Energy Consumption Monitoring Method and Device for Finishing Rolling Process Based on Nonlinear Dynamic Feature Extraction

Abstract

This invention provides a method and device for monitoring energy consumption in the finishing rolling process based on nonlinear dynamic feature extraction, belonging to the field of intelligent monitoring technology for industrial production processes. The method includes: constructing a dataset based on a mutual information calculation method using a finishing rolling process energy consumption dataset to obtain a historical process dataset and energy consumption levels; extracting features from the historical process dataset using a feature extraction network to obtain a set of historical stand energy consumption features; calculating statistical quantities using a support vector data description algorithm based on the historical stand energy consumption feature set to obtain monitoring thresholds for each stand; and calculating statistical quantities based on actual stand energy consumption data to determine whether energy consumption exceeds the standard, and calculating the level of excessive energy consumption to obtain the level of excessive energy consumption. This invention is a precise and efficient distributed energy consumption monitoring method for the finishing rolling process of hot strip steel based on nonlinear dynamic feature extraction.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for industrial production processes, and in particular to a method and device for monitoring energy consumption in finishing rolling processes based on nonlinear dynamic feature extraction. Background Technology

[0002] The steel industry, as an energy and resource-intensive industry, has high carbon emissions. With the continuous advancement of industrial manufacturing and intelligent technologies, energy management and process optimization are becoming increasingly important in modern production. Hot strip rolling is a crucial process in steel production. The motor power of the finishing mill is much higher than that of other equipment, accounting for a significant proportion of the total energy consumption of the production line. It is a key energy-consuming equipment and a process requiring close monitoring throughout the hot strip rolling process. The finishing zone includes seven stands through which slabs are continuously rolled to the required specifications. The finishing rolling process is characterized by complex features such as continuous multi-stage production, the inheritance of energy consumption information between stages, nonlinearity of various influencing factors, and dynamic time series. Maintaining stable energy consumption and promoting energy conservation and reduction during stable production requires effective monitoring of energy consumption in the finishing rolling process.

[0003] Hot strip rolling is a large-scale industrial production process characterized by nonlinear and highly dynamic process variables, as well as complex features such as severe coupling between operating units and strong correlations between process variables. Existing technologies propose dividing large-scale industrial processes into different sub-blocks and developing data models for each sub-block based on its data characteristics. Distributed Principal Component Analysis (PCA), multi-block PCA, and kernel PCA are widely used for monitoring large-scale processes; multi-block PLS and distributed parallel PCA are also applied for monitoring and fault diagnosis. Traditional distributed multivariate statistical process monitoring (MSPM) methods typically rely on data with clear statistical regularities. However, in large-scale industrial processes, data often exhibits complex characteristics such as nonlinearity, strong dynamism, and variable coupling. Therefore, traditional distributed MSPM monitoring methods struggle to deeply explore the latent characteristics and correlations of variables and lack sufficient data representation capabilities for large-scale complex processes. How to efficiently extract effective features from operational data is a crucial issue in energy consumption monitoring and fault diagnosis.

[0004] In the existing technology, there is a lack of an accurate and efficient distributed energy consumption monitoring method for the hot rolling and finishing process of strip steel based on nonlinear dynamic feature extraction. Summary of the Invention

[0005] To address the technical problem of existing technologies failing to quickly locate the cause of anomalies and the position of the mill stand in global monitoring, this invention provides a method and device for monitoring energy consumption in the finishing rolling process based on nonlinear dynamic feature extraction. The technical solution is as follows:

[0006] On the one hand, a method for monitoring energy consumption in the finishing rolling process based on nonlinear dynamic feature extraction is provided. This method is implemented by an energy consumption monitoring device for the finishing rolling process and includes:

[0007] S1. Obtain the energy consumption dataset of the finishing rolling process; Based on the mutual information calculation method, construct the dataset according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level;

[0008] S2. Construct a feature extraction network based on the autoencoder structure and the long short-term memory network structure;

[0009] S3. Based on the historical process dataset, feature extraction is performed through the feature extraction network to obtain a set of historical rack energy consumption features;

[0010] S4. Based on the historical rack energy consumption feature set, use the support vector data description algorithm to calculate statistics and obtain the monitoring statistics of each rack; determine the monitoring statistics of each rack as the monitoring threshold of each rack.

[0011] S5. During the hot rolling and finishing process of strip steel, data is collected from the stand industrial control server to obtain the actual stand energy consumption data; based on the energy consumption level and the monitoring threshold of each stand, the excess energy consumption level is calculated by using the feature extraction network and support vector data description algorithm according to the actual stand energy consumption data, and the excess energy consumption level is obtained.

[0012] On the other hand, a finishing rolling process energy consumption monitoring device based on nonlinear dynamic feature extraction is provided. This device is applied to the finishing rolling process energy consumption monitoring method based on nonlinear dynamic feature extraction. The device includes:

[0013] The dataset construction module is used to obtain the energy consumption dataset of the finishing rolling process; based on the mutual information calculation method, the dataset is constructed according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level;

[0014] The network construction module is used to build a feature extraction network based on the autoencoder structure and the long short-term memory network structure.

[0015] The feature extraction module is used to extract features from the historical process dataset through the feature extraction network to obtain a set of historical rack energy consumption features.

[0016] The monitoring threshold calculation module is used to calculate statistics based on the historical rack energy consumption feature set using the support vector data description algorithm to obtain the monitoring statistics for each rack; and to determine the monitoring statistics for each rack as the monitoring threshold for each rack.

[0017] The excessive energy consumption level monitoring module is used to collect data from the stand industrial control server during the hot rolling and finishing process of strip steel to obtain the actual stand energy consumption data; based on the energy consumption level and the monitoring threshold of each stand, the excessive energy consumption level is calculated by the feature extraction network and support vector data description algorithm according to the actual stand energy consumption data, and the excessive energy consumption level is obtained.

[0018] On the other hand, a finishing rolling process energy consumption monitoring device is provided, the finishing rolling process energy consumption monitoring device comprising: a processor; a memory, the memory storing computer-readable instructions, the computer-readable instructions being executed by the processor to implement any of the above-described finishing rolling process energy consumption monitoring methods based on nonlinear dynamic feature extraction.

[0019] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement any of the above-described methods for monitoring energy consumption in the finishing rolling process based on nonlinear dynamic feature extraction.

[0020] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0021] This invention proposes an energy consumption monitoring method for the finishing rolling process based on nonlinear dynamic feature extraction. It extracts nonlinear and dynamic time-series fusion features from the hot strip finishing rolling process data using a feature fusion extraction model, significantly improving the accuracy of energy consumption exceeding limits monitoring. Considering the impact of energy consumption information transmission between stands, the features extracted from the current stand are processed using a minimum redundancy maximum correlation method to remove redundant features, retaining the most correlated features, and then passed to the next stand, solving the problem of energy consumption information inheritance between series stands. Exceeding energy consumption limits are classified into energy consumption levels, displaying the degree of exceedance, facilitating timely adjustments to the production line status and achieving energy conservation and consumption reduction. This invention is a precise and efficient distributed energy consumption monitoring method for the hot strip finishing rolling process based on nonlinear dynamic feature extraction. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of an energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction provided by an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the production process of hot continuous strip rolling finishing provided by an embodiment of the present invention;

[0025] Figure 3 This is a block diagram of an energy consumption monitoring device for finishing rolling process based on nonlinear dynamic feature extraction provided in an embodiment of the present invention;

[0026] Figure 4 This is a schematic diagram of the structure of an energy consumption monitoring device for the finishing rolling process provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0028] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0029] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0030] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0031] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0032] This invention provides a method for monitoring energy consumption in a finishing rolling process based on nonlinear dynamic feature extraction. This method can be implemented using an energy consumption monitoring device for the finishing rolling process, which can be a terminal or a server. Figure 1 The flowchart shown is for a method of monitoring energy consumption in the finishing rolling process based on nonlinear dynamic feature extraction. The processing flow of this method may include the following steps:

[0033] S1. Obtain the energy consumption dataset of the finishing rolling process; Based on the mutual information calculation method, construct the dataset according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level.

[0034] Optionally, based on the mutual information calculation method, a dataset is constructed from the energy consumption dataset of the finishing rolling process to obtain historical process datasets and energy consumption levels, including:

[0035] S11. Preprocess the energy consumption dataset of the finishing rolling process to obtain the processed stand energy consumption dataset.

[0036] S12. Based on the type of steel grade in the finishing mill, and according to the processed stand energy consumption dataset, calculate the finishing mill deformation energy consumption of the stands in the finishing mill area to obtain the unit deformation energy consumption data of each stand; and determine the unit deformation energy consumption data of each stand as the energy consumption index data.

[0037] S13. Based on the energy consumption index data, use the one-dimensional k-means clustering method to classify the energy consumption levels and obtain the energy consumption levels.

[0038] S14. Based on the energy consumption index data, perform mutual information calculation according to the processed rack energy consumption data to obtain the first mutual information coefficient set.

[0039] S15. Based on a preset coefficient threshold, a second mutual information coefficient set is obtained by filtering according to the first mutual information coefficient set.

[0040] S16. Based on the processed rack energy consumption data, select the energy consumption data corresponding to the second mutual information coefficient set to construct a dataset and obtain the historical process dataset.

[0041] In one feasible implementation, the motor power of the finishing mill in the hot strip rolling process of this invention is much larger than that of other equipment. It accounts for a significant proportion of the total energy consumption of the production line and is a key energy-consuming device in the entire hot strip rolling process. The finishing mill area includes seven stands, through which slabs are continuously rolled to the required specifications, such as... Figure 2 As shown.

[0042] The production process data of the seven stands were collected and classified according to steel grade. The unit deformation energy consumption of each stand was calculated as an energy consumption index variable. In the finishing rolling process, the changes in thickness, length and width of the product during the rolling process have a significant impact on the amount of energy consumption.

[0043] The unit deformation energy consumption is constructed as an energy consumption index data to screen the energy consumption data of rack selection. When the process variables and the characteristics between racks are inherited, the variable with the greatest correlation is retained. The unit deformation energy consumption is calculated as follows (1).

[0044] (1);

[0045] in, For the first Rack power consumption For the first Variation in frame thickness For the first Change in frame length For the first Variation in rack width. When the variation is small, it can be ignored.

[0046] The first mutual information coefficient between the process variable and the energy consumption index variable is calculated. The mathematical expression for this step is as follows: (2)

[0047] (2);

[0048] in, It is represented as the mutual information coefficient between production process variables and energy consumption index variables. It is the joint probability distribution function of X and Y. and X and Y are the marginal probability distribution functions, respectively, where x and y are the sample values ​​of the process variable and energy consumption index variable, respectively.

[0049] Mutual information is used to evaluate the correlation between production process variables and energy consumption index variables. A second mutual information coefficient is selected, and production process variables with a correlation dependence threshold greater than the second mutual information coefficient are selected to form a historical process dataset.

[0050] The energy consumption levels include Level 1 (low energy consumption), Level 2 (high energy consumption), and Level 3 (ultra-high energy consumption).

[0051] In one feasible implementation, the present invention performs one-dimensional k-means clustering on the energy consumption change index, which is the energy consumption per unit deformation. It sets three cluster centers and divides them into three levels, with the energy consumption per unit change volume increasing sequentially. Level 1 is low energy consumption, Level 2 is high energy consumption, Level 3 is ultra-high energy consumption, and Levels 2 and 3 are both above-standard energy consumption.

[0052] S2. Construct a feature extraction network based on the autoencoder structure and the long short-term memory network structure.

[0053] The feature extraction network includes a nonlinear feature extraction module and a dynamic feature extraction module.

[0054] The nonlinear feature extraction module is an unsupervised neural network structure; the nonlinear feature extraction module includes an input layer, hidden layers, and an output layer;

[0055] The dynamic feature extraction module includes an input gate, a forget gate, and an output gate.

[0056] In a feasible implementation manner, the feature extraction network proposed by the present invention includes a non-linear feature extraction module and a dynamic feature extraction module.

[0057] An autoencoder is used to construct the non-linear feature extraction module, which has a symmetric three-layer unsupervised neural network structure, divided into an input layer, a hidden layer (encoder), and an output layer (decoder); the encoder maps the input data into a low-dimensional latent space, and the decoder reconstructs the input data from the representation in this latent space. The training objective of the non-linear feature extraction module is to minimize the difference between the input and the reconstructed output. Usually, the network parameters are adjusted by minimizing the reconstruction error, so that the hidden layer learns the features of the input and obtains the best data representation.

[0058] When the input data is , representing the th rack, the encoder maps the input data into the latent space (l < d), and the mathematical expression of this process is as follows in Equation (3):

[0059] (3);

[0060] where is the first weight matrix of the encoder, is the first bias vector, is the Relu activation function. Through the mapping, the low-dimensional latent representation extracted by the encoder from the input data.

[0061] The decoder then reconstructs the input data from the representation in the latent space, and this process is shown as follows in Equation (4):

[0062]

[0063] (4);

[0063] where is the second weight matrix of the encoder, is the second bias vector, is a preset activation function. The error between the input data and the reconstructed data is expressed as: , and the model parameters are trained by the gradient descent method to extract the non-linear feature .

[0064] A dynamic feature extraction module is constructed using a Long Short-Term Memory (LSTM) network. Input gates, forget gates, and output gates are introduced to enable the network to effectively capture long-term dependencies in time series. Through these gates, the LSTM network can selectively retain or discard information, transmitting key information within the sequence while avoiding interference from outdated information. The calculation formulas for the dynamic feature extraction module are as follows: (5), (6), (7), (8), (9), (10):

[0065] (5);

[0066] (6);

[0067] (7);

[0068] (8);

[0069] (9);

[0070] (10);

[0071] in, , , Representing the first The outputs of the rack forget gate, input gate, and output gate. This indicates the state of candidate cells. For input data, This is the hidden state from the previous moment. It represents the cell state at the previous moment. Here is the weight matrix for the forget gate, input gate, candidate memory units, and output gate. These are the bias factors for the forget gate, input gate, candidate memory units, and output gate. tanh and tanh are activation functions.

[0072] For dynamic time series problems, LSTM receives data from multiple time series through its input layer and applies its gating mechanism at each time step to extract and update dynamic time series features. The features extracted by the network are as follows. The features of each sub-process extracted using an autoencoder (AE) and LSTM are represented as follows: .

[0073] S3. Based on the historical process dataset, feature extraction is performed using a feature extraction network to obtain a set of historical rack energy consumption features.

[0074] Optionally, based on the historical process dataset, features are extracted using a feature extraction network to obtain a set of historical rack energy consumption features, including:

[0075] S31. Based on the historical process dataset, use a feature extraction network to extract features from the first rack to obtain the energy consumption features of the first rack; save the energy consumption features of the first rack to the historical rack energy consumption feature set.

[0076] S32. The number of stands in the finishing rolling zone is 7. The current stand number is b. Let b=2.

[0077] S33. Determine if b is greater than 7. If b is greater than 7, proceed to step S38. If b is less than or equal to 7, proceed to step S34.

[0078] S34. Based on the processed energy consumption data, extract the features of the b-th rack through a feature extraction network to obtain the features of the b-th rack.

[0079] S35. Based on the characteristics of the (b-1)th rack, use the minimum redundancy maximum correlation analysis method to eliminate redundant features and obtain the optimized characteristics of the (b-1)th rack;

[0080] S36. Perform feature fusion between the optimized features of rack b-1 and rack b to obtain the fused rack features;

[0081] S37. Determine the characteristics of the converged rack as the characteristics of the b-th rack and save it to the historical rack energy consumption characteristic set, let b = b + 1; execute step S33;

[0082] S38, Output historical rack energy consumption characteristics set.

[0083] In one feasible implementation, for the features already extracted from the b-th rack, minimum redundancy maximum correlation analysis is used to eliminate redundant features of the energy consumption related features of the current rack, retain the feature with the maximum correlation to the energy consumption index variable of the (b+1)-th rack, and pass the processed features to the next rack for fusion with the features already extracted from the next rack. The objective function used in this process is as follows (11):

[0084] (11);

[0085] in, Representing features and target variables Maximize the mutual information between features and select the feature most relevant to the target. This represents the mutual information between features; minimizing this term removes features with high redundancy. This is achieved by maximizing the objective function. And by obtaining approximately optimal variable characteristics .

[0086] If the characteristics of the previous rack are inherited to the next rack, then the set of energy consumption-related variables for the next rack can be represented as follows: .

[0087] S4. Based on the historical rack energy consumption feature set, use the support vector data description algorithm to calculate the statistics and obtain the monitoring statistics of each rack; determine the monitoring statistics of each rack as the monitoring threshold of each rack.

[0088] In one feasible implementation, in this step, based on the historical normal operating energy consumption characteristics of each rack, support vector description is used to monitor each rack separately, according to the characteristics of each rack. Construct a hypersphere such that normal data points are contained within it, while outlier data points are located outside the hypersphere. Use features extracted from normal sample data. Calculate the radius of a hypersphere As thresholds for monitoring statistics of each rack, for the test samples Features were extracted Calculate the distance of the b-th rack from the center of the hypersphere. The calculation formula for this process is as follows (12):

[0089] (12);

[0090] in, It is the first The supersphere center of each sub-process, It is a Lagrange multiplier.

[0091] The formula for calculating the distance between the test sample and the center of the sphere is as follows (13):

[0092] (13);

[0093] Comparison of sample distances Compared with normal sample radius ,like If the energy consumption is normal, then the energy consumption is normal; otherwise, the energy consumption exceeds the standard.

[0094] S5. During the hot rolling and finishing process of strip steel, data is collected from the industrial control server of the stand to obtain the actual stand energy consumption data; based on the energy consumption level and the monitoring threshold of each stand, the excess energy consumption level is calculated by using a feature extraction network and support vector data description algorithm according to the actual stand energy consumption data, and the excess energy consumption level is obtained.

[0095] Optionally, based on the energy consumption level and the monitoring threshold of each rack, and according to the actual rack energy consumption data, the excessive energy consumption level is calculated using a feature extraction network and a support vector data description algorithm to obtain the excessive energy consumption level, including:

[0096] S51. Based on the actual rack energy consumption data, use a feature extraction network to extract features and obtain the actual rack energy consumption feature set.

[0097] S52. Based on the actual rack energy consumption characteristic set, use the support vector data description algorithm to calculate the statistics and obtain the actual monitoring statistics of each rack.

[0098] S53. Based on the monitoring threshold of each rack, verify the actual monitoring statistics of each rack and obtain the actual monitoring verification results.

[0099] S54. Based on the actual rack energy consumption feature set, select the excessive energy consumption features to construct the actual rack energy consumption excess feature set according to the actual monitoring and verification results;

[0100] S55. Based on the energy consumption level, calculate the energy consumption level exceeding the standard according to the actual rack energy consumption exceeding the standard characteristic set, and obtain the energy consumption level exceeding the standard.

[0101] In one feasible implementation, based on the monitoring thresholds for each rack and the actual rack energy consumption data, the feature extraction network and support vector data description algorithm are used to determine whether the energy consumption exceeds the standard. For energy consumption exceeding the statistical threshold, the level of energy consumption exceeding the standard is calculated to determine the degree of energy consumption exceeding the standard. The features of the energy consumption exceeding the standard are then analyzed. Calculate its energy consumption level probability. The classification formula is as follows (14):

[0102] (14);

[0103] in, Represents energy consumption level. It is the first The radius of the hypersphere described by the class sample support vectors. Is the sample up to the first The distance from the center of the hypersphere of the sample class is given by exp, which is an exponential operation. If If the value is within the specified threshold, it is classified as Level 2 energy consumption exceeding the standard; otherwise, it is classified as Level 3 energy consumption exceeding the standard. Production data samples are collected in real time, and their monitoring statistics are calculated. When the statistics exceed the threshold, it is determined that energy consumption is exceeding the standard in current production, and the level of energy consumption exceeding the standard is calculated.

[0104] This invention proposes an energy consumption monitoring method for the finishing rolling process based on nonlinear dynamic feature extraction. It extracts nonlinear and dynamic time-series fusion features from the hot strip finishing rolling process data using a feature fusion extraction model, significantly improving the accuracy of energy consumption exceeding limits monitoring. Considering the impact of energy consumption information transmission between stands, the features extracted from the current stand are processed using a minimum redundancy maximum correlation method to remove redundant features, retaining the most correlated features, and then passed to the next stand, solving the problem of energy consumption information inheritance between series stands. Exceeding energy consumption limits are classified into energy consumption levels, displaying the degree of exceedance, facilitating timely adjustments to the production line status and achieving energy conservation and consumption reduction. This invention is a precise and efficient distributed energy consumption monitoring method for the hot strip finishing rolling process based on nonlinear dynamic feature extraction.

[0105] Figure 3 This is a block diagram illustrating an energy consumption monitoring device for a finishing rolling process based on nonlinear dynamic feature extraction, according to an exemplary embodiment. The device is used in a finishing rolling process energy consumption monitoring method based on nonlinear dynamic feature extraction. (Refer to...) Figure 3 The device includes a dataset construction module 310, a network construction module 320, a feature extraction module 330, a monitoring threshold calculation module 340, and an excessive energy consumption level monitoring module 350. Among them:

[0106] The dataset construction module 310 is used to obtain the energy consumption dataset of the finishing rolling process; based on the mutual information calculation method, the dataset is constructed according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level;

[0107] Network building module 320 is used to build a feature extraction network based on the autoencoder structure and the long short-term memory network structure;

[0108] Feature extraction module 330 is used to extract features from historical process datasets through a feature extraction network to obtain a set of historical rack energy consumption features;

[0109] The monitoring threshold calculation module 340 is used to calculate statistics based on the historical rack energy consumption feature set using the support vector data description algorithm to obtain the monitoring statistics of each rack; and to determine the monitoring statistics of each rack as the monitoring threshold of each rack.

[0110] The Excess Energy Consumption Level Monitoring Module 350 is used to collect data from the stand industrial control server during the hot rolling and finishing process of strip steel to obtain the actual stand energy consumption data. Based on the energy consumption level and the monitoring threshold of each stand, the module calculates the excess energy consumption level through a feature extraction network and a support vector data description algorithm according to the actual stand energy consumption data.

[0111] Optionally, the dataset building module 310 is further used for:

[0112] S11. Preprocess the energy consumption dataset of the finishing rolling process to obtain the processed stand energy consumption dataset.

[0113] S12. Based on the type of steel grade in the finishing mill, and according to the processed stand energy consumption dataset, calculate the finishing mill deformation energy consumption of the stands in the finishing mill area to obtain the unit deformation energy consumption data of each stand; and determine the unit deformation energy consumption data of each stand as the energy consumption index data.

[0114] S13. Based on the energy consumption index data, use the one-dimensional k-means clustering method to classify the energy consumption levels and obtain the energy consumption levels.

[0115] S14. Based on the energy consumption index data, perform mutual information calculation according to the processed rack energy consumption data to obtain the first mutual information coefficient set.

[0116] S15. Based on a preset coefficient threshold, a second mutual information coefficient set is obtained by filtering according to the first mutual information coefficient set.

[0117] S16. Based on the processed rack energy consumption data, select the energy consumption data corresponding to the second mutual information coefficient set to construct a dataset and obtain the historical process dataset.

[0118] The energy consumption levels include Level 1 (low energy consumption), Level 2 (high energy consumption), and Level 3 (ultra-high energy consumption).

[0119] The feature extraction network includes a nonlinear feature extraction module and a dynamic feature extraction module.

[0120] The nonlinear feature extraction module is an unsupervised neural network structure; the nonlinear feature extraction module includes an input layer, hidden layers, and an output layer;

[0121] The dynamic feature extraction module includes an input gate, a forget gate, and an output gate.

[0122] Optionally, the feature extraction module 330 is further used for:

[0123] S31. Based on the historical process dataset, use a feature extraction network to extract features from the first rack to obtain the energy consumption features of the first rack; save the energy consumption features of the first rack to the historical rack energy consumption feature set.

[0124] S32. The number of stands in the finishing rolling zone is 7. The current stand number is b. Let b=2.

[0125] S33. Determine if b is greater than 7. If b is greater than 7, proceed to step S38. If b is less than or equal to 7, proceed to step S34.

[0126] S34. Based on the processed energy consumption data, extract the features of the b-th rack through a feature extraction network to obtain the features of the b-th rack.

[0127] S35. Based on the characteristics of the (b-1)th rack, use the minimum redundancy maximum correlation analysis method to eliminate redundant features and obtain the optimized characteristics of the (b-1)th rack;

[0128] S36. Perform feature fusion between the optimized features of rack b-1 and rack b to obtain the fused rack features;

[0129] S37. Determine the characteristics of the converged rack as the characteristics of the b-th rack and save it to the historical rack energy consumption characteristic set, let b = b + 1; execute step S33;

[0130] S38, Output historical rack energy consumption characteristics set.

[0131] Optionally, the energy consumption level monitoring module 350 can be further used for:

[0132] S51. Based on the actual rack energy consumption data, use a feature extraction network to extract features and obtain the actual rack energy consumption feature set.

[0133] S52. Based on the actual rack energy consumption characteristic set, use the support vector data description algorithm to calculate the statistics and obtain the actual monitoring statistics of each rack.

[0134] S53. Based on the monitoring threshold of each rack, verify the actual monitoring statistics of each rack and obtain the actual monitoring verification results.

[0135] S54. Based on the actual rack energy consumption feature set, select the excessive energy consumption features to construct the actual rack energy consumption excess feature set according to the actual monitoring and verification results;

[0136] S55. Based on the energy consumption level, calculate the energy consumption level exceeding the standard according to the actual rack energy consumption exceeding the standard characteristic set, and obtain the energy consumption level exceeding the standard.

[0137] This invention proposes an energy consumption monitoring method for the finishing rolling process based on nonlinear dynamic feature extraction. It extracts nonlinear and dynamic time-series fusion features from the hot strip finishing rolling process data using a feature fusion extraction model, significantly improving the accuracy of energy consumption exceeding limits monitoring. Considering the impact of energy consumption information transmission between stands, the features extracted from the current stand are processed using a minimum redundancy maximum correlation method to remove redundant features, retaining the most correlated features, and then passed to the next stand, solving the problem of energy consumption information inheritance between series stands. Exceeding energy consumption limits are classified into energy consumption levels, displaying the degree of exceedance, facilitating timely adjustments to the production line status and achieving energy conservation and consumption reduction. This invention is a precise and efficient distributed energy consumption monitoring method for the hot strip finishing rolling process based on nonlinear dynamic feature extraction.

[0138] Figure 4This is a schematic diagram of the structure of an energy consumption monitoring device for the finishing rolling process provided in an embodiment of the present invention, as shown below. Figure 4 As shown, the energy consumption monitoring equipment for the finishing rolling process may include the above-mentioned Figure 3 The illustrated device is an energy consumption monitoring device for the finishing rolling process based on nonlinear dynamic feature extraction. Optionally, the energy consumption monitoring device 410 for the finishing rolling process may include a first processor 2001.

[0139] Optionally, the finishing rolling process energy consumption monitoring device 410 may also include a memory 2002 and a transceiver 2003.

[0140] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0141] The following is combined with Figure 4 A detailed introduction to each component of the finishing rolling process energy consumption monitoring equipment 410 is provided below:

[0142] The first processor 2001 is the control center of the finishing rolling process energy consumption monitoring device 410. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0143] Optionally, the first processor 2001 can perform various functions of the finishing rolling process energy consumption monitoring device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0144] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 4 CPU0 and CPU1 are shown in the diagram.

[0145] In a specific implementation, as one example, the energy consumption monitoring device 410 for the finishing rolling process may also include multiple processors, for example... Figure 4The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0146] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0147] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently, and may be connected via the interface circuit of the finishing mill process energy consumption monitoring device 410. Figure 4 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0148] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0149] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 4 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0150] Optionally, the transceiver 2003 can be integrated with the first processor 2001, or it can exist independently and be connected to the interface circuit of the finishing rolling process energy consumption monitoring device 410. Figure 4 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0151] It should be noted that, Figure 4 The structure of the finishing rolling process energy consumption monitoring device 410 shown does not constitute a limitation on the router. The actual knowledge structure identification device may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0152] Furthermore, the technical effect of the finishing rolling process energy consumption monitoring device 410 can be referred to the technical effect of the finishing rolling process energy consumption monitoring method based on nonlinear dynamic feature extraction described in the above method embodiments, and will not be repeated here.

[0153] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0154] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0155] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0156] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0157] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0158] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0159] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0161] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0162] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0164] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0165] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for monitoring energy consumption in a finishing rolling process based on nonlinear dynamic feature extraction, characterized in that, The method includes: S1. Obtain the energy consumption dataset of the finishing rolling process; based on the mutual information calculation method, construct the dataset according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level; S2. Construct a feature extraction network based on the autoencoder structure and the long short-term memory network structure; S3. Based on the historical process dataset, feature extraction is performed through the feature extraction network to obtain a set of historical rack energy consumption features; S4. Based on the historical rack energy consumption feature set, use the support vector data description algorithm to calculate statistics and obtain the monitoring statistics of each rack; determine the monitoring statistics of each rack as the monitoring threshold of each rack. S5. During the hot rolling and finishing process of strip steel, data is collected from the stand industrial control server to obtain the actual stand energy consumption data; based on the energy consumption level and the monitoring threshold of each stand, the excess energy consumption level is calculated by using the feature extraction network and support vector data description algorithm according to the actual stand energy consumption data, and the excess energy consumption level is obtained.

2. The energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction according to claim 1, characterized in that, The mutual information-based calculation method constructs a dataset based on the finishing rolling process energy consumption dataset to obtain historical process datasets and energy consumption levels, including: S11. Preprocess the energy consumption dataset of the finishing rolling process to obtain the processed stand energy consumption dataset. S12. Based on the type of steel grade in the finishing mill, and according to the processed stand energy consumption dataset, calculate the finishing mill deformation energy consumption of the stands in the finishing mill area to obtain the unit deformation energy consumption data of each stand; and determine the unit deformation energy consumption data of each stand as energy consumption index data. S13. Based on the energy consumption index data, use the one-dimensional k-means clustering method to classify energy consumption levels and obtain energy consumption levels. S14. Based on the energy consumption index data, perform mutual information calculation according to the processed rack energy consumption data to obtain a first set of mutual information coefficients. S15. Based on a preset coefficient threshold, filter according to the first mutual information coefficient set to obtain a second mutual information coefficient set; S16. Based on the processed rack energy consumption data, select the energy consumption data corresponding to the second mutual information coefficient set to construct a dataset and obtain a historical process dataset.

3. The energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction according to claim 1, characterized in that, The energy consumption levels include Level 1 (low energy consumption), Level 2 (high energy consumption), and Level 3 (ultra-high energy consumption).

4. The energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction according to claim 1, characterized in that, The feature extraction network includes a nonlinear feature extraction module and a dynamic feature extraction module; The nonlinear feature extraction module is an unsupervised neural network structure; the nonlinear feature extraction module includes an input layer, a hidden layer, and an output layer; The dynamic feature extraction module includes an input gate, a forget gate, and an output gate.

5. The energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction according to claim 2, characterized in that, The step of extracting features from the historical process dataset using the feature extraction network to obtain a set of historical rack energy consumption features includes: S31. Based on the historical process dataset, use the feature extraction network to extract features of the first rack to obtain the energy consumption features of the first rack; save the energy consumption features of the first rack to the historical rack energy consumption feature set; S32. The number of stands in the finishing rolling zone is 7. The current stand number is b. Let b=2. S33. Determine if b is greater than 7. If b is greater than 7, proceed to step S38. If b is less than or equal to 7, proceed to step S34. S34. Based on the processed energy consumption data, the features of the b-th rack are extracted through the feature extraction network to obtain the features of the b-th rack. S35. Based on the characteristics of the (b-1)th rack, use the minimum redundancy maximum correlation analysis method to eliminate redundant features and obtain the optimized characteristics of the (b-1)th rack. S36. The optimized features of the (b-1)th rack are fused with the features of the bth rack to obtain the fused rack features; S37. Determine the fusion rack feature as the b-th rack feature and save it to the historical rack energy consumption feature set, let b = b + 1; execute step S33; S38. Output the set of historical rack energy consumption characteristics.

6. The energy consumption monitoring method for finishing rolling process based on nonlinear dynamic feature extraction according to claim 1, characterized in that, Based on the energy consumption level and the monitoring thresholds for each rack, and according to the actual rack energy consumption data, the excess energy consumption level is calculated using the feature extraction network and support vector data description algorithm to obtain the excess energy consumption level, including: S51. Based on the actual rack energy consumption data, use a feature extraction network to extract features and obtain an actual rack energy consumption feature set. S52. Based on the actual rack energy consumption feature set, use the support vector data description algorithm to calculate the statistics and obtain the actual monitoring statistics of each rack. S53. Based on the monitoring threshold of each rack, verify the actual monitoring statistics of each rack to obtain the actual monitoring verification results; S54. Based on the actual rack energy consumption feature set, and according to the actual monitoring and verification results, select the excessive energy consumption features to construct the actual rack energy consumption excess feature set; S55. Based on the energy consumption level, calculate the excess energy consumption level according to the actual rack energy consumption excess feature set to obtain the excess energy consumption level.

7. A finishing rolling process energy consumption monitoring device based on nonlinear dynamic feature extraction, wherein the finishing rolling process energy consumption monitoring device based on nonlinear dynamic feature extraction is used to implement the finishing rolling process energy consumption monitoring method based on nonlinear dynamic feature extraction as described in any one of claims 1-6, characterized in that, The device includes: The dataset construction module is used to obtain the energy consumption dataset of the finishing rolling process; based on the mutual information calculation method, the dataset is constructed according to the energy consumption dataset of the finishing rolling process to obtain the historical process dataset and energy consumption level; The network construction module is used to build a feature extraction network based on the autoencoder structure and the long short-term memory network structure. The feature extraction module is used to extract features from the historical process dataset through the feature extraction network to obtain a set of historical rack energy consumption features. The monitoring threshold calculation module is used to calculate statistics based on the historical rack energy consumption feature set using the support vector data description algorithm to obtain the monitoring statistics for each rack; and to determine the monitoring statistics for each rack as the monitoring threshold for each rack. The excessive energy consumption level monitoring module is used to collect data from the stand industrial control server during the hot rolling and finishing process of strip steel to obtain the actual stand energy consumption data; based on the energy consumption level and the monitoring threshold of each stand, the excessive energy consumption level is calculated by the feature extraction network and support vector data description algorithm according to the actual stand energy consumption data, and the excessive energy consumption level is obtained.

8. The energy consumption monitoring device for finishing rolling process based on nonlinear dynamic feature extraction according to claim 7, characterized in that, The dataset construction module is further used for: S11. Preprocess the energy consumption dataset of the finishing rolling process to obtain the processed stand energy consumption dataset. S12. Based on the type of steel grade in the finishing mill, and according to the processed stand energy consumption dataset, calculate the finishing mill deformation energy consumption of the stands in the finishing mill area to obtain the unit deformation energy consumption data of each stand. The energy consumption data per unit deformation of each rack is determined as energy consumption index data; S13. Based on the energy consumption index data, use the one-dimensional k-means clustering method to classify energy consumption levels and obtain energy consumption levels. S14. Based on the energy consumption index data, perform mutual information calculation according to the processed rack energy consumption data to obtain a first set of mutual information coefficients. S15. Based on a preset coefficient threshold, filter according to the first mutual information coefficient set to obtain a second mutual information coefficient set; S16. Based on the processed rack energy consumption data, select the energy consumption data corresponding to the second mutual information coefficient set to construct a dataset and obtain a historical process dataset.

9. An energy consumption monitoring device for a finishing rolling process, characterized in that, The energy consumption monitoring equipment for the finishing rolling process includes: processor; A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 6.

Images