Discrete manufacturing enterprise energy consumption prediction method suitable for dynamic production

Abstract

The invention discloses a discrete manufacturing enterprise energy consumption prediction method suitable for dynamic production, and the method comprises the steps: collecting the time sequence energy consumption data of a discrete manufacturing enterprise, and constructing a data set, so as to obtain an energy consumption curve of the discrete manufacturing enterprise; rearranging the 1D time sequence energy consumption data into a 2D matrix by adopting a preset calculation mode; processing the 2D matrix by adopting preset convolution modules, carrying out weighted summation calculation on the data after data processing, then stacking the convolution modules and connecting the convolution modules by adopting residual errors; and inputting an output result of the convolution module into a full connection layer, and generating a prediction result at the next moment, thereby realizing energy consumption prediction of the discrete manufacturing enterprise. According to the method, the time sequence periodic information mining method is adopted to identify the energy consumption time sequence change of the discrete manufacturing enterprise caused by the dynamic change of production and equipment, so that the energy consumption prediction precision of the discrete manufacturing enterprise is improved, and the accurate prediction of the energy consumption of the discrete manufacturing enterprise with large energy consumption fluctuation and insignificant consumption rule is realized.

Description

Technical Field

[0001] The present invention relates to cross-technical fields such as industrial big data analysis and industrial energy management, and in particular to an energy consumption prediction method for discrete manufacturing enterprises that adapts to dynamic production. Background Art

[0002] Energy consumption forecasting for discrete manufacturing companies is crucial for understanding the energy demand of future production and manufacturing processes and achieving balanced scheduling of energy supply and use. Especially in situations such as energy shortages, accurate energy consumption forecasting can help companies develop reasonable energy supply plans and prioritize the energy needs of important production activities, thereby reducing manufacturing energy costs while promoting green and sustainable development of companies. However, due to the variability of production tasks in discrete manufacturing processes, the energy consumption of discrete manufacturing companies in the manufacturing process fluctuates greatly and is difficult to accurately predict. How to accurately use the large amount of production energy consumption data that emerges in the production process to achieve accurate prediction of the energy consumption of the entire park of discrete manufacturing companies has become a hot issue that needs to be urgently addressed in the current field of industrial energy conservation and carbon reduction.

[0003] Early research on energy consumption prediction for discrete manufacturing enterprises focused on modeling energy consumption mechanisms at multiple levels, including equipment, process, workshop, and enterprise, and formed a prediction method driven by a mechanism model. Since such methods require the construction of multivariate associations between production information and multi-equipment operation mechanisms, they are mostly used in process or semi-process scenarios with fixed production modes and stable production conditions, but they are still difficult to meet the demands of discrete manufacturing enterprises with complex production and changing production plans for fast and low-cost prediction.

[0004] With the development of industrial artificial intelligence technology, enterprises can monitor and collect time-series energy consumption data in the production process by deploying smart devices and the industrial Internet. Based on the collected time-series energy consumption data, researchers have proposed a data-driven modeling and prediction method for the energy consumption system of discrete manufacturing enterprises, using random forests, support vector machines and other methods for time-series prediction. Traditional machine learning-based methods have good prediction effects on stable time-series energy consumption data, but cannot meet the prediction of large changes in energy consumption data.

[0005] In summary, the accurate prediction of energy consumption in discrete manufacturing enterprises with large fluctuations in energy consumption and unclear consumption patterns still faces the following technical difficulties: the time span of the time series energy consumption data of discrete manufacturing enterprises is long, the data volume is large, and the prediction model requires high stability; the frequent changes in production tasks and plans cause the time series energy consumption data to show complex periodicity and random fluctuations, making feature extraction and trend identification difficult. Summary of the invention

[0006] The present invention provides an energy consumption prediction method for discrete manufacturing enterprises that adapt to dynamic production. The method adopts a time series periodic information mining method to identify the time series changes in energy consumption of discrete manufacturing enterprises caused by production and equipment changes, so as to improve the accuracy of energy consumption prediction for discrete manufacturing enterprises and realize accurate prediction of energy consumption for discrete manufacturing enterprises with large energy consumption fluctuations and unclear consumption patterns.

[0007] The present invention provides a discrete manufacturing enterprise energy consumption prediction method adapted to dynamic production, comprising:

[0008] Collecting the time series energy consumption data of discrete manufacturing enterprises and constructing a data set to obtain the energy consumption curve of the discrete manufacturing enterprises; wherein the time series energy consumption data is 1D time series energy consumption data;

[0009] Rearranging the 1D time series energy consumption data into a 2D matrix using a preset calculation method so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data;

[0010] The 2D matrix is ​​processed by a preset convolution module, and the weighted sum of the data is calculated, and then the convolution modules are stacked, and residual connections are used to ensure that information is effectively transmitted and aggregated in the deep network;

[0011] The output result of the convolution module is input into the fully connected layer to generate a prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

[0012] Furthermore, the step of rearranging the 1D time series energy consumption data into a 2D matrix by using a preset calculation method so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data includes:

[0013] Fourier transform is used to analyze the frequency domain of the time series energy consumption data to identify the main cycle in the energy consumption data, and then the first k frequencies whose amplitudes exceed the set value are selected by calculating the frequency spectrum to determine the length of the corresponding cycle;

[0014] Based on the identified periodic features, the selected k original 1D time series energy consumption data are rearranged into k 2D matrices to capture the dynamic characteristics of short-term and long-term time series energy consumption data respectively; wherein the columns of the matrix represent the energy consumption changes within a cycle, and the rows of the matrix represent the energy consumption changes at the same time points between different cycles.

[0015] Furthermore, the step of using Fourier transform to analyze the frequency domain of the time series energy consumption data to identify the main cycle in the energy consumption data, and then selecting the first k frequencies whose amplitudes exceed the set value by calculating the frequency spectrum to determine the length of the corresponding cycle includes:

[0016] The cycle of the time series energy consumption data is calculated using the formula:

[0017] S=F ave (F amp (F FFT (X 1D )))

[0018]

[0019]

[0020] in, T×C represents the time series energy consumption data of enterprises with length T and variable C; F FFT (·) indicates Fourier transform of the time series energy consumption data to convert the time domain features into frequency domain features; F amp (·) indicates the amplitude value in the frequency domain curve of the time series energy consumption; F ave (·) means averaging the amplitudes obtained at different frequencies; argFirstk(·) means selecting the first k frequencies whose amplitudes exceed the set value {f 1 ,…f k}, and its corresponding period length is {p 1 ,…p k}, k is a hyperparameter.

[0021] Furthermore, the step of rearranging the selected k original 1D time series energy consumption data into k 2D matrices based on the identified periodic features so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data respectively includes:

[0022] Based on the selected frequency {f 1 ,…f k} and the corresponding period length {p 1 ,…p k}, rearrange the 1D time series energy consumption data into a 2D matrix; where the columns represent the energy consumption changes within a cycle, and the rows represent the energy consumption changes at the same time point in different cycles. The representation of the 2D matrix is ​​as follows:

[0023]

[0024] Among them, F pad (·) indicates that the original time series energy consumption data is filled with 0; F res (·) represents the reconstruction of 1D time series energy consumption data into a 2D matrix according to the cycle information, where f i represents the frequency of occurrence of the i-th frequency in the time series energy consumption data, p i Indicates the length of the cycle.

[0025] Furthermore, the steps of using a preset convolution module to process the 2D matrix, performing weighted sum calculation on the data after the data, and then stacking the convolution modules, and using residual connections to ensure effective transmission and aggregation of information in the deep network include:

[0026] The Inception module is used to convert the processed 2D matrix back to 1D form; wherein the Inception module is a module that uses multi-scale 2D convolution kernels;

[0027] The calculated cycle amplitude is used as the weight, and the Softmax function is used to perform weighted summation on different 1D time series energy consumption data, and then fused as the output of the current layer;

[0028] The convolution modules are stacked and residual connections are used to ensure effective transmission and aggregation of information in the deep network.

[0029] Furthermore, the step of converting the processed 2D matrix back to 1D form using the Inception module includes:

[0030] The Inception module is used to perform convolution on the converted 2D matrix to extract features, and aggregate the changes in the enterprise time series energy consumption data within a cycle and at the same time point between different cycles:

[0031]

[0032] Among them, F inception (·) is a deep learning module that processes 2D matrices in parallel using convolution kernels of different sizes to capture the short-term and long-term variation characteristics of time-series energy consumption data; Contains multi-scale periodicity and time-dependent features extracted from time-series energy consumption data.

[0033] Convert the convolutional 2D matrix to 1D and truncate the previously padded part:

[0034]

[0035] Among them, F tru (·) indicates truncation, which is used to remove the filling time point, that is, F pad (·) The 0 added in the middle restores the time series energy consumption data to its original length; Indicates that the processed 2D matrix is ​​converted back into 1D time series energy consumption data.

[0036] Furthermore, the step of taking the calculated period amplitude as a weight, using the Softmax function to perform weighted summation on different 1D time series energy consumption data, and fusing them into the output of the current layer includes:

[0037] The Softmax function is used to perform weighted fusion on different 1D time series energy consumption data to achieve adaptive aggregation of the hierarchical convolution module:

[0038]

[0039]

[0040] Among them, F softmax (·) Convert the amplitude of each frequency into a weight to indicate the importance of the frequency in the overall feature; It means that the 1D features corresponding to multiple frequencies are aggregated by weighted summation to obtain the output.

[0041] Furthermore, in the step of stacking the convolution modules and using residual connections to ensure effective transmission and aggregation of information in the deep network, the calculation formula is:

[0042]

[0043] in, represents the output of the convolution module; Φ(·) represents the convolution module; Represents the output of the l-1th.

[0044] The present invention also provides a discrete manufacturing enterprise energy consumption prediction device adapted to dynamic production, comprising:

[0045] A collection module, used to collect the time series energy consumption data of discrete manufacturing enterprises and construct a data set to obtain the energy consumption curve of the discrete manufacturing enterprises; wherein the time series energy consumption data is 1D time series energy consumption data;

[0046] An arrangement module, used for rearranging the 1D time series energy consumption data into a 2D matrix by a preset calculation method, so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data;

[0047] The calculation module processes the 2D matrix using a preset convolution module, performs weighted sum calculation on the data, and then stacks the convolution modules, using residual connections to ensure effective transmission and aggregation of information in the deep network;

[0048] The generation module is used to input the output result of the convolution module into the fully connected layer to generate a prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

[0049] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0050] The present invention also provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the steps of the above method when executed by a processor.

[0051] The beneficial effects of the present invention are:

[0052] 1. The present invention converts one-dimensional time series energy consumption data into a two-dimensional matrix containing multi-time period period information, which can effectively identify the periodic characteristics in the enterprise's time series consumption data and separate local fluctuations from global trends, thereby overcoming the characteristic of the overall dynamic change of energy consumption of discrete manufacturing enterprises and realizing accurate prediction of energy consumption.

[0053] 2. The present invention fully considers the dynamic nature of production tasks and plans of discrete manufacturing enterprises, and adopts a two-dimensional convolution method, so that the model can simultaneously handle the complex relationships within and between cycles, thereby improving the feature extraction capability, and is suitable for realizing accurate prediction of the energy consumption of discrete manufacturing enterprises at the next moment. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 The present invention is a flow chart of a method for predicting energy consumption of discrete manufacturing enterprises that is adapted to dynamic production.

[0055] Figure 2 The present invention is a schematic diagram of the structure of the energy consumption prediction device for discrete manufacturing enterprises that is adapted to dynamic production.

[0056] Figure 3 It is a schematic diagram of the internal structure of the computer device in the present invention.

[0057] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0058] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0059] like Figure 1 As shown, the present invention provides a discrete manufacturing enterprise energy consumption prediction method adapted to dynamic production, comprising:

[0060] S1. Collecting the time series energy consumption data of discrete manufacturing enterprises and constructing a data set to obtain the energy consumption curve of the discrete manufacturing enterprises; wherein the time series energy consumption data is 1D time series energy consumption data;

[0061] S2. Rearranging the 1D time series energy consumption data into a 2D matrix using a preset calculation method, so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data; specifically including:

[0062] S21. In order to process the time series energy consumption data of discrete manufacturing enterprises, the frequency domain of the time series energy consumption data is analyzed by Fourier transform to identify the main cycles in the energy consumption data. By analyzing the frequency spectrum, the first k frequencies whose amplitudes exceed the set value are selected to determine the length of the corresponding cycle;

[0063] The cycle of the time series energy consumption data is calculated using the formula:

[0064] S=F ave (F amp (F FFT (X 1D )))

[0065]

[0066]

[0067] in, T×C represents the time series energy consumption data of enterprises with length T and variable C; F FFT (·) indicates Fourier transform of the time series energy consumption data to convert the time domain features into frequency domain features; F amp (·) indicates the amplitude value in the frequency domain curve of the time series energy consumption; F ave (·) means averaging the amplitudes obtained at different frequencies; argFirstk(·) means selecting the first k frequencies whose amplitudes exceed the set value {f 1 ,…f k}, the period lengths corresponding to these frequencies are {p 1 ,…p k}, k is a hyperparameter. The above three formulas can be abbreviated as:

[0068] S,{f 1 ,…f k},{p 1 ,…p k}=F(X 1D )

[0069] S22. Based on the identified periodic features, the selected k original 1D time series energy consumption data are rearranged into k 2D matrices so as to capture the dynamic characteristics of short-term and long-term time series energy consumption data respectively; wherein the columns of the matrix represent the energy consumption changes within a cycle, and the rows of the matrix represent the energy consumption changes at the same time points between different cycles.

[0070] Based on the selected frequency {f 1 ,…f k} and the corresponding period length {p 1 ,…p k}, the 1D time series energy consumption data is rearranged into a 2D matrix; where the columns represent the energy consumption changes within a cycle, and the rows represent the energy consumption changes at the same time point in different cycles. The 2D matrix representation can better capture complex time change patterns, as shown in the following formula:

[0071]

[0072] Among them, F pad (·) indicates that the original time series energy consumption data is padded with 0 so that its length can adapt to the dimension requirements in the next step of reconstruction to avoid data loss or dimension mismatch; F res (·) represents the reconstruction of 1D time series energy consumption data into a 2D matrix according to the cycle information, where f i represents the frequency of occurrence of the i-th frequency in the time series energy consumption data, p i Indicates the length of the cycle.

[0073] S3, using a preset convolution module to process the 2D matrix, and performing weighted sum calculation on the data after the data, and then stacking the convolution modules, using residual connections to ensure effective transmission and aggregation of information in the deep network; specifically including:

[0074] S31, Inception module processes time series energy consumption data through multi-scale 2D convolution kernels to capture changes in different periods and avoid misidentification of short-term fluctuations as long-term trends. The Inception module is used to convert the processed 2D matrix back to 1D form for subsequent calculations;

[0075] The Inception module is used to perform convolution on the converted 2D matrix to extract features, and aggregate the changes in the enterprise time series energy consumption data within a cycle and at the same time point between different cycles:

[0076]

[0077] Among them, F inception (·) is a deep learning module that processes 2D matrices in parallel using convolution kernels of different sizes to capture the short-term and long-term variation characteristics of time-series energy consumption data; Contains multi-scale periodicity and time-dependent features extracted from time-series energy consumption data.

[0078] Convert the convolutional 2D matrix to 1D and truncate the previously padded part:

[0079]

[0080] Among them, F tru (·) indicates truncation, which is used to remove the filling time point, that is, F pad(·) The 0 added in the middle restores the time series energy consumption data to its original length; Indicates that the processed 2D matrix is ​​converted back into 1D time series energy consumption data.

[0081] S32, taking the calculated cycle amplitude as the weight, using the Softmax function to perform weighted summation on different 1D time series energy consumption data, and merging them into the output of the current layer;

[0082] The Softmax function is used to perform weighted fusion on different 1D time series energy consumption data to achieve adaptive aggregation of the hierarchical convolution module:

[0083]

[0084]

[0085] Among them, F softmax (·) Convert the amplitude of each frequency into a weight to indicate the importance of the frequency in the overall feature; It means that the 1D features corresponding to multiple frequencies are aggregated by weighted summation to obtain the output.

[0086] S33, stack the convolution modules and use residual connections to ensure effective transmission and aggregation of information in the deep network. The calculation formula is:

[0087]

[0088] in, represents the output of the convolution module; Φ(·) represents the convolution module; Represents the output of the l-1th.

[0089] S4. Input the output result of the convolution module into the fully connected layer to generate a prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

[0090] The data used in this paper comes from a total of 31,395 power consumption values ​​of a discrete manufacturing enterprise from May 2023 to June 2024, with a data collection interval of 15 minutes. The experiment divides the data set into a training set, a validation set, and a test set, with a ratio of 8:1:1.

[0091] The hyperparameters involved in the training and optimization process used in the experiment are shown in Table 1, the evaluation indicators used are shown in Table 2, and the experimental results are shown in Table 3.

[0092] Table 1. Training and optimization hyperparameters used in the experiments

[0093]

[0094] Table 2 Evaluation indicators used in the experiment

[0095]

[0096] Table 3 Experimental results

[0097]

[0098] It can be seen from the experimental results that the method of the present invention is superior to the Transformer model. The method of the present invention considers the overall dynamic variability and can superimpose long-term information, which can effectively overcome the influence of short-term dynamics.

[0099] The present invention discloses a method for predicting energy consumption of discrete manufacturing enterprises considering production dynamics. By acquiring the historical time series energy consumption data of discrete manufacturing enterprises, a data set is constructed, and the overall energy consumption curve of the enterprise is fitted; the frequency domain characteristics of the time series energy consumption data are analyzed by Fourier transform, and multiple frequencies with large amplitudes are selected to determine the cycle length; a convolution module is constructed to convert the one-dimensional time series energy consumption data into a two-dimensional matrix, the columns represent the changes in energy consumption of the enterprise within a cycle, and the rows represent the changes in energy consumption of the enterprise in the same time period between different cycles; convolution processing is performed on the two-dimensional matrix, the characteristics of the enterprise within a cycle and the same time period during the same cycle are aggregated, and the two-dimensional matrix is ​​converted into a one-dimensional sequence for weighted fusion; residual connection is performed on the outputs of multiple convolution modules, and the results are output, thereby realizing the energy consumption prediction of discrete manufacturing enterprises. The present invention can realize the accurate prediction of energy consumption of discrete manufacturing enterprises, thereby promoting the green and sustainable development of enterprises while realizing the reduction of manufacturing energy costs.

[0100] like Figure 2 As shown, the present invention also provides a discrete manufacturing enterprise energy consumption prediction device adapted to dynamic production, comprising:

[0101] The acquisition module 1 is used to collect the time series energy consumption data of the discrete manufacturing enterprise and construct a data set to obtain the energy consumption curve of the discrete manufacturing enterprise; wherein the time series energy consumption data is 1D time series energy consumption data;

[0102] Arrangement module 2, used for rearranging the 1D time series energy consumption data into a 2D matrix by a preset calculation method, so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data;

[0103] The calculation module 3 processes the 2D matrix using a preset convolution module, performs weighted sum calculation on the data after the data is processed, and then stacks the convolution modules, using residual connections to ensure effective transmission and aggregation of information in the deep network;

[0104] The generation module 4 is used to input the output result of the convolution module into the fully connected layer to generate the prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

[0105] In one embodiment, the arrangement module 2 comprises:

[0106] An identification unit, used for analyzing the frequency domain of the time series energy consumption data by Fourier transform to identify the main cycle in the energy consumption data, and then selecting the first k frequencies whose amplitudes exceed the set value by calculating the frequency spectrum to determine the length of the corresponding cycle;

[0107] The arrangement unit is used to rearrange the selected k original 1D time series energy consumption data into k 2D matrices based on the identified periodic features, so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data respectively; wherein the columns of the matrix represent the energy consumption changes within a cycle, and the rows of the matrix represent the energy consumption changes at the same time point between different cycles.

[0108] In one embodiment, the identification unit comprises:

[0109] The cycle of the time series energy consumption data is calculated using the formula:

[0110] S=F ave (F amp (F FFT (X 1D )))

[0111]

[0112]

[0113] in, T×C represents the time series energy consumption data of enterprises with length T and variable C; F FFT (·) indicates Fourier transform of the time series energy consumption data to convert the time domain features into frequency domain features; F amp (·) indicates the amplitude value in the frequency domain curve of the time series energy consumption; F ave (·) means averaging the amplitudes obtained at different frequencies; argFirstk(·) means selecting the first k frequencies whose amplitudes exceed the set value {f 1 ,…f k}, and its corresponding period length is {p 1 ,…p k}, k is a hyperparameter.

[0114] In one embodiment, the arranging unit comprises:

[0115] Based on the selected frequency {f 1 ,…f k} and the corresponding period length {p 1 ,…p k}, rearrange the 1D time series energy consumption data into a 2D matrix; where the columns represent the energy consumption changes within a cycle, and the rows represent the energy consumption changes at the same time point in different cycles. The representation of the 2D matrix is ​​as follows:

[0116]

[0117] Among them, F pad (·) indicates that the original time series energy consumption data is filled with 0; F res (·) represents the reconstruction of 1D time series energy consumption data into a 2D matrix according to the cycle information, where f i represents the frequency of occurrence of the i-th frequency in the time series energy consumption data, p i Indicates the length of the cycle.

[0118] In one embodiment, the calculation module 3 includes:

[0119] A conversion unit, used to convert the processed 2D matrix back to 1D form using an Inception module; wherein the Inception module is a module using a multi-scale 2D convolution kernel;

[0120] The weighting unit is used to take the calculated period amplitude as the weight, use the Softmax function to perform weighted summation on different 1D time series energy consumption data, and fuse them into the output of the current layer;

[0121] The stacking unit is used to stack convolutional modules and adopt residual connections to ensure effective transmission and aggregation of information in deep networks.

[0122] In one embodiment, the conversion unit includes:

[0123] The extraction subunit is used to perform convolution on the converted 2D matrix using the Inception module to extract features and aggregate the changes in the enterprise time series energy consumption data within a cycle and at the same time point between different cycles:

[0124]

[0125] Among them, F inception (·) is a deep learning module that processes 2D matrices in parallel using convolution kernels of different sizes to capture the short-term and long-term variation characteristics of time-series energy consumption data; Contains multi-scale periodicity and time-dependent features extracted from time-series energy consumption data.

[0126] The transformation subunit is used to convert the 2D matrix after convolution to 1D and truncate the previously filled part:

[0127]

[0128] Among them, F tru (·) indicates truncation, which is used to remove the filling time point, that is, F pad (·) The 0 added in the middle restores the time series energy consumption data to its original length; Indicates that the processed 2D matrix is ​​converted back into 1D time series energy consumption data.

[0129] In one embodiment, the weighting unit comprises:

[0130] The Softmax function is used to perform weighted fusion on different 1D time series energy consumption data to achieve adaptive aggregation of the hierarchical convolution module:

[0131]

[0132]

[0133] Among them, F softmax (·) Convert the amplitude of each frequency into a weight to indicate the importance of the frequency in the overall feature; It means that the 1D features corresponding to multiple frequencies are aggregated by weighted summation to obtain the output.

[0134] In one embodiment, in the stacking unit, the calculation formula is:

[0135]

[0136] in, represents the output of the convolution module; Φ(·) represents the convolution module; Represents the output of the l-1th.

[0137] The above modules, units and sub-units are used to execute the corresponding steps in the above discrete manufacturing enterprise energy consumption prediction method adapted to dynamic production. The specific implementation method thereof is described in the above method embodiment and will not be repeated here.

[0138] like Figure 3 As shown, the present invention also provides a computer device, which can be a server, and its internal structure can be as shown in Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store all data required for the process of the energy consumption prediction method for discrete manufacturing enterprises that adapts to dynamic production. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the energy consumption prediction method for discrete manufacturing enterprises that adapts to dynamic production is implemented.

[0139] Those skilled in the art will understand that Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0140] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, any one of the above-mentioned discrete manufacturing enterprise energy consumption prediction methods adapted to dynamic production is implemented.

[0141] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used 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), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0142] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0143] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting energy consumption of discrete manufacturing enterprises that adapts to dynamic production, characterized in that: include: Collecting the time series energy consumption data of discrete manufacturing enterprises and constructing a data set to obtain the energy consumption curve of the discrete manufacturing enterprises; wherein the time series energy consumption data is 1D time series energy consumption data; Rearranging the 1D time series energy consumption data into a 2D matrix using a preset calculation method so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data; The 2D matrix is ​​processed by a preset convolution module, and the weighted sum of the data is calculated, and then the convolution modules are stacked, and residual connections are used to ensure that information is effectively transmitted and aggregated in the deep network; The output result of the convolution module is input into the fully connected layer to generate a prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

2. The energy consumption prediction method for discrete manufacturing enterprises adapted to dynamic production according to claim 1 is characterized in that: The step of rearranging the 1D time series energy consumption data into a 2D matrix by using a preset calculation method so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data includes: Fourier transform is used to analyze the frequency domain of the time series energy consumption data to identify the main cycle in the energy consumption data, and then the first k frequencies whose amplitudes exceed the set value are selected by calculating the frequency spectrum to determine the length of the corresponding cycle; Based on the identified periodic features, the selected k original 1D time series energy consumption data are rearranged into k 2D matrices to capture the dynamic characteristics of short-term and long-term time series energy consumption data respectively; wherein the columns of the matrix represent the energy consumption changes within a cycle, and the rows of the matrix represent the energy consumption changes at the same time points between different cycles.

3. The energy consumption prediction method for discrete manufacturing enterprises adapted to dynamic production according to claim 2 is characterized in that: The step of using Fourier transform to analyze the frequency domain of the time series energy consumption data to identify the main cycle in the energy consumption data, and then selecting the first k frequencies whose amplitudes exceed the set value by calculating the frequency spectrum to determine the length of the corresponding cycle includes: The cycle of the time series energy consumption data is calculated using the formula: S=F ave (F amp (F FFT (X 1D ))) in, T×C represents the time series energy consumption data of enterprises with length T and variable C; F FFT (·) indicates Fourier transform of the time series energy consumption data to convert the time domain features into frequency domain features; F amp (·) indicates the amplitude value in the frequency domain curve of the time series energy consumption; F ave (·) means averaging the amplitudes obtained at different frequencies; argFirstk(·) means selecting the first k frequencies {f1,…f k }, and its corresponding period length is {p1,…p k }, k is a hyperparameter.

4. The energy consumption prediction method for discrete manufacturing enterprises adapted to dynamic production according to claim 3 is characterized in that: The step of rearranging the selected k original 1D time series energy consumption data into k 2D matrices based on the identified periodic features so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data respectively includes: Based on the selected frequencies {f1,…f k } and the corresponding period lengths {p1,…p k }, rearrange the 1D time series energy consumption data into a 2D matrix; where the columns represent the energy consumption changes within a cycle, and the rows represent the energy consumption changes at the same time point in different cycles. The representation of the 2D matrix is ​​shown in the following formula: Among them, F pad (·) indicates that the original time series energy consumption data is filled with 0; F res (·) represents the reconstruction of 1D time series energy consumption data into a 2D matrix according to the cycle information, where f i represents the frequency of occurrence of the i-th frequency in the time series energy consumption data, p i Indicates the length of the cycle.

5. The energy consumption prediction method for discrete manufacturing enterprises adapted to dynamic production according to claim 1 is characterized in that: The steps of using a preset convolution module to process the 2D matrix, performing weighted sum calculation on the data after the data, and then stacking the convolution modules, and using residual connections to ensure effective transmission and aggregation of information in the deep network include: The Inception module is used to convert the processed 2D matrix back to 1D form; wherein the Inception module is a module that uses multi-scale 2D convolution kernels; The calculated cycle amplitude is used as the weight, and the Softmax function is used to perform weighted summation on different 1D time series energy consumption data, and then fused as the output of the current layer; The convolution modules are stacked and residual connections are used to ensure effective transmission and aggregation of information in the deep network.

6. The energy consumption prediction method for discrete manufacturing enterprises adapted to dynamic production according to claim 5 is characterized in that: The step of using the Inception module to convert the processed 2D matrix back to 1D form includes: The Inception module is used to perform convolution on the converted 2D matrix to extract features, and aggregate the changes in the enterprise time series energy consumption data within a cycle and at the same time point between different cycles: Among them, F inception (·) is a deep learning module that processes 2D matrices in parallel using convolution kernels of different sizes to capture the short-term and long-term variation characteristics of time-series energy consumption data; Contains multi-scale periodicity and time-dependent features extracted from time-series energy consumption data. Convert the convolutional 2D matrix to 1D and truncate the previously padded part: Among them, F tru (·) indicates truncation, which is used to remove the filling time point, that is, F pad (·) The 0 added in the middle restores the time series energy consumption data to its original length; Indicates that the processed 2D matrix is ​​converted back into 1D time series energy consumption data.

7. The method for predicting energy consumption of discrete manufacturing enterprises adapted to dynamic production according to claim 6 is characterized in that: The step of taking the calculated period amplitude as a weight, using the Softmax function to perform weighted summation on different 1D time series energy consumption data, and fusing them into the output of the current layer includes: The Softmax function is used to perform weighted fusion on different 1D time series energy consumption data to achieve adaptive aggregation of the hierarchical convolution module: Among them, F softmax (·) Convert the amplitude of each frequency into a weight to indicate the importance of the frequency in the overall feature; It means that the 1D features corresponding to multiple frequencies are aggregated by weighted summation to obtain the output.

8. The method for predicting energy consumption of discrete manufacturing enterprises adapted to dynamic production according to claim 5 is characterized in that: In the step of stacking convolution modules and using residual connections to ensure effective transmission and aggregation of information in the deep network, the calculation formula is: in, represents the output of the convolution module; Φ(·) represents the convolution module; Represents the output of the l-1th.

9. A discrete manufacturing enterprise energy consumption prediction device adapted to dynamic production, characterized in that: include: A collection module, used to collect the time series energy consumption data of discrete manufacturing enterprises and construct a data set to obtain the energy consumption curve of the discrete manufacturing enterprises; wherein the time series energy consumption data is 1D time series energy consumption data; An arrangement module, used for rearranging the 1D time series energy consumption data into a 2D matrix by a preset calculation method, so as to capture the dynamic characteristics of the short-term and long-term time series energy consumption data; The calculation module processes the 2D matrix using a preset convolution module, performs weighted sum calculation on the data, and then stacks the convolution modules, using residual connections to ensure effective transmission and aggregation of information in the deep network; The generation module is used to input the output result of the convolution module into the fully connected layer to generate a prediction result for the next moment, thereby realizing the energy consumption prediction of the discrete manufacturing enterprise.

10. 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, the steps of the method according to any one of claims 1 to 8 are implemented.

Images