss between the predicted value and the actual value: (12) in, Indicates the first Predicted energy consumption values ​​at each time step. Indicates the first The actual energy consumption value at each time step Indicates the number of samples; After each round of training, the energy consumption prediction network for the manufacturing enterprise cluster is evaluated. parameters Optimize: (13) in, Indicates the learning rate. This represents the optimized energy consumption prediction network for the manufacturing enterprise cluster. Parameters; save The energy consumption prediction network for manufacturing enterprise clusters after rounds of training final parameters Training is over.

6. A device for predicting the energy consumption situation of a manufacturing enterprise cluster based on heterogeneous energy consumption characteristic analysis, characterized in that, include: The data acquisition module is used to collect electricity consumption data from manufacturing enterprises and construct an electricity consumption dataset for the manufacturing enterprises based on the electricity consumption data. Data set of overall electricity consumption of manufacturing enterprise clusters The manufacturing cluster includes multiple manufacturing enterprises. The building blocks are used to construct a network for predicting the energy consumption of manufacturing enterprise clusters. The network Includes feature analysis module and situation prediction module The characteristic analysis module Used to perform the steps of energy consumption data frequency selection, two-dimensional reconstruction of energy consumption data, heterogeneous feature extraction, one-dimensional reconstruction of energy consumption data and importance weighting as described in claim 1; A pre-training module is used to utilize the electricity consumption dataset of the manufacturing enterprise. Energy consumption prediction network for the aforementioned manufacturing enterprise cluster Pre-training is performed to optimize the network parameters by minimizing the loss function, and the final parameters from the model fitting process are saved. This enables the analysis of heterogeneous energy consumption characteristics among different manufacturing enterprises within a manufacturing enterprise cluster; The training module is used to utilize the overall electricity consumption dataset of the manufacturing enterprise cluster. Energy consumption prediction network for the aforementioned manufacturing enterprise cluster Training is performed, and the final parameters are... The initial parameters are used for transfer learning, and the network parameters are optimized by minimizing the loss function. The final parameters used for model fitting are then saved. ; The prediction module is used to input the real-time collected overall electricity consumption data of the manufacturing enterprise cluster into the trained energy consumption situation prediction network of the manufacturing enterprise cluster. This enables the prediction of the overall energy consumption status of manufacturing enterprise clusters based on the analysis of heterogeneous energy consumption characteristics.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images