Data management and performance prediction method and device for hot rolling process of strip steel based on cloud-edge collaboration

Through cloud-edge collaboration, data governance and performance prediction of strip hot rolling process are achieved, and the problems of low data utilization value and insufficient process evaluation in the existing technology are solved, and prediction accuracy and data asset value are improved. Digital and intelligent transformation.

CN118797274BActive Publication Date: 2024-11-15ANSTEEL AUTOMAION CO +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411263330.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-15
Estimated Expiration
2044-09-10

AI Technical Summary

Technical Problem

The existing technology cannot effectively achieve data management and high-precision prediction of strip hot rolling production process data, resulting in low data utilization value, insufficient process evaluation, and lack of comprehensive digital and intelligent transformation support.

Method used

Using cloud-edge collaboration method, detect outliers through isolated forest algorithms, robust normalization and partially robust regression processing data, establish a global metadata view with six-dimensional data models, filter auxiliary variables with knowledge of mechanisms, and build time-space volumes The network model is accumulated and comprehensive evaluation is used to achieve interaction and performance prediction of multi-source heterogeneous data.

Benefits of technology

The data management and performance prediction of the hot rolling process of strip steel is realized, the prediction accuracy is improved, and valuable data assets are formed, which supports digital and intelligent transformation, reduces the frequency of inspection and testing, brings economic profits, energy conservation and consumption reduction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118797274B_ABST
    Figure CN118797274B_ABST
Patent Text Reader

Abstract

The present invention discloses a cloud-edge collaborative strip hot rolling process data governance and performance prediction method and device, the method comprising: obtaining strip hot rolling production process data for rapid detection of outliers; performing non-ideal data expansion on cloud-side data; using a multi-source heterogeneous data description paradigm of a six-dimensional data model to establish a global view of the strip hot rolling process metadata; combining the knowledge of the strip hot rolling process mechanism and mRMR to screen performance-related auxiliary variables; constructing an entity embedding model to achieve the interaction and fusion of the multi-source heterogeneous data; constructing a time-space convolutional network soft sensor model on the edge; and using a multi-performance index comprehensive evaluation description method based on the entropy weight method on the cloud side to fuse local prediction results and complete the performance evaluation of the whole process of strip hot rolling. The technical problem that the existing technology cannot achieve data governance of the strip hot rolling production process, and cannot effectively achieve high-precision performance prediction and full-process evaluation is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring of industrial operating systems, and in particular to a cloud-edge collaborative strip hot rolling process data management and performance prediction method and device. Background Art

[0002] The hot strip rolling process is a fully automated production line with the characteristics of high speed, high efficiency, high quality, multi-level, multi-process, and multi-condition. When the performance of the strip product is abnormal, if the performance abnormality can be detected quickly, and the abnormal process or even the equipment can be accurately diagnosed, and the operator intervenes and controls in time, unnecessary economic losses, personal safety and other problems can be avoided. At the same time, a large amount of structured, semi-structured and unstructured heterogeneous coupled data will be generated throughout the life cycle of the hot strip rolling process. The data quality is unbalanced, the key production factor data is difficult to perceive in real time, and the utilization rate of multi-source heterogeneous data resources is low.

[0003] Under the wave of transformation and upgrading of the manufacturing industry, the structure, function, process flow, control environment and evaluation indicators of traditional manufacturing systems are becoming increasingly complex, evolving into new complex product manufacturing processes. The corresponding digital representation, factor modeling, production control and production line optimization are key points for the future core competitiveness of the manufacturing industry, and have attracted great attention and long-term exploration from academia and industry. Data governance is the most important basic technology for digital manufacturing. In recent years, domestic and foreign scholars have made phased progress in the research of digital representation and performance modeling, production prediction and optimization. Multi-dimensional feature perception, multi-source heterogeneous data fusion, and knowledge extraction and expression at the data level have become the focus of digital governance; data-driven performance analysis and correlation modeling for industrial process decision-making, optimization, performance prediction, control and other applications are still in their infancy.

[0004] At present, the research on data governance in the hot strip rolling process mainly focuses on the basic characterization of data, and there is no detailed model or system to manage multi-source data. In terms of performance prediction, soft sensor modeling methods are mainly divided into mechanism-based, data-based, and mechanism-and-data hybrid-driven modeling methods. However, the traditional data governance of the hot strip rolling process is not systematic, the value of data utilization is shallow, the research on full-process performance prediction and evaluation is relatively small, and there are almost no application cases. Relevant technologies are urgently needed to support the digital transformation of the steel manufacturing process.

[0005] Therefore, there is a need in the existing technology to improve the data governance and performance prediction methods of the hot rolling process of strip steel with cloud-edge collaboration. Summary of the invention

[0006] In view of this, the purpose of an embodiment of the present invention is to propose a cloud-edge collaborative strip hot rolling process data management and performance prediction method and device to solve the technical problem that the existing technology cannot realize strip hot rolling production process data management, and cannot effectively realize high-precision performance prediction and full-process evaluation.

[0007] Based on the above purpose, an embodiment of the present invention provides a cloud-edge collaborative strip hot rolling process data management and performance prediction method, including:

[0008] The S1 cloud-side server obtains the production process data of hot-rolled strip steel and uses the isolation forest algorithm to quickly detect outliers;

[0009] S2 performs non-ideal data expansion on cloud-side data based on non-ideal data processing methods based on robust normalization and partial robust regression;

[0010] S3 uses the multi-source heterogeneous data description paradigm of the six-dimensional data model to establish a global view of the metadata of the strip hot rolling process, and saves the data of the entire line and the entire life cycle to the cloud-side server to achieve data governance;

[0011] S4 combines the knowledge of the hot rolling process mechanism of strip steel and the maximum relevance minimum redundancy algorithm to screen the performance-related auxiliary variables;

[0012] S5 builds an entity embedding model and processes unstructured and semi-structured production data to achieve interaction and fusion of the multi-source heterogeneous data;

[0013] S6 constructs a time-space convolutional network soft sensor model on the edge, uses the data synchronized by the cloud side to achieve local performance prediction, and feeds the result back to the cloud side;

[0014] S7 uses a multi-performance index comprehensive evaluation description method based on the entropy weight method to fuse local prediction results on the cloud side to complete the performance evaluation of the entire process of hot rolling of strip steel.

[0015] In some embodiments, in S1, using the isolation forest algorithm to quickly detect outliers includes:

[0016] The production process data includes three levels of data: L1, L2, and L3. The isolation forest algorithm is used to quickly detect data outliers on the three levels of data, build a binary tree, and specify the cutting dimension and cutting point P;

[0017] Repeat the classification and placement of sample points in the training sample set until there is only one data in the node or the maximum height of the tree reaches the limit;

[0018] Loop through the new binary trees to generate An isolated tree iTree is constructed, multiple isolated trees are constructed, the average path length of each isolated tree is calculated, the path length of the data point in each tree is calculated to determine the isolated value, outliers are quickly detected based on the set threshold and the isolated value, and the outliers are marked.

[0019] In some implementations, in S2, performing non-ideal data expansion on cloud-side data based on a non-ideal data processing method of robust normalization and partial robust regression includes:

[0020] S21 normalizes the original data on the cloud side to eliminate the influence of different dimensions;

[0021] S22 uses robust normalization to estimate the mean and variance of the measured data;

[0022] S23 uses a partial robust M regression algorithm to further process the strip hot rolling process and quality data to eliminate outliers.

[0023] In some implementations, in S3, using a multi-source heterogeneous data description paradigm of a six-dimensional data model to establish a global view of the metadata of the hot rolling process of the strip steel includes:

[0024] Constructing a four-layer metamodel of the steel rolling process, and establishing a six-dimensional data model for the steel rolling process data based on the four-layer metamodel structure, wherein the six-dimensional data model includes a time domain T, a space domain S, an entity object domain I, a relationship domain R, an attribute P, and an attribute value V;

[0025] On the basis of the six-dimensional data model, the multi-level information space mapping relationship from physical levels such as process and equipment to comprehensive production indicators, operation indicators and control indicators is combined to complete the unified description of multi-source heterogeneous data of the hot rolling process of the strip steel;

[0026] Metadata is used to construct a metadata structure model. According to the data management and service flow of the steel rolling process, a strip hot rolling data metadata system is constructed by dividing it into technical metadata, management metadata, and business metadata based on application scenarios.

[0027] In some embodiments, in S4, combining the knowledge of the hot rolling process mechanism of the steel strip and the maximum relevance minimum redundancy algorithm to screen the performance-related auxiliary variables includes:

[0028] Based on the maximum correlation minimum redundancy algorithm as the correlation measurement basis, the features associated with the performance variables are screened according to the production process of hot rolling of strip steel, and the initial feature subsets of each process are obtained. , calculated as follows:

[0029]

[0030]

[0031] in, is the feature set, is the number of features in the feature set, is the performance target variable, Features and the target variable The mutual information between The features and target variables in the feature set The mean mutual information between Features and Features The mutual information between is the mutual information between each feature in the feature set;

[0032] Combining the maximum correlation and minimum redundancy results in the minimum redundancy maximum correlation criterion, which is calculated as follows:

[0033]

[0034] in, It is the minimum redundancy maximum correlation measurement indicator function;

[0035] Analyze process data using knowledge of the hot rolling process mechanism of strip steel, screen out variables related to performance, and form a knowledge feature selection subset, wherein the process data includes raw material composition and switch quantity;

[0036] Based on the soft sensor auxiliary variable screening of data and knowledge fusion, the data feature subsets preliminarily selected by the maximum correlation and minimum redundancy algorithms are Knowledge feature subsets with knowledge-based filtering Fusion is performed to obtain the final auxiliary variable .

[0037] In some implementations, in S5, processing the unstructured and semi-structured production data to achieve the interaction and fusion of the multi-source heterogeneous data includes:

[0038] One-hot encoding is used for encoding. The one-hot encoding of is expressed as follows:

[0039]

[0040] in, Indicates The one-hot encoding of the natural state, Is the length The vector of Expressed as The number of values ​​of is the mapping symbol;

[0041] Given input The output of the additional linear neuron is defined as follows:

[0042]

[0043] in, is the number of neurons in the one-hot encoding layer, The weights connecting each neuron in the one-hot encoding layer, For the Layer Coding Layer The one-hot encoding representation of is the weight connecting the one-hot encoding layer to the embedding layer, is the index of the embedding layer;

[0044] After the entity embedding represents all categorical features, the inputs of all embedding layers and all continuous variables are concatenated, where the dimension of the embedding layer is The hyperparameters that need to be defined by themselves are converted into dense variables of fixed size after two fully connected layers, and finally combined with the auxiliary variables through the merging layer as the input data of the soft measurement model.

[0045] In some embodiments, in S6, constructing a time-space convolutional network (TSTN) soft sensor model on the side includes:

[0046] Build a time-space convolutional neural network model, including time convolution layer, space convolution layer, pooling layer, and fully connected layer. By combining time and space convolution, the time and space information is fused together to effectively extract the time and space features of the data;

[0047] The soft sensor model is trained and deployed to the edge server. The auxiliary variable data is obtained from the metadata database and input into the constructed time-space convolutional network neural network for training. The loss function of the model training is:

[0048]

[0049] in, is the actual value of performance, is the predicted value of the soft sensor model, are the parameters of the model;

[0050] The edge server obtains the online real-time metadata of the hot rolling process of the strip from the cloud side, and inputs the selected auxiliary variables into the trained time-space convolutional network soft measurement model on the edge to complete the real-time performance prediction.

[0051] In some embodiments, the temporal convolution layer extracts temporal features through a one-dimensional convolution operation to process time series data, the spatial convolution layer captures spatial features through a two-dimensional convolution operation to process structural data between different processes, and the pooling layer is used for dimensionality reduction and extraction of significant features.

[0052] In some implementations, in S7, fusing the local prediction results using a multi-performance index comprehensive evaluation description method based on an entropy weight method on the cloud side includes:

[0053] A comprehensive evaluation description method for multiple performance indicators based on the entropy weight method. The entropy weight method can determine the importance of indicators according to the amount of information of different indicators, so as to obtain the evaluation results by weighting multiple indicators. Information entropy represents the degree of variation of data under a certain indicator, which is expressed as:

[0054]

[0055] Among them, n represents a total of n indicators. Indicated in Among the indicators The proportion of performance index values ​​to the total sample is expressed as:

[0056]

[0057] The weight is expressed as:

[0058]

[0059] in Indicates the number of performance indicators, No. The information entropy of the performance indicators, the final performance score can be expressed as:

[0060]

[0061] in, Indicates performance indicator weights, Indicates The performance indicator weights are used to visualize the final performance evaluation score and the performance prediction results of each process on the cloud side and the edge side based on the Django framework.

[0062] On the other hand, the present invention also provides a cloud-edge collaborative strip hot rolling process data management and performance prediction device, which is used to implement the above method, including:

[0063] A cloud-side acquisition module configured to acquire strip hot rolling production data;

[0064] A cloud-side data governance module is configured to construct a metadata panorama of hot-rolled strip steel. The data governance module includes a data outlier detection module, a non-ideal data expansion module, a metadata construction module, a metadata system construction module, and a data information description expression solution module;

[0065] The edge input module is configured to obtain cloud-side production data and input it into the constructed soft measurement model;

[0066] A side output module configured to obtain a process performance prediction result according to the cloud-side production data and the side soft measurement model;

[0067] The cloud-side output module is configured to fuse the sub-process performance prediction results and output the final performance evaluation results.

[0068] The present invention has at least the following beneficial technical effects:

[0069] (1) The proposed six-dimensional data model, the multi-source heterogeneous data description paradigm, the constructed hot strip rolling metadata database, and the established data information description expression scheme will provide a complete set of data governance technologies for hot strip rolling, form valuable data assets for hot strip rolling, and at the same time serve as a digital base to promote the digital and intelligent transformation of hot strip rolling.

[0070] (2) The established time-space convolutional network soft sensor model can capture the characteristics of the time series and process of hot rolling production of strip steel, greatly improving the accuracy of performance prediction. It can replace destructive offline tests such as mechanical properties testing, reduce the frequency of inspections and tests, bring direct economic benefits, and achieve energy saving and consumption reduction.

[0071] (3) The “cloud-edge” collaborative mechanism is taken into consideration, which allows real-time monitoring of the product performance of sub-processes and the status of the entire production process, providing information support for operators. Based on performance prediction, the cause of the failure can be traced, and production can be intervened in a timely manner to achieve quality and efficiency improvements. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] 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 in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

[0073] Figure 1 It is a schematic diagram of the execution flow of the "cloud-edge" collaborative strip hot rolling process data management and performance prediction and evaluation method provided by an embodiment of the present invention;

[0074] Figure 2 It is the production flow chart of hot rolling of strip steel;

[0075] Figure 3 It is a steel rolling process metadata model diagram provided by an embodiment of the present invention;

[0076] Figure 4 It is an example diagram of the knowledge description of the finishing mill process and its attribute outlet thickness provided by an embodiment of the present invention;

[0077] Figure 5 The steel rolling process metadata database provided by the embodiment of the present invention;

[0078] Figure 6 It is a structural schematic diagram of a "cloud-edge" collaborative strip hot rolling process data management and performance prediction and evaluation device provided by an embodiment of the present invention;

[0079] Figure 7 It is a system block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0080] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0081] First of all, it should be noted that in the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "exemplarily" is intended to present the concept in a concrete way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0082] Moreover, in the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0083] In addition, in the embodiments of the present invention, sometimes a subscript (such as W1) may be expressed in a non-subscript form (such as W1). When the difference is not emphasized, the meanings to be expressed are consistent.

[0084] This embodiment provides a cloud-edge collaborative strip hot rolling process data management and performance prediction and evaluation method to achieve data management and performance prediction and evaluation of the strip hot rolling production process, which can not only improve product quality, but also provide operation guidance for production process personnel. The method can be implemented by an electronic device, which can be a terminal or a server.

[0085] The execution flow of this method is as follows Figure 1 As shown, the following steps are included:

[0086] The S1 cloud-side server obtains the production process data of hot-rolled strip steel and uses the isolation forest algorithm to quickly detect outliers;

[0087] S2 performs non-ideal data expansion on cloud-side data based on non-ideal data processing methods based on robust normalization and partial robust regression;

[0088] S3 uses the multi-source heterogeneous data description paradigm of the six-dimensional data model to establish a global view of the metadata of the strip hot rolling process, and saves the data of the entire line and the entire life cycle to the cloud-side server to achieve data governance;

[0089] S4 combines the knowledge of the hot rolling process mechanism of strip steel and the maximum relevant minimum redundancy algorithm (mRMR) to screen the performance-related auxiliary variables;

[0090] S5 builds an entity embedding model and processes unstructured and semi-structured production data to achieve interaction and fusion of the multi-source heterogeneous data;

[0091] S6 constructs a time-space convolutional network (TSCN) soft sensor model on the edge, uses the data synchronized by the cloud side to achieve local performance prediction, and feeds the result back to the cloud side;

[0092] S7 uses a multi-performance index comprehensive evaluation description method based on the entropy weight method to fuse local prediction results on the cloud side to complete the performance evaluation of the entire process of hot rolling of strip steel.

[0093] The method steps are further described below in conjunction with specific embodiments.

[0094] The S1 cloud-side server obtains the production process data of hot-rolled strip steel and uses the isolation forest algorithm to quickly detect outliers;

[0095] Specifically, in this embodiment, the outlier rapid detection process includes:

[0096] Use the isolation forest algorithm to quickly detect data outliers. Build a binary tree, specify the cutting dimension and cutting point P;

[0097] Place the sample points in the training sample set that are greater than or equal to P in the right node of the binary tree, and place the samples that are less than or equal to P in the left node;

[0098] Recursively repeat the first two steps until there is only one data in the node or the maximum height of the tree reaches the limit;

[0099] Repeat the previous step, select a new binary tree, and generate An isolated tree iTree is constructed, and multiple isolated trees are finally constructed;

[0100] Compute the average path length for each isolated tree:

[0101]

[0102] in, is a harmonic number, defined as , is the total number of data points in the isolated tree. The following formula is used to calculate the path length of the data point in each tree to determine the isolated value (abnormal score). The shorter the path, the easier it is for the data point to be isolated, that is, the greater the possibility of an abnormal value:

[0103]

[0104] in, It is a data point The average path length in all trees, is the sample size of the training data, is the average path length of the tree, which is used for normalization. Finally, based on the calculated anomaly score, a threshold is set to determine whether a data point is an outlier and the outlier is marked.

[0105] S2 performs non-ideal data expansion on the cloud-side data based on a non-ideal data processing method of robust normalization and partial robust regression; specifically, in this embodiment, the non-ideal data preprocessing combining robust normalization and partial robust M regression includes:

[0106] S21: Normalization of raw data. The presence of outliers will have an adverse effect on subsequent modeling. Raw data should be normalized to eliminate the influence of different dimensions. Generally, the normalization process can be defined as:

[0107]

[0108] in, and Respectively represent the measurement data If the measured data contains outliers, their mean and variance will change significantly and deviate from the true value.

[0109] S22: Robust normalization estimates mean and variance. To solve the problem described in S31, the mean and variance are estimated by robust normalization. The mean value in robust normalization is represented by the total square loss center and the robust center, in the following form:

[0110]

[0111] Among them, the variance value can be calculate, . represents an integer function, is the correction factor. Can be The data is sorted in ascending order. By multiplying Therefore, the mean and variance The desired result is robust normalization.

[0112] S23: Partially robust M regression algorithm (PRMR) eliminates outliers. The hot rolling process and quality data of the strip are and , further processed using the PRMR algorithm to remove outliers. and It can be obtained, among which It is a diagonal weight matrix, which is implemented as follows:

[0113] Use process and quality data to replace U respectively, and use the mean and variance obtained by robust normalization to standardize and obtain process data and quality data .

[0114] Initialize weights , calculate the leverage weight ,use Instead of the score vector , calculate the residual weight , , and finally calculate the comprehensive weight .

[0115] use Respectively and Perform weighted calculations, establish regression relationships for weighted data based on the PLS algorithm, and calculate sample residuals , update the comprehensive weight .

[0116] If the relative difference between two consecutive calculations is less than the threshold (set in this section ), the iteration is considered to converge, otherwise it returns to the weight initialization step. The final comprehensive weight , weighted to get new data and .

[0117] S3 uses the multi-source heterogeneous data description paradigm of the six-dimensional data model to establish a global view of the metadata of the strip hot rolling process, and saves the data of the entire line and the entire life cycle to the cloud-side server to achieve data governance;

[0118] Specifically, the description method of the six-dimensional data model multi-source heterogeneous data description paradigm includes:

[0119] S31: Data metamodel construction and multi-source heterogeneous data description paradigm. Construct a four-layer metamodel of the steel rolling process, such as Figure 3 As shown, they are meta-metamodel layer, metamodel layer, model layer, and instance layer, from top to bottom, from abstract to concrete, and from top layer to bottom layer implementation.

[0120] Based on the four-layer metamodel structure, a six-dimensional data model (time domain T, space domain S, entity object domain I, relationship domain R, attribute P, attribute value V) for steel rolling process data is proposed. On the basis of the data metamodels of each system, each process, and each stage, the multi-level information space mapping relationship from the physical level of process, equipment, etc. to the comprehensive production indicators, operation indicators, and control indicators is considered to complete the quantitative, formal, and logical unified description of multi-source heterogeneous data at multiple levels (multi-stage-multi-process-multi-system) of the hot rolling process of strip steel. Through the description of the six-dimensional data model, the mutual conversion of structured, semi-structured, and unstructured data and the unified description of information are finally realized. Taking the knowledge description of the finishing mill process and its attribute outlet thickness as an example, Figure 4 shown.

[0121] S32: Construct a metadata system for hot-rolled strip data. Use metadata to construct a metadata structure model. Based on the data management and service process of the steel rolling process, divide the steel rolling process metadata system into technical metadata, management metadata, and business metadata according to the application scenario. Metadata of various functions form an internally interrelated metadata system.

[0122] Among them, technical metadata refers to all data involved in the information supply chain throughout the entire process of the product life cycle, mainly including data source metadata, storage metadata, calculation metadata, operation metadata, sharing metadata, service metadata, process metadata, etc., such as the execution status, execution time, execution results, and metadata generated during the execution of data ETL tasks and query analysis. Management metadata mainly refers to the description of the management attributes of data, including the description of information on the processing, archiving, structure, technical processing, access control, management department, management responsible person, and related systems of digital objects. Clarifying management attributes is conducive to implementing data management responsibilities to departments and individuals, and is the basis of data security management. Business metadata refers to user-oriented metadata that describes technical metadata and related business, including data sources, types, connotations, relationships, constraints and meanings, as well as related business rules, terms and attributes, etc., mainly including model metadata, application metadata, analysis metadata, etc. Various metadata are interrelated and collaborative to achieve a unified description of data.

[0123] The steel rolling process metadata database based on the unified metadata standard can achieve effective data management, solve the problem of unified steel rolling process data information and format, complete the establishment of the strip hot rolling production process metadata database, and provide a data basis for subsequent performance prediction. The metadata framework is as follows Figure 5 shown.

[0124] S33: Construct a data information description and expression scheme to achieve "cloud-side" data governance. Establish a set of standard steel rolling process data information description and expression schemes on the "cloud side" to achieve the circulation and sharing of data information among various systems based on the establishment of a comprehensive steel rolling process metadata model. Construct a metadata structure model based on the multi-level, multi-scale, and difficult-to-unify characteristics of steel rolling process data. Each model layer corresponds to different information description formats and instance characteristics.

[0125] S4, combining the knowledge of the hot rolling process mechanism of strip steel and the maximum relevance minimum redundancy algorithm mRMR to screen the performance-related auxiliary variables;

[0126] Specifically, in this embodiment, the variables highly correlated with the performance are extracted as auxiliary variables by combining the knowledge of the hot rolling process mechanism of the strip steel and the maximum relevance minimum redundancy mRMR, including:

[0127] S41: Taking the maximum relevant minimum redundancy mRMR as the correlation metric, the features associated with the performance variables are screened according to the production process of hot rolling of strip steel, and the initial feature subsets of each process are obtained. , that is, The feature subset of process data is mainly for float and int data in the hot rolling process of strip steel, and the calculation is as follows:

[0128]

[0129]

[0130] in, is the feature set, is the number of features in the feature set, is the performance target variable, Features and the target variable The mutual information between The features and target variables in the feature set The mean mutual information between Features and Features The mutual information between is the mutual information between the features in the feature set.

[0131] Combining the maximum correlation and minimum redundancy results in the minimum redundancy maximum correlation criterion, as follows:

[0132]

[0133] in, It is the minimum redundancy and maximum relevance measurement indicator function.

[0134] S42: Use the knowledge of the hot rolling process mechanism of strip steel to analyze process data (including raw material composition, switch quantity and other unstructured data), including control model, production process and other knowledge, screen out performance-related variables, and form a knowledge feature selection subset .

[0135] S43: Soft-sensing auxiliary variable screening based on data and knowledge fusion. The data feature subset initially selected by mRMR Knowledge feature subsets with knowledge-based filtering Fusion is performed to obtain the final auxiliary variable , as follows:

[0136]

[0137] in, is the feature subset of process data, is a subset of data features.

[0138] S5. Build entity embedding models to process unstructured (such as raw material composition) and semi-structured data (such as switch control quantity) to achieve interaction and fusion of multi-source heterogeneous data;

[0139] One-hot encoding is used for encoding. The one-hot encoding of is expressed as follows.

[0140]

[0141] in, Indicates The one-hot encoding of the natural state, Is the length The vector of Expressed as The number of values ​​of For mapping symbols.

[0142] Given input The output of the additional linear neuron is defined as follows,

[0143]

[0144] in, is the number of neurons in the one-hot encoding layer, The weights connecting each neuron in the one-hot encoding layer, For the Layer Coding Layer The one-hot encoding representation of is the weight connecting the one-hot encoding layer to the embedding layer, is the index of the embedding layer.

[0145] After the entity embedding represents all categorical features, the inputs of all embedding layers and all continuous variables are concatenated. is a hyperparameter that needs to be defined by yourself. After passing through two fully connected layers, it is converted into a dense variable of fixed size, and finally combined with the auxiliary variable through the merging layer as the input data of the soft sensor model.

[0146] S6, build a time-space convolutional network (TSTN) time series soft sensor model on the "edge", use the data synchronized on the "cloud" side to achieve local performance prediction, and feed the results back to the "cloud" side;

[0147] Specifically, in this embodiment, the “side” construction of a time-space convolutional network (TSCN) time series soft sensor model includes:

[0148] S61: Construct a TSCN neural network model, including temporal convolution layer, spatial convolution layer, pooling layer, and fully connected layer. By combining temporal and spatial convolution, the spatiotemporal information is fused together to effectively extract the spatiotemporal features of the data.

[0149] The temporal convolution layer is used to process time series data and extract temporal features through one-dimensional convolution operations; the spatial convolution layer is used to process structural data between different processes and capture spatial features through two-dimensional convolution operations; the pooling layer is used to reduce dimensionality and extract the most significant features; the fully connected layer is used to learn nonlinear relationships and map them to the output space.

[0150] S62: Soft sensor model training and deployment to the edge server. The auxiliary variable data obtained in S53 is obtained from the metadata database and input into the constructed TSCN neural network for training. The loss function of the model training is:

[0151]

[0152] in, is the actual value of performance, is the predicted value of the soft sensor model, are the parameters of the model.

[0153] S63: The "side" server obtains the online real-time metadata of the hot rolling process of the strip from the "cloud side", and inputs the selected auxiliary variables into the TSCN soft measurement model that has been trained on the "side" to complete the real-time performance prediction, and uploads the prediction results to the "cloud side" server.

[0154] On the "cloud side", S7 uses a comprehensive evaluation description method of multiple performance indicators based on the entropy weight method to integrate local prediction results and complete the performance evaluation of the entire process of hot rolling of strip steel, which can be viewed in real time by operators and managers.

[0155] Specifically, in the implementation case, a comprehensive evaluation description method of multiple performance indicators based on the entropy weight method is used to integrate local performance prediction results to achieve comprehensive evaluation of multiple performance indicators, including:

[0156] S71: A comprehensive evaluation description method for multiple performance indicators based on the entropy weight method. The entropy weight method can determine the importance of indicators according to the amount of information of different indicators, and then weight the multiple indicators to obtain the evaluation results. Specifically, information entropy can represent the degree of variation of data under a certain indicator, which is defined as

[0157]

[0158] Among them, n represents a total of n indicators. Indicated in Among the indicators, The proportion of performance index values ​​to the total sample, that is

[0159]

[0160] The larger the information entropy, the greater the variation of the indicator, that is, the greater the discrimination between different samples under the indicator. In the entropy weight method, the indicator with large information entropy will get a larger weight. The weight is expressed as

[0161]

[0162] in Indicates the number of performance indicators, No. The information entropy (degree of variation) of each performance indicator. Then the final performance score can be expressed as

[0163]

[0164] in, No. performance indicator weights, Indicated in Among the indicators, The proportion of performance index values ​​in the total sample

[0165] S72: Based on the Django framework, the final performance evaluation score (cloud side) and the performance prediction results of each process (edge ​​side) are visualized, allowing operators and managers to view the production status in real time.

[0166] Furthermore, an embodiment of the present invention also provides a "cloud-edge" collaborative strip hot rolling production process data governance and performance prediction and evaluation device, wherein it should be noted that since the "cloud-edge" collaborative strip hot rolling production process data governance and performance prediction and evaluation device provided in this embodiment corresponds to the above-mentioned "cloud-edge" collaborative strip hot rolling production process data governance and performance prediction and evaluation method specific implementation method, the device can achieve the purpose of the present invention by executing the process steps in the above-mentioned method specific implementation method. Therefore, the explanations in the above-mentioned "cloud-edge" collaborative strip hot rolling production process data governance and performance prediction and evaluation method specific implementation method are also applicable to the "cloud-edge" collaborative strip hot rolling production process data governance and performance prediction and evaluation device. Therefore, the specific implementation method of the device will not be repeated in the following specific implementation methods of the present invention.

[0167] like Figure 6 As shown in the figure, the "cloud-edge" collaborative strip hot rolling production process data management and performance prediction and evaluation device includes:

[0168] The cloud side acquisition module 201 is used to acquire the strip hot rolling production data;

[0169] The cloud-side data governance module 202 is used to construct a panoramic view of the hot-rolled strip steel metadata to implement data governance;

[0170] A data outlier detection module 203 is used to detect outliers in the acquired data;

[0171] A non-ideal data expansion module 204, used for expanding non-ideal data;

[0172] Metadata building module 205, used to describe multivariate heterogeneous data in the process of hot rolling of steel strip;

[0173] A metadata system construction module 206 is used to construct a metadata system for the strip hot rolling process;

[0174] Data information description expression scheme module 207, used for data information sharing and communication in the metadata system of the strip hot rolling process;

[0175] The side input module 208 is used to obtain the “cloud side” production data and input it into the constructed soft measurement model;

[0176] The edge output module 209 is used to obtain the process performance prediction result according to the "cloud side" production data and the "edge side" soft measurement model;

[0177] The cloud-side output module 210 is used to integrate the sub-process performance prediction results to obtain the final performance evaluation results for use by management personnel and operators;

[0178] In summary, the "cloud-edge" collaborative strip hot rolling production process data management and performance prediction and evaluation device solution in this embodiment has at least the following beneficial effects:

[0179] (1) The proposed six-dimensional data model, the multi-source heterogeneous data description paradigm, the constructed hot strip rolling metadata database, and the established data information description expression scheme will provide a complete set of data governance technologies for hot strip rolling, form valuable data assets for hot strip rolling, and at the same time serve as a digital base to promote the digital and intelligent transformation of hot strip rolling.

[0180] (2) The established time-space convolutional network soft sensor model can capture the characteristics of the time series and process of hot rolling production of strip steel, greatly improving the accuracy of performance prediction. It can replace destructive offline tests such as mechanical properties testing, reduce the frequency of inspections and tests, bring direct economic benefits, and achieve energy saving and consumption reduction.

[0181] (3) The “cloud-edge” collaborative mechanism is taken into consideration, which allows real-time monitoring of the product performance of sub-processes and the status of the entire production process, providing information support for operators. Based on performance prediction, the cause of the failure can be traced, and production can be intervened in a timely manner to achieve quality and efficiency improvements.

[0182] The present invention also provides a computer readable storage medium, such as Figure 7As shown, the electronic device includes: a processor and a memory; wherein the processor and the memory can be connected via a communication bus; the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the method of the first embodiment. In addition, the electronic device may also include a transceiver, the processor and the transceiver can be connected via a communication bus, and the transceiver is used to communicate with other devices.

[0183] Next, combine Figure 7 The following is a detailed introduction to the various components of the electronic device:

[0184] Among them, the processor is the control center of the electronic device, and the electronic device may include multiple processors, each of which may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may be a processor or a general term for multiple processing elements. For example, the processor is one or more central processing units (CPUs), or other general-purpose processors, application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement an embodiment of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (field programmable gate arrays, FPGAs), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor may execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0185] In a specific implementation, as an embodiment, the processor may include one or more CPUs, such as Figure 7 The CPU0 and CPU1 shown in the figure are, of course, only exemplary.

[0186] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0187] Optionally, the memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and accessed through the interface circuit ( Figure 7 The processor is coupled to the processor (not shown), which is not specifically limited in this embodiment of the present invention.

[0188] The transceiver may include a receiver and a transmitter ( Figure 7 The receiver is used to implement the receiving function, and the transmitter is used to implement the sending function. The transceiver can be integrated with the processor or exist independently and communicate with the electronic device through the interface circuit ( Figure 7 (not shown) is coupled to the processor, which is not specifically limited in this embodiment of the present invention.

[0189] In addition, it should be noted that Figure 7 The structure of the electronic device shown in the figure does not constitute a limitation on the device, and the actual device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. In addition, the technical effects achieved by the electronic device when executing the method of the first embodiment above can refer to the technical effects described in the first embodiment above, so they are not repeated here.

[0190] In addition, the method disclosed in the embodiment of the present invention can also be implemented as a computer program executed by a processor, and the computer program can be stored in a computer-readable storage medium. The storage medium stores at least one instruction, and the instruction is loaded and executed by the processor to implement the method of the first embodiment above. Among them, the computer-readable storage medium can be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device. The instructions stored therein can be loaded by the processor in the terminal and the above method can be executed.

[0191] In addition, it should be noted that the present invention can be provided as a method, an apparatus or a computer program product. Therefore, the embodiment of the present invention can be in the form of a full or partial hardware embodiment, a full or partial software embodiment or an embodiment combining software and hardware. Moreover, when implemented using software, the embodiment of the present invention can be in the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. 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, the process or function described in the embodiment of the present invention is generated in whole or in part. 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 computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center containing one or more available media sets. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium. The semiconductor medium may be a solid state drive.

[0192] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0193] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0194] It should also 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. The terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or terminal 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 terminal device. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements. In addition, the term "and / or" is only an association relationship describing the associated objects, indicating that there can be three relationships, for example, A and / or B, which can represent: A exists alone, A and B exist at the same time, and B exists alone, wherein A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding. "At least one" means one or more, and "plurality" means two or more. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or plural.

[0195] In addition, it can be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean 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.

[0196] Those skilled in the art will appreciate that the units and algorithm steps of each example 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 performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can 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 the present invention.

[0197] In several embodiments provided by the present invention, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of functional modules / units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0198] If the method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0199] Finally, it should be noted that the above is a preferred embodiment of the present invention. It should be pointed out that although the preferred embodiment of the present invention has been described, for those skilled in the art, once the basic creative concept of the present invention is known, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention. Therefore, the attached claims are intended to be interpreted as including the preferred embodiment and all changes and modifications that fall within the scope of the embodiments of the present invention.

Claims

1. A cloud-edge collaborative strip hot rolling process data management and performance prediction method, characterized in that: include: The S1 cloud-side server obtains the production process data of hot-rolled strip steel and uses the isolation forest algorithm to quickly detect outliers; S2 performs non-ideal data expansion on cloud-side data based on non-ideal data processing methods based on robust normalization and partial robust regression; S3 uses the multi-source heterogeneous data description paradigm of the six-dimensional data model to establish a global view of the metadata of the strip hot rolling process, and saves the data of the entire line and the entire life cycle to the cloud-side server to achieve data governance; S4 combines the knowledge of the hot rolling process mechanism of strip steel and the maximum relevance minimum redundancy algorithm to screen the performance-related auxiliary variables; S5 builds an entity embedding model and processes unstructured and semi-structured production data to achieve interaction and fusion of the multi-source heterogeneous data; S6 constructs a time-space convolutional network soft sensor model on the edge, uses the data synchronized by the cloud side to achieve local performance prediction, and feeds the result back to the cloud side; S7 uses a multi-performance index comprehensive evaluation description method based on the entropy weight method to fuse local prediction results on the cloud side to complete the performance evaluation of the entire process of hot rolling of strip steel.

2. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S1, the isolation forest algorithm is used to quickly detect outliers, including: The production process data includes three levels of data: L1, L2, and L3. The isolation forest algorithm is used to quickly detect data outliers on the three levels of data, build a binary tree, and specify the cutting dimension and cutting point P; Repeat the classification and placement of sample points in the training sample set until there is only one data in the node or the maximum height of the tree reaches the limit; A new binary tree is selected cyclically, i isolated trees iTree are generated, multiple isolated trees are constructed, the average path length of each isolated tree is calculated, the path length of the data point in each tree is calculated to determine the isolated value, outliers are quickly detected based on the set threshold and the isolated value, and the outliers are marked.

3. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S2, the non-ideal data processing method based on robust normalization and partial robust regression is used to perform non-ideal data expansion on the cloud side data, including: S21 normalizes the original data on the cloud side to eliminate the influence of different dimensions; S22 uses robust normalization to estimate the mean and variance of the measured data; S23 uses a partial robust M regression algorithm to further process the strip hot rolling process and quality data to eliminate outliers.

4. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S3, a six-dimensional data model is used to describe multi-source heterogeneous data and establish a global view of the metadata of the hot rolling process of strip steel, including: Constructing a four-layer metamodel of the steel rolling process, and based on the four-layer metamodel structure, establishing a six-dimensional data model for the steel rolling process data, wherein the six-dimensional data model includes a time domain T, a space domain S, an entity object domain I, a relationship domain R, an attribute P, and an attribute value V; On the basis of the six-dimensional data model, the multi-level information space mapping relationship from the physical level of process and equipment to comprehensive production indicators, operation indicators and control indicators is combined to complete the unified description of multi-source heterogeneous data of the hot rolling process of the strip steel; Metadata is used to construct a metadata structure model. According to the steel rolling process data management and service flow, a strip hot rolling data metadata system is constructed based on application scenarios and divided into technical metadata, management metadata, and business metadata.

5. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S4, the performance-related auxiliary variables are screened by combining the knowledge of the hot rolling process mechanism of the strip steel and the maximum relevance minimum redundancy algorithm, including: Based on the maximum correlation minimum redundancy algorithm as the correlation measurement basis, the features associated with the performance variables are screened according to the production process of hot rolling of strip steel, and the initial feature subsets of each process are obtained. , calculated as follows: ; ; in, is the feature set, is the number of features in the feature set, is the performance target variable, Features and the target variable The mutual information between The features and target variables in the feature set The mean mutual information between Features and Features The mutual information between is the mutual information between each feature in the feature set; Combining the maximum correlation and minimum redundancy results in the minimum redundancy maximum correlation criterion, which is calculated as follows: ; in, It is the minimum redundancy maximum correlation measurement indicator function; Analyze process data using knowledge of the hot rolling process mechanism of strip steel, screen out variables related to performance, and form a knowledge feature selection subset, wherein the process data includes raw material composition and switch quantity; Based on the soft sensor auxiliary variable screening of data and knowledge fusion, the data feature subsets preliminarily selected by the maximum correlation and minimum redundancy algorithms are Knowledge feature subsets with knowledge-based filtering Fusion is performed to obtain the final auxiliary variable .

6. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S5, processing unstructured and semi-structured production data to achieve interaction and fusion of the multi-source heterogeneous data includes: One-hot encoding is used for encoding. The one-hot encoding of is expressed as follows: ; in, represents the one-hot encoding of the i-th natural state, is the length For the vector, Expressed as The number of values ​​of is the mapping symbol; Given input The output of the additional linear neuron is defined as follows: ; in, is the number of neurons in the one-hot encoding layer, The weights connecting each neuron in the one-hot encoding layer, For the Layer Coding Layer The one-hot encoding representation of is the weight connecting the one-hot encoding layer to the embedding layer, is the index of the embedding layer; After the entity embedding represents all categorical features, the inputs of all embedding layers and all continuous variables are concatenated, where the dimension of the embedding layer is The hyperparameters that need to be defined by themselves are converted into dense variables of fixed size after two fully connected layers, and finally combined with the auxiliary variables through the merging layer as the input data of the soft measurement model.

7. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: In S6, the time-space convolutional network TSTN soft sensor model is constructed on the side including: Build a time-space convolutional neural network model, including time convolution layer, space convolution layer, pooling layer, and fully connected layer. By combining time and space convolution, the time and space information is fused together to effectively extract the time and space features of the data; The soft sensor model is trained and deployed to the edge server. The auxiliary variable data is obtained from the metadata database and input into the constructed time-space convolutional network neural network for training. The loss function of the model training is: ; in, is the actual value of performance, is the predicted value of the soft sensor model, are the parameters of the model; The edge server obtains the online real-time metadata of the hot rolling process of the strip from the cloud side, and inputs the selected auxiliary variables into the trained time-space convolutional network soft measurement model on the edge to complete the real-time performance prediction.

8. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 7 is characterized in that: The temporal convolution layer extracts temporal features through a one-dimensional convolution operation to process time series data, the spatial convolution layer captures spatial features through a two-dimensional convolution operation to process structural data between different processes, and the pooling layer is used for dimensionality reduction and extraction of significant features.

9. The cloud-edge collaborative strip hot rolling process data management and performance prediction method according to claim 1 is characterized in that: The cloud side uses a multi-performance index comprehensive evaluation description method based on the entropy weight method to fuse local prediction results, including: A comprehensive evaluation description method for multiple performance indicators based on the entropy weight method. The entropy weight method can determine the importance of indicators according to the amount of information of different indicators, so as to obtain the evaluation results by weighting multiple indicators. Information entropy represents the degree of variation of data under a certain indicator, which is expressed as: ; Among them, n represents a total of n indicators. Indicated in Among the indicators The proportion of performance index values ​​to the total sample is expressed as: ; The weight is expressed as: ; in Indicates the number of performance indicators, No. The information entropy of the performance indicators is expressed as: ; in, Indicates performance indicator weights, Represents the weight of the jth performance indicator, and visualizes the final performance evaluation score based on the Django framework to predict the performance of each process on the cloud side and the edge side.

10. A cloud-edge collaborative strip hot rolling process data management and performance prediction device, used to implement the cloud-edge collaborative strip hot rolling process data management and performance prediction method as described in any one of claims 1 to 9, characterized in that: include: A cloud-side acquisition module configured to acquire strip hot rolling production data; A cloud-side data governance module is configured to construct a metadata panorama of hot-rolled strip steel. The data governance module includes a data outlier detection module, a non-ideal data expansion module, a metadata construction module, a metadata system construction module, and a data information description expression solution module; The edge input module is configured to obtain cloud-side production data and input it into the constructed soft measurement model; A side output module configured to obtain a process performance prediction result according to the cloud-side production data and the side soft measurement model; The cloud-side output module is configured to fuse the sub-process performance prediction results and output the final performance evaluation results.

Citation Information

Patent Citations

  • Finish rolling whole process distributed operation performance evaluation and non-optimal reason tracing method

    CN114048592A

  • Strip steel hot continuous rolling non-stationary process tiny fault detection method and device

    CN115994337A

  • Rolling process data element model and data space construction method and system

    CN117573880A

  • KR20240057765A