Industrial scene energy consumption prediction method and device, equipment and storage medium

Abstract

The invention discloses an industrial scene energy consumption prediction method and device, equipment and a storage medium, relates to the field of energy consumption prediction, is applied to a preset energy consumption prediction system, and comprises the following steps: screening connection functions according to historical industrial scene data obtained from a local intranet server; respectively determining a Spearman correlation coefficient between each variable and energy consumption in the historical industrial scene data and a weight coefficient respectively corresponding to each variable through a target connection function obtained after screening and an analytic hierarchy process, and carrying out variable screening; carrying out preprocessing and data division on the obtained screened data; and based on the target training set, the target verification set and the target test set obtained after division, performing model training, verification and test on an initial energy consumption prediction model pre-constructed based on the deep belief network to obtain a target energy consumption prediction model so as to perform energy consumption prediction by using the target energy consumption prediction model based on the external network server. And the model prediction precision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of energy consumption prediction, and in particular to an industrial scene energy consumption prediction method, device, equipment and storage medium. Background Art

[0002] With the rapid development of the Internet of Things and machine learning technologies, energy consumption prediction in industrial scenarios has gradually attracted the attention of more and more professionals. However, existing solutions directly build models for prediction by collecting massive amounts of data (including production data, equipment status, and energy prices, which may contain a large amount of sensitive data). However, due to the complexity and variability of the industrial production process, as well as factors such as equipment aging, process adjustments, and raw material changes that may affect energy consumption, directly building a prediction model with massive data will not be able to guarantee the prediction accuracy of the model, which will affect the user experience. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide an industrial scene energy consumption prediction method, device, equipment and storage medium, which can effectively improve the model prediction accuracy, thereby improving the accuracy of energy consumption prediction, and further improving user experience. The specific scheme is as follows:

[0004] In a first aspect, the present application provides an industrial scenario energy consumption prediction method for presetting an energy consumption prediction system, comprising:

[0005] Filter each connection function according to the preset function screening rules and the historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function;

[0006] Determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and perform variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data;

[0007] Performing corresponding preprocessing and data division on the screened data to obtain corresponding target training set, target verification set and target test set;

[0008] Based on the target training set, the target verification set and the target test set, the initial energy consumption prediction model pre-constructed based on the deep belief network is trained, verified and tested to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on an external network server and using the target energy consumption prediction model.

[0009] Optionally, the preset energy consumption prediction system also includes a network firewall for securely isolating the intranet server and the external network server through its own embedded data channel control system; wherein the intranet server is a server connected to a trusted network or domain, and the external network server is a server connected to a non-trusted network or domain.

[0010] Optionally, the energy consumption prediction based on the external network server and using the target energy consumption prediction model includes:

[0011] Collect data of the current industrial scene through the local intranet server to obtain corresponding scene data;

[0012] The scene data is transferred to the external network server through the network gateway, so that the external network server performs energy consumption prediction based on the received scene data and using the target energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0013] Optionally, the method of screening each connection function according to a preset function screening rule and historical industrial scene data obtained from a local intranet server to obtain a corresponding target connection function includes:

[0014] Get the pre-configured target experience connection function;

[0015] Inputting historical industrial scene data obtained from a local intranet server into each connection function and a target experience connection function respectively, so as to obtain function value information corresponding to each connection function and the target experience connection function respectively;

[0016] Based on the function value information, the Euclidean distance information between each of the connection functions and the target empirical connection function is determined respectively, so as to determine the corresponding target connection function according to the size relationship between each of the Euclidean distance information.

[0017] Optionally, the target connection function and the analytic hierarchy process are used to respectively determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable, and variable screening is performed based on the Spearman correlation coefficient and the weight coefficient, including:

[0018] After determining the Spearman correlation coefficient between each variable in the historical industrial scene data and energy consumption based on the target connection function, constructing a corresponding initial analysis structure model based on each variable in the historical industrial scene data;

[0019] Performing a corresponding consistency check on the initial analysis structure model based on a preset check rule, and obtaining a corresponding target analysis structure model when the obtained check result indicates that the check passes;

[0020] Determine the weight coefficients corresponding to the variables based on the target analysis structural model, and use the corresponding weight coefficients to weight the corresponding Spearman correlation coefficients to obtain weighted correlation coefficients corresponding to the variables;

[0021] Variable screening is performed using each of the weighted correlation coefficients and based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data.

[0022] Optionally, the performing model training, verification and testing on the initial energy consumption prediction model pre-constructed based on the deep belief network based on the target training set, the target verification set and the target test set includes:

[0023] Obtain an initial energy consumption prediction model pre-built based on a deep belief network;

[0024] After completing the hyperparameter configuration operation corresponding to the initial energy consumption prediction model through the grid optimization method, the initial energy consumption prediction model is trained, verified, and model parameters are adjusted based on the target training set and the target verification set until the preset iterative training conditions are met to obtain the corresponding trained model;

[0025] The trained model is tested based on the target test set and the preset model evaluation index, and when the obtained model test result indicates that the test passes, the corresponding target energy consumption prediction model is determined.

[0026] Optionally, the performing model training, model verification, and model parameter adjustment on the initial energy consumption prediction model based on the target training set and the verification set includes:

[0027] In the process of model training the initial energy consumption prediction model, unsupervised training and supervised training are performed on each restricted Boltzmann machine network and back propagation network in the initial energy consumption prediction model based on the target training set, and weights and biases are updated based on the target verification set to complete the corresponding parameter adjustment operation.

[0028] In a second aspect, the present application provides an industrial scene energy consumption prediction device, which is applied to a preset energy consumption prediction system, including:

[0029] A function screening module is used to screen each connection function according to preset function screening rules and historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function;

[0030] A variable screening module, used to determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and perform variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data;

[0031] A data processing module is used to perform corresponding preprocessing and data division on the screened data to obtain corresponding target training sets, target verification sets and target test sets;

[0032] The energy consumption prediction model is used to perform model training, verification and testing on the initial energy consumption prediction model pre-constructed based on the deep belief network based on the target training set, the target verification set and the target test set to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on the external network server and using the target energy consumption prediction model.

[0033] In a third aspect, the present application provides an electronic device, including:

[0034] Memory, used to store computer programs;

[0035] The processor is used to execute the computer program to implement the steps of the aforementioned industrial scenario energy consumption prediction method.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the steps of the aforementioned industrial scenario energy consumption prediction method.

[0037] It can be seen that in this application, by presetting the energy consumption prediction system, firstly, each connection function is screened according to the preset function screening rules and the historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function; the Spearman correlation coefficient between each variable and energy consumption in the historical industrial scene data and the weight coefficient corresponding to each variable are determined by the target connection function and the hierarchical analysis method, and the variable is screened based on the Spearman correlation coefficient and the weight coefficient to obtain the corresponding screened data; the screened data is preprocessed and divided accordingly to obtain the corresponding target training set, target verification set and target test set; based on the target training set, the target verification set and the target test set, the initial energy consumption prediction model pre-built based on the deep belief network is trained, verified and tested to obtain the corresponding target energy consumption prediction model, so as to predict energy consumption based on the external network server and using the target energy consumption prediction model. That is, in this application, the variables in the historical industrial scene data are carefully selected by the connection function and the hierarchical analysis method, and then the obtained screened data is used to train the initial energy consumption prediction model pre-built based on the deep belief network. In this way, the model prediction accuracy can be effectively improved, thereby improving the accuracy of energy consumption prediction and further improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0039] Figure 1 A flow chart of an industrial scenario energy consumption prediction method provided for this application;

[0040] Figure 2 A schematic diagram of the architecture of a preset energy consumption prediction system provided for this application;

[0041] Figure 3 A schematic diagram of a specific industrial scenario energy consumption prediction method provided for this application;

[0042] Figure 4 A schematic diagram of model training for an initial energy consumption prediction model provided in this application;

[0043] Figure 5 A schematic diagram of the structure of an industrial scene energy consumption prediction device provided in this application;

[0044] Figure 6 A structural diagram of an electronic device provided for this application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Existing solutions directly build models for prediction by collecting massive amounts of data (including production data, equipment status, and energy prices, which may contain a large amount of sensitive data). However, due to the complexity and variability of industrial production processes, as well as factors such as equipment aging, process adjustments, and changes in raw materials that may affect energy consumption, directly building prediction models with massive data will not be able to guarantee the prediction accuracy of the model, which will affect the user experience. To this end, this application provides an industrial scenario energy consumption prediction solution that can effectively improve the model prediction accuracy, thereby improving the accuracy of energy consumption predictions, and thus improving the user experience.

[0047] See also Figure 1 As shown, an embodiment of the present invention discloses an industrial scene energy consumption prediction method, which is applied to a preset energy consumption prediction system, including:

[0048] Step S11: Filter each connection function according to preset function filtering rules and historical industrial scene data obtained from a local intranet server to obtain a corresponding target connection function.

[0049] Specifically, Figure 2As shown, in this embodiment, regarding the preset energy consumption prediction system, the preset energy consumption prediction system includes a network gate for securely isolating the intranet server and the extranet server through its own embedded data channel control system; wherein the intranet server is a server connected to a trusted network or domain, including a corresponding intranet isolation switching module (the module has a corresponding switching chip 2), which is used for data collection and data storage; the extranet server is a server connected to an untrusted network or domain, including a corresponding extranet isolation switching module (the module has a corresponding switching chip 1), which is used for energy consumption prediction. In addition, the intranet server and the extranet server can communicate securely through the network gate. During this process, the network gate uses a unique control logic and a dedicated communication protocol to completely control the real-time exchange of data, ensure the disconnection of any connection between the trusted network (domain) and the untrusted network (domain), and completely block the TCP / IP (Transmission Control Protocol / Internet Protocol) protocol and other network protocols. In this way, the "2+1" architecture is adopted in hardware, that is, through the design of dual servers plus network firewalls, secure isolation between trusted networks and untrusted networks can be achieved, thereby ensuring the security of data during collection, storage and transmission, and preventing unauthorized access and tampering.

[0050] In this embodiment, combined with Figure 3 As shown, in order to improve the accuracy of energy consumption prediction, the correlation of various influencing factors is fully considered when performing energy consumption prediction in industrial scenarios to select accurate characteristic variables. Specifically, it is necessary to first screen the target connection function, that is, in this embodiment, the historical data of each influencing factor variable related to energy consumption prediction is input into the alternative Copula function (i.e., the connection function) to obtain the alternative Copula function value. Among them, there are 5 types of Copula functions involved in this embodiment: Gaussian-Copula, t-Copula, Gumbel-Copula, Clayton-Copula and Frank-Copula.

[0051] Afterwards, by defining , compare the Euclidean distance between each candidate Copula function and the defined empirical Copula function (i.e., the target empirical connection function) to achieve the optimization of the Copula function. The empirical Copula function defined in this embodiment As follows: Let x i ,y i (i=1,2,…,n) are sample data taken from variables X and Y respectively, F x (x), F y(y) are the empirical distribution functions of X and Y respectively, then:

[0052] ;

[0053] In the formula, I (if) For example, when the if condition is true, I (if) =1 if the value is set to 0, otherwise it is 0.

[0054] The Euclidean distance d between the empirical Copula function and each candidate Copula function j The formula is as follows:

[0055] ;

[0056] In the formula, a i For x i The corresponding empirical distribution function value F x (x i );b i for y i The corresponding empirical distribution function value F y (y i );C j (a i ,b i ) are the five candidate Copula functions, j=[0,1,2,3,4,5]. The candidate Copula function corresponding to the minimum Euclidean distance is the optimal Copula function, that is, the target connection function.

[0057] Step S12, determining the Spearman correlation coefficient between each variable in the historical industrial scene data and energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and performing variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data.

[0058] Furthermore, in this embodiment, after obtaining the target connection function, after determining the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption based on the target connection function, a corresponding initial analysis structure model is constructed based on each variable in the historical industrial scene data; the initial analysis structure model is subjected to a corresponding consistency check based on a preset check rule, and when the check result obtained indicates that the check is passed, a corresponding target analysis structure model is obtained; based on the target analysis structure model, weight coefficients corresponding to each of the variables are determined, and the corresponding Spearman correlation coefficients are weighted using the corresponding weight coefficients to obtain weighted correlation coefficients corresponding to each of the variables; variables are screened using the weighted correlation coefficients and based on the Spearman correlation coefficients and the weight coefficients to obtain corresponding screened data.

[0059] Specifically, in this embodiment, the target link function is used to calculate the Spearman rank correlation coefficient (ie, the Spearman correlation coefficient) ζ to characterize the correlation between variables X and Y. The calculation formula is as follows:

[0060] ;

[0061] Where a and b are the marginal distribution functions G(x) and H(y) of variables X and Y respectively; d is the differential sign.

[0062] Afterwards, combined Figure 3 As shown in the figure, the Analytic Hierarchy Process (AHP) is used to calculate the weight coefficient σ of each variable. ij , where AHP can decompose the problem into different components according to the nature of the problem and the overall goal to be achieved, and cluster and combine the factors at different levels according to the interrelated influences and affiliations between the factors to form a multi-level analysis structure model, so that the problem can be ultimately reduced to the determination of the relative importance of the lowest level (schemes and measures for decision-making, etc.) relative to the highest level (overall goal) or the ranking of the relative order of superiority and inferiority. Therefore, when using AHP to calculate the weight coefficient:

[0063] (1) Clarify the program objectives and build a hierarchical structure. The structure is divided into the goal layer, the criterion layer and the program layer, and each layer is composed of several factors and indicators.

[0064] (2) Construct a judgment matrix. Each indicator in each layer can be compared with an indicator in the previous layer as a criterion. When an indicator in the previous layer is used as an evaluation criterion, the judgment matrix (where A represents the judgment matrix; n represents the number of rows and columns in the matrix, and n*n here means that the matrix is ​​a square matrix) ijIndicates the importance of the i-th element in this layer relative to the j-th factor, a ij Generally, positive integers from 1 to 9 and their reciprocals are used.

[0065] (3) Find the judgment matrix The maximum eigenvalue λ max And the corresponding eigenvector ω, the ω' after normalizing ω is the weight of the level indicator.

[0066] (4) Consistency test. First calculate the consistency index M C1 , ; Then check the average random consistency index M according to the order n R1 ; Finally, calculate the consistency ratio , if N CR When <0.10, the consistency of the judgment matrix is ​​considered acceptable, otherwise it should be appropriately modified.

[0067] (5) By combining the weights of each level, we can finally get the weights of each solution level factor on the target level and the criterion level.

[0068] In the present application, when using the obtained weighted correlation coefficient for screening, a threshold can be set in advance for screening, and variables greater than or equal to the threshold are determined as target variables, and the variable data corresponding to the target variable in the historical industrial scene data is retained to obtain the screened data, which will be used as input features of the model for training in subsequent steps.

[0069] Step S13: performing corresponding preprocessing and data division on the screened data to obtain corresponding target training set, target verification set and target test set.

[0070] In this embodiment, the screened data obtained after variable screening needs to be preprocessed, including but not limited to data cleaning, normalization and labeling. After that, the preprocessed screened data is divided to obtain the corresponding target training set, target validation set and target test set.

[0071] Step S14, based on the target training set, the target verification set and the target test set, the initial energy consumption prediction model pre-constructed based on the deep belief network is trained, verified and tested to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on the external network server and using the target energy consumption prediction model.

[0072] Specifically, in this embodiment, the initial energy consumption prediction model based on DBN (Deep Belief Networks) is composed of multiple restricted Boltzmann machines (RBM) and a single-layer back propagation (BP) network. After that, the remaining hyperparameters are determined by the grid optimization method, and then the network model is trained until the preset number of iterations is reached.

[0073] It is important to understand that, combined with Figure 4 As shown, in the process of model training, after initializing the model parameters, the present embodiment first trains each layer of RBM network unsupervisedly so that the feature vector is mapped to different feature spaces, thereby retaining as much feature information as possible; then, a BP network is set in the last layer of DBN to receive the output feature vector of RBM as its input feature vector, and supervised training of the entity relationship classifier is performed. That is, in the process of model training of the initial energy consumption prediction model, unsupervised training and supervised training are performed on each restricted Boltzmann machine network and the back propagation network in the initial energy consumption prediction model based on the target training set, and weights and biases are updated based on the target verification set to complete the corresponding parameter adjustment operation. Among them, each layer of RBM network can only ensure that the weights in its own layer are optimally mapped to the feature vector of that layer, not the feature vector of the entire DBN, so the back propagation network also propagates the error information from top to bottom to each layer of RBM to fine-tune the entire DBN network. The process of RBM network training model can be regarded as the initialization of the weight parameters of a deep BP network, which enables DBN to overcome the shortcomings of BP network that it is easy to fall into local optimality and has a long training time due to random initialization of weight parameters.

[0074] It can be understood that, in this embodiment, regarding the determination of the target energy consumption prediction model, first obtain the initial energy consumption prediction model pre-built based on the deep belief network; then after completing the hyperparameter configuration operation corresponding to the initial energy consumption prediction model through the grid optimization method, the initial energy consumption prediction model is trained, verified, and the model parameters are adjusted based on the target training set and the target verification set until the preset iterative training conditions are met to obtain the corresponding trained model; then the trained model is tested based on the target test set and the preset model evaluation index, and when the obtained model test result indicates that the test passes, the corresponding target energy consumption prediction model is determined. Among them, the original low-dimensional features in the target training set are converted into high-dimensional features layer by layer through multiple hidden layers, so that the model learns the implicit mapping relationship. The Adam optimization algorithm is used to input the feature quantity in the target verification set into the trained DBN energy consumption prediction model, and then the output energy consumption prediction result is compared with the true value and the loss function is calculated to adjust the model parameters generation by generation according to the loss function. Finally, the mean squared error (MSE), root mean square error (RMSE) and mean absolute error (MAE) are used as model evaluation indicators. When the evaluation results meet expectations, the target energy consumption prediction model can be used to perform energy consumption prediction operations related to industrial scenarios.

[0075] In summary, this embodiment proposes an industrial scenario energy consumption prediction method based on data security. It can be divided into two aspects: energy consumption prediction and data security. In terms of energy consumption prediction, the energy consumption data correlation analysis method based on AHP-Copula theory is first used to calculate the weighted Spearman correlation coefficient of each variable; then the weighted correlation coefficient is used as the basis for selecting the input features of the energy consumption prediction model DBN; finally, the historical data set of the input features is preprocessed and input into the energy consumption prediction model DBN, and network structure hyperparameter selection, network model training, network parameter tuning, and model performance evaluation are performed to obtain energy consumption prediction results. In terms of data security, the "2+1" architecture is adopted in hardware, that is, the design of dual servers and network firewalls is used to achieve secure isolation between trusted networks and untrusted networks.

[0076] It can be seen that in the embodiment of the present application, by presetting the energy consumption prediction system, firstly, each connection function is screened according to the preset function screening rules and the historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function; the Spearman correlation coefficient between each variable and energy consumption in the historical industrial scene data and the weight coefficient corresponding to each variable are determined by the target connection function and the hierarchical analysis method, and the variable is screened based on the Spearman correlation coefficient and the weight coefficient to obtain the corresponding screened data; the screened data is preprocessed and divided accordingly to obtain the corresponding target training set, target verification set and target test set; the initial energy consumption prediction model pre-built based on the deep belief network is trained, verified and tested based on the target training set, the target verification set and the target test set to obtain the corresponding target energy consumption prediction model, so as to predict energy consumption based on the external network server and using the target energy consumption prediction model. That is, in the present application, the variables in the historical industrial scene data are carefully selected by the connection function and the hierarchical analysis method, and then the obtained screened data is used to train the initial energy consumption prediction model pre-built based on the deep belief network. In this way, the model prediction accuracy can be effectively improved, thereby improving the accuracy of energy consumption prediction and further improving user experience.

[0077] See also Figure 5 As shown, the embodiment of the present application also discloses an industrial scene energy consumption prediction device, which is applied to a preset energy consumption prediction system, including:

[0078] The function screening module 11 is used to screen each connection function according to preset function screening rules and historical industrial scene data obtained from the local intranet server to obtain a corresponding target connection function;

[0079] A variable screening module 12 is used to determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and perform variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data;

[0080] The data processing module 13 is used to perform corresponding preprocessing and data division on the screened data to obtain corresponding target training sets, target verification sets and target test sets;

[0081] The energy consumption prediction model 14 is used to perform model training, verification and testing on the initial energy consumption prediction model pre-constructed based on the deep belief network based on the target training set, the target verification set and the target test set to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on the external network server and using the target energy consumption prediction model.

[0082] Among them, for more specific working processes of the above-mentioned modules, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0083] It can be seen that in the embodiment of the present application, the variables in the historical industrial scene data are carefully selected through the connection function and the hierarchical analysis method, and then the obtained filtered data is used to train the initial energy consumption prediction model pre-built based on the deep belief network. In this way, the model prediction accuracy can be effectively improved, thereby improving the accuracy of energy consumption prediction, and thus improving the user experience.

[0084] In some specific embodiments, the energy consumption prediction model 14 may specifically include:

[0085] A data acquisition unit, used to collect data on the current industrial scene through the local intranet server to obtain corresponding scene data;

[0086] The prediction result acquisition unit is used to transfer the scene data to the external network server through the network gate, so that the external network server performs energy consumption prediction based on the received scene data and using the target energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0087] In some specific embodiments, the function screening module 11 may specifically include:

[0088] A function acquisition unit, used to acquire a pre-configured target experience connection function;

[0089] A function value acquisition unit, used to input the historical industrial scene data acquired from the local intranet server into each connection function and the target experience connection function, respectively, to obtain function value information corresponding to each of the connection functions and the target experience connection function;

[0090] The target function determination unit is used to determine the Euclidean distance information between each of the connection functions and the target empirical connection function based on the function value information, so as to determine the corresponding target connection function according to the size relationship between each of the Euclidean distance information.

[0091] In some specific embodiments, the variable screening module 12 may specifically include:

[0092] An initial analysis structure model building unit, configured to build a corresponding initial analysis structure model based on each variable in the historical industrial scene data after determining the Spearman correlation coefficient between each variable in the historical industrial scene data and energy consumption based on the target connection function;

[0093] A target analysis structure model acquisition unit is used to perform a corresponding consistency check on the initial analysis structure model based on a preset check rule, and obtain a corresponding target analysis structure model when the obtained check result indicates that the check is passed;

[0094] A coefficient weighting unit, used for determining the weight coefficients corresponding to the variables based on the target analysis structure model, and weighting the corresponding Spearman correlation coefficients using the corresponding weight coefficients to obtain weighted correlation coefficients corresponding to the variables;

[0095] The screened data acquisition unit is used to perform variable screening based on the Spearman correlation coefficient and the weight coefficient using each of the weighted correlation coefficients to obtain corresponding screened data.

[0096] In some specific embodiments, the energy consumption prediction module 14 may specifically include:

[0097] An initial energy consumption prediction model acquisition unit, used to acquire an initial energy consumption prediction model pre-built based on a deep belief network;

[0098] A post-training model acquisition unit is used to perform model training, model verification, and model parameter adjustment on the initial energy consumption prediction model based on the target training set and the target verification set after completing the hyperparameter configuration operation corresponding to the initial energy consumption prediction model through the grid optimization method, until the preset iterative training conditions are met, so as to obtain the corresponding post-training model;

[0099] The target energy consumption prediction model acquisition unit is used to test the trained model based on the target test set and the preset model evaluation index, and when the obtained model test result indicates that the test passes, determine the corresponding target energy consumption prediction model.

[0100] In some specific embodiments, the energy consumption prediction module 14 may specifically include:

[0101] A model training unit is used to perform unsupervised training and supervised training on each restricted Boltzmann machine network and back propagation network in the initial energy consumption prediction model based on the target training set during the model training of the initial energy consumption prediction model, and to update the weights and biases based on the target verification set to complete the corresponding parameter adjustment operation.

[0102] Furthermore, the present application also discloses an electronic device. Figure 6 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content in the diagram cannot be regarded as any limitation on the scope of use of the present application.

[0103] Figure 6 A schematic diagram of the structure of an electronic device 20 provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the industrial scene energy consumption prediction method disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0104] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0105] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0106] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, which can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program that can be used to complete the industrial scene energy consumption prediction method performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program that can be used to complete other specific tasks.

[0107] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the aforementioned industrial scene energy consumption prediction method is implemented. The specific steps of the method can refer to the corresponding contents disclosed in the aforementioned embodiments, and will not be repeated here.

[0108] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0109] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0110] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0111] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0112] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for predicting energy consumption in industrial scenarios, characterized in that: Applied to preset energy consumption prediction system, including: Filter each connection function according to the preset function screening rules and the historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function; Determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and perform variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data; Performing corresponding preprocessing and data division on the screened data to obtain corresponding target training set, target verification set and target test set; Based on the target training set, the target verification set and the target test set, the initial energy consumption prediction model pre-constructed based on the deep belief network is trained, verified and tested to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on an external network server and using the target energy consumption prediction model.

2. The industrial scene energy consumption prediction method according to claim 1 is characterized in that: The preset energy consumption prediction system also includes a network firewall for securely isolating the intranet server and the external network server through its own embedded data channel control system; wherein the intranet server is a server connected to a trusted network or domain, and the external network server is a server connected to an untrusted network or domain.

3. The industrial scene energy consumption prediction method according to claim 2 is characterized in that: The energy consumption prediction based on the external network server and using the target energy consumption prediction model includes: Collect data of the current industrial scene through the local intranet server to obtain corresponding scene data; The scene data is transferred to the external network server through the network gateway, so that the external network server performs energy consumption prediction based on the received scene data and using the target energy consumption prediction model to obtain corresponding energy consumption prediction results.

4. The industrial scene energy consumption prediction method according to claim 1 is characterized in that: The method of screening each connection function according to the preset function screening rules and the historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function includes: Get the pre-configured target experience connection function; Inputting historical industrial scene data obtained from a local intranet server into each connection function and a target experience connection function respectively, so as to obtain function value information corresponding to each connection function and the target experience connection function respectively; Based on the function value information, the Euclidean distance information between each of the connection functions and the target empirical connection function is determined respectively, so as to determine the corresponding target connection function according to the size relationship between each of the Euclidean distance information.

5. The industrial scene energy consumption prediction method according to claim 1 is characterized in that: The target connection function and the analytic hierarchy process are used to respectively determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable, and variable screening is performed based on the Spearman correlation coefficient and the weight coefficient, including: After determining the Spearman correlation coefficient between each variable in the historical industrial scene data and energy consumption based on the target connection function, constructing a corresponding initial analysis structure model based on each variable in the historical industrial scene data; Performing a corresponding consistency check on the initial analysis structure model based on a preset check rule, and obtaining a corresponding target analysis structure model when the obtained check result indicates that the check passes; Determine the weight coefficients corresponding to the variables based on the target analysis structural model, and use the corresponding weight coefficients to weight the corresponding Spearman correlation coefficients to obtain weighted correlation coefficients corresponding to the variables; Variable screening is performed using each of the weighted correlation coefficients and based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data.

6. The industrial scene energy consumption prediction method according to any one of claims 1 to 5, characterized in that: The performing model training, verification and testing on the initial energy consumption prediction model pre-built based on the deep belief network based on the target training set, the target verification set and the target test set includes: Obtain an initial energy consumption prediction model pre-built based on a deep belief network; After completing the hyperparameter configuration operation corresponding to the initial energy consumption prediction model through the grid optimization method, the initial energy consumption prediction model is trained, verified, and model parameters are adjusted based on the target training set and the target verification set until the preset iterative training conditions are met to obtain the corresponding trained model; The trained model is tested based on the target test set and the preset model evaluation index, and when the obtained model test result indicates that the test passes, the corresponding target energy consumption prediction model is determined.

7. The industrial scene energy consumption prediction method according to claim 6 is characterized in that: The performing model training, model verification, and model parameter adjustment on the initial energy consumption prediction model based on the target training set and the verification set includes: In the process of model training the initial energy consumption prediction model, unsupervised training and supervised training are performed on each restricted Boltzmann machine network and back propagation network in the initial energy consumption prediction model based on the target training set, and weights and biases are updated based on the target verification set to complete the corresponding parameter adjustment operation.

8. An industrial scene energy consumption prediction device, characterized in that: Applied to preset energy consumption prediction system, including: A function screening module is used to screen each connection function according to preset function screening rules and historical industrial scene data obtained from the local intranet server to obtain the corresponding target connection function; A variable screening module, used to determine the Spearman correlation coefficient between each variable in the historical industrial scene data and the energy consumption and the weight coefficient corresponding to each variable through the target connection function and the hierarchical analysis method, and perform variable screening based on the Spearman correlation coefficient and the weight coefficient to obtain corresponding screened data; A data processing module is used to perform corresponding preprocessing and data division on the screened data to obtain corresponding target training sets, target verification sets and target test sets; The energy consumption prediction model is used to perform model training, verification and testing on the initial energy consumption prediction model pre-constructed based on the deep belief network based on the target training set, the target verification set and the target test set to obtain the corresponding target energy consumption prediction model, so as to perform energy consumption prediction based on the external network server and using the target energy consumption prediction model.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the industrial scenario energy consumption prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, which, when executed by a processor, implements the industrial scenario energy consumption prediction method as described in any one of claims 1 to 7.

Images