A comprehensive evaluation method for the operation status of the entire finishing rolling process

Abstract

The present invention discloses a comprehensive evaluation method for the operation status of the entire process of finishing rolling, including: obtaining historical sample data generated in the entire process of finishing rolling in the past, and establishing a historical sample database; dividing the entire process of finishing rolling into three subsystems, namely, an upstream subsystem, a midstream subsystem, and a downstream subsystem; constructing an operation status evaluation model corresponding to each subsystem, and training the constructed operation status evaluation model corresponding to each subsystem based on the historical sample database, and using the trained operation status evaluation model corresponding to each subsystem to respectively realize the operation status evaluation of each subsystem; fusing the operation status evaluation results of each subsystem to realize the operation status evaluation of the entire process of finishing rolling. The present invention can effectively monitor and judge the operation status of the production process in a timely and accurate manner; it aims to explore effective solutions to key challenges in the operation status evaluation of modern process industries, and has important practical application and promotion value.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial production process control and monitoring, and in particular to a comprehensive evaluation method for the operation status of the entire finishing rolling process. Background Art

[0002] With the development of automation and information technology, modern industrial processes are becoming more and more integrated and complex. Due to uncertain factors such as raw material changes, equipment aging, and changes in the operating environment, industrial processes may gradually deviate from the optimal working point and enter a non-optimal operating state, affecting the economic benefits of the enterprise. Therefore, developing an effective process operating status assessment method is an inevitable trend in process monitoring and an urgent need for improvement. Taking the entire process of finishing rolling as an example, we can further deepen our understanding of the necessity and importance of industrial process operating status assessment.

[0003] The whole process of finishing rolling is highly automated, with main equipment closely connected and streamlined. Due to the frequent changes in raw materials and the drastic fluctuations in working modes, the whole process of finishing rolling has inherent nonlinear and dynamic characteristics. The whole process of finishing rolling mainly consists of seven stands, which can be divided into three subsystems: upstream, midstream and downstream through in-depth research on process knowledge. In the normal production process, the non-optimal operating state of any working area will affect the quality of the entire production line and the final product, so it is necessary to evaluate its operating state in real time. In addition, once a fault occurs, it is also necessary to effectively identify the type and severity of the fault and formulate different maintenance strategies for different faults. For example, the thickness abnormality caused by the fault of the upstream stand can be compensated to a certain extent by adjusting the rolling operation of the subsequent stand, and this minor fault does not need to be alarmed. However, when the fault is difficult to compensate and the output thickness of the midstream and downstream stands is abnormal, the fault needs to be alarmed and the equipment maintained. Once the maintenance operation is performed, the entire finishing rolling process must be stopped and the slab must be heated in the heating furnace for a long time. This will not only lead to excessive oxidation of the slab and reduce the product yield, but also reduce product quality, resulting in production waste and huge economic losses.

[0004] In order to simultaneously handle process nonlinearity and dynamics, long short-term memory (LSTM) networks have been widely used. Compared with traditional recurrent neural networks (RNNs), LSTM can solve the gradient vanishing and gradient exploding problems. However, the LSTM model has a complex structure and takes a long time to train.

[0005] In addition, traditional process monitoring methods often focus on how to distinguish between "normal" and "faulty" operating conditions. This rough evaluation method obviously cannot meet the actual production needs. Based on process monitoring, the operating status evaluation is to conduct a more detailed evaluation of the severity of process failures and the quality of operating performance by analyzing the abnormalities of process state parameters and possible deviations in operation control during the production process. In recent years, operating status evaluation has gradually received widespread attention from academia and industry. Existing methods can be mainly divided into two categories: evaluation methods based on quantitative information and evaluation methods based on qualitative information. Multivariate statistical methods are the most widely used quantitative evaluation methods, which are suitable for processes with less prior knowledge of the process. The evaluation method based on the probability framework requires prior knowledge to assist in determining the probability density function. The method based on neural networks is favored by researchers because of its strong learning ability and nonlinear processing ability. The evaluation method based on qualitative information deals with the evaluation of abnormal conditions based on qualitative information. The most commonly used methods for processing qualitative information are Bayesian networks, fuzzy theory, and rough set theory.

[0006] Strictly speaking, most industrial process variables are not linearly related. In particular, due to changes in operating conditions, the nonlinearity between process variables becomes increasingly obvious. Traditional linear methods cannot effectively extract and analyze process potential characteristics.

[0007] Therefore, nonlinear methods have been widely used in recent years. Based on the total latent structure projection (TPLS), the nuclear total latent structure projection (TKPLS) was proposed and applied to the hydrometallurgical process. In order to avoid the tedious data alignment work, an evaluation method based on nonlinear optimal relevant variation information (NORVI) was proposed. With the development of deep learning, evaluation methods based on deep neural networks have received attention. Yan Xuefeng et al. used autoencoders (AE) to extract nonlinear features and established a corresponding nonlinear process monitoring model. (Nonlinear)

[0008] Traditional operating status assessment methods often assume that process data is static and ignore the time correlation of variables. Modern industrial processes usually have dynamic characteristics, which are closely related to the internal mechanisms of the process, including the operating stage of the process, physical and chemical reaction mechanisms, noise and interference, etc. In order to solve this problem, relevant scholars have proposed many dynamic models. Zou Xiaoyu and Zhao Chunhui used the method of cointegration analysis to analyze the long-term equilibrium relationship of dynamic processes. Qin Sizhao and others proposed a new dynamic principal component analysis (DiPCA) model for dynamic data modeling and process monitoring. Zhang Shumei and others used canonical variable analysis (CVA) and slow feature analysis (SFA) to accurately identify the operating status of dynamic processes. (Dynamic)

[0009] Abnormal conditions caused by instrument failure, process actuator failure, improper operation, etc. may lead to paralysis of the entire system or even catastrophic accidents. It is not sufficient to use only conventional operating status assessment. The assessment of the degree of fault under abnormal conditions is equally important. However, the existing technology makes it difficult to complete the evaluation of the operating status level under normal conditions and the evaluation of the degree of fault under abnormal conditions in the entire finishing process where nonlinearity and dynamics coexist. Therefore, it is necessary to study the comprehensive evaluation of the operating status. The so-called comprehensive evaluation includes two levels: first, this method can simultaneously deal with the nonlinearity and dynamic characteristics of the process; second, this method can effectively evaluate the operating status level under normal conditions (excellent / good / medium / poor) and the degree of fault under abnormal conditions (minor fault / medium fault / serious fault), such as Figure 1 shown. Summary of the invention

[0010] The present invention provides a comprehensive evaluation method for the operating status of the entire finishing rolling process to solve the technical problem in the prior art that it is difficult to complete the operating status level evaluation under normal conditions and the fault degree evaluation under abnormal conditions in the entire finishing rolling process where nonlinearity and dynamics coexist.

[0011] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0012] On the one hand, the present invention provides a method for comprehensively evaluating the operation status of the entire finishing rolling process, comprising:

[0013] Obtain historical sample data generated in the entire process of finishing rolling and establish a historical sample database;

[0014] The whole process of finishing rolling is divided into three subsystems: upstream subsystem, midstream subsystem and downstream subsystem;

[0015] Constructing operation status evaluation models corresponding to each subsystem respectively, and training the constructed operation status evaluation models corresponding to each subsystem based on the historical sample database, and using the trained operation status evaluation models corresponding to each subsystem to respectively implement operation status evaluation of each subsystem;

[0016] The operating status evaluation results of each subsystem are integrated to realize the operating status evaluation of the entire finishing rolling process.

[0017] Furthermore, the whole finishing rolling process includes a first stand, a second stand, a third stand, a fourth stand, a fifth stand, a sixth stand and a seventh stand;

[0018] The historical sample data include the roll gap, rolling force, rolling speed, current, voltage corresponding to the first frame, the second frame, the third frame, the fourth frame, the fifth frame, the sixth frame and the seventh frame when steel plates of different steel grades and different plate thicknesses were generated in the past, and the bending roll force corresponding to the second frame, the third frame, the fourth frame, the fifth frame, the sixth frame and the seventh frame.

[0019] Further, the upstream subsystem includes a first rack, a second rack and a third rack; the midstream subsystem includes a fourth rack and a fifth rack; and the downstream subsystem includes a sixth rack and a seventh rack.

[0020] Furthermore, the operating status evaluation model is a Siamese gated recurrent unit network (SGRU) formed by combining a Siamese neural network (SNN) with a gated recurrent unit (GRU).

[0021] Furthermore, SGRU has two inputs: s in and re ; Among them, s in is the sample of the training set, s re is a reference sample randomly selected from the training set and used to calculate the similarity;

[0022] The GRU encoder-decoder is used to extract s in and re Dynamic nonlinear characteristics of and

[0023]

[0024] In the formula, and represents the update gate of GRU, and Indicates candidate status; and is transferred to the decoder GRU, output and To reconstruct sequence data; It represents the dynamic nonlinear characteristics at time T-1, and T represents the length of the time series.

[0025] Furthermore, SGRU is divided into a similarity comparison unit and a classification unit; among them,

[0026] In the similarity comparison unit, by calculating and The Euclidean distance of is used to compare the similarity S of dynamic nonlinear features:

[0027]

[0028] In the classification unit, the fully connected layer uses a softmax function to perform fault identification and operation status evaluation; Represents input The probability of being identified as class k, the output of the k-class classifier for:

[0029]

[0030] Where K represents the number of categories, Represents the softmax model parameters; Represents the dynamic nonlinear features extracted by the GRU encoder-decoder.

[0031] Furthermore, based on the historical sample database, the operation status assessment model corresponding to each subsystem is trained, including:

[0032] According to the subsystems corresponding to the historical sample data, the data in the historical sample database is divided into an upstream subsystem data set, a midstream subsystem data set and a downstream subsystem data set;

[0033] The principal component analysis algorithm is used to compress and transform the data sets corresponding to each subsystem, and the compressed data set X corresponding to the upstream subsystem is obtained. 上 , the compressed transformed data set X corresponding to the midstream subsystem 中 , and the compressed transformed dataset X corresponding to the downstream subsystem 下 ;

[0034] Construct the fusion data set corresponding to each subsystem and obtain the upstream subsystem fusion data set I 上 , Midstream Subsystem Fusion Dataset I 中 Fusion of datasets with downstream subsystems I 下 ;in,

[0035]

[0036] Where P 中 For X 中 The corresponding compression matrix, P 下 For X 下 The corresponding compression matrix, P 上 For X 上 The corresponding compression matrix;

[0037] Use sliding window to 上 ,I 中 and I下 They are constructed as time series data respectively;

[0038] Utilize I 上 ,I 中 and I 下 The corresponding time series data are used to train the operation status assessment models corresponding to the upstream subsystem, the midstream subsystem and the downstream subsystem respectively to obtain the trained operation status assessment models.

[0039] Furthermore, after constructing the fusion data set corresponding to each subsystem, the method further includes:

[0040] Construct a mechanism structure matrix S that meets the interconnection of subsystems;

[0041] According to the mechanism structure matrix S, the fusion data set corresponding to each subsystem is updated according to the following rules:

[0042]

[0043] In the formula, s ib represents the element corresponding to the i-th row and the b-th column in the mechanism structure matrix S, P i Represents X i The corresponding compression matrix, i,b∈{upper, middle, lower}.

[0044] Furthermore, the operating status evaluation results of each subsystem are integrated to realize the operating status evaluation of the entire finishing rolling process, including:

[0045] According to the thickness deviation of the strip steel product, the operating status is divided into four levels: excellent, good, medium and poor, which are represented by serial numbers 1, 2, 3 and 4 respectively; the fault types are fault 1, fault 2...fault F, and each fault type is numbered 5, 6,..., F+4;

[0046] At time t, the operating status evaluation result G of the upstream subsystem is 上 (t), the operating status evaluation result of the midstream subsystem G 中 (t) and the operating status evaluation result G of the downstream subsystem 下 (t) is fused as follows to obtain:

[0047] G(t)=max{G 上 (t),G 中 (t),G 下 (t)}

[0048] Among them, G(t) represents the global evaluation result;

[0049] Based on G(t), the operating status and fault type are determined.

[0050] Further, the determining of the operating state and the fault type based on G(t) includes:

[0051] If G(t) = 1, the process is determined to be in the optimal operating state;

[0052] If 1<G(t)≤4, it is determined that the process is in a non-optimal operating state;

[0053] If G(t)>4, it is determined that the process is in an abnormal state. At this time, the fault degree is evaluated according to the following rules:

[0054] If G 下 (t)>4 indicates a serious fault. In this case, the entire production line needs to be stopped to repair the faulty equipment to ensure production safety and product quality;

[0055] If G 中 (t)>4, indicating a medium fault, in which case a fault warning is issued;

[0056] If G 下 (t)≤4, maintenance will be performed after the current batch is completed;

[0057] If G 上 (t)>4 indicates a minor fault, in which case the evaluation results are recorded and maintenance is performed during downtime.

[0058] On the other hand, the present invention further provides an electronic device, comprising a processor and a memory; wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the above method.

[0059] On the other hand, the present invention further provides a computer-readable storage medium, wherein at least one instruction is stored in the storage medium, and the instruction is loaded and executed by a processor to implement the above method.

[0060] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0061] The comprehensive evaluation method for the operation status of the entire process of finishing rolling provided by the present invention is a distributed comprehensive evaluation method for operation status. The method provides a new idea and approach for the evaluation of the operation status of the entire process of finishing rolling based on data and knowledge, and makes up for the problem of the coexistence of nonlinearity and dynamics in the evaluation of the operation status of the entire process of finishing rolling that is difficult to solve by the traditional operation status evaluation method. It aims to face the key challenges in the operation status evaluation and fault degree evaluation of the modern process industry, explore practical and effective solutions, and has important practical application and promotion value. Compared with the traditional operation status evaluation method, the evaluation accuracy of the present invention is improved by 6%. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 It is a schematic diagram of the connotation of comprehensive operation status evaluation of finishing rolling;

[0064] Figure 2 It is a schematic diagram of the execution flow of the comprehensive evaluation method for the operation status of the entire finishing rolling process provided by an embodiment of the present invention;

[0065] Figure 3 It is a schematic diagram of the whole process of finishing rolling;

[0066] Figure 4 It is a schematic diagram of a specific execution flow of the method for comprehensive evaluation of the operation status of the entire finishing rolling process provided by an embodiment of the present invention;

[0067] Figure 5 is a schematic diagram of the SGRU network structure provided by an embodiment of the present invention;

[0068] Figure 6 Schematic diagram of a distributed comprehensive evaluation system implemented by the comprehensive evaluation method for the operation status of the entire finishing rolling process provided by an embodiment of the present invention;

[0069] Figure 7 It is a schematic diagram of evaluation results evaluated by the evaluation method provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0070] 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.

[0071] First embodiment

[0072] This embodiment provides a method for comprehensively evaluating the operation status of the entire finishing rolling process. The method can be implemented by an electronic device. The execution process of the method is as follows: Figure 2 As shown, the following steps are included:

[0073] S1, obtaining historical sample data generated in the entire process of finishing rolling in the past and establishing a historical sample database;

[0074] It should be noted that the entire finishing rolling process includes a total of 7 frames; based on this, the historical sample data obtained in this embodiment are various operating status level data groups generated in the past when different steel grades and steel plates of different thicknesses were generated. Specifically, the data group includes the roll gap, rolling force, rolling speed, current, voltage of frames 1-7, and the bending roll force of frames 2-7 (frame 1 has no bending roll force) and other process variables.

[0075] S2, based on the finishing rolling process knowledge and the transmission relationship between material flow, energy flow and information flow, the whole finishing rolling process is divided into three subsystems: upstream subsystem, midstream subsystem and downstream subsystem;

[0076] It should be noted that the entire process of finishing rolling is composed of multiple production equipments, which cooperate and communicate with each other. In this regard, in this embodiment, rack 1, rack 2 and rack 3 are divided into upstream subsystems; rack 4 and rack 5 are divided into midstream subsystems; and rack 6 and rack 7 are divided into downstream subsystems. Figure 3 shown.

[0077] S3, constructing the operation status evaluation model corresponding to each subsystem respectively, and training the constructed operation status evaluation model corresponding to each subsystem based on the historical sample database, and using the trained operation status evaluation model corresponding to each subsystem to respectively implement the operation status evaluation of each subsystem;

[0078] Specifically, in this embodiment, Figure 4 As shown, the implementation process of the above S3 is as follows:

[0079] S31, in order to achieve effective operation status evaluation, according to the process knowledge and the subsystem corresponding to the historical sample data, the data in the historical sample database is divided into different subsystems X = [X 上 ,X 中 ,X 下 ], namely: upstream subsystem data set, midstream subsystem data set and downstream subsystem data set;

[0080] S32, using the principal component analysis (PCA) algorithm to compress and transform the data sets corresponding to each subsystem, and obtain the compressed and transformed data set X corresponding to the upstream subsystem 上 , the compressed transformed data set X corresponding to the midstream subsystem 中 , and the compressed transformed dataset X corresponding to the downstream subsystem 下 ; Among them, the expression of compression transformation is as follows:

[0081]

[0082] Where, Tb For subsystem X b The score matrix, P b is the corresponding compression matrix, E b is the residual matrix.

[0083] S33, then subsystem X b As the center, a fusion dataset I containing information of other subsystems is constructed. b , thereby constructing the fusion data set corresponding to each subsystem and obtaining the upstream subsystem fusion data set I 上 , Midstream Subsystem Fusion Dataset I 中 Fusion of datasets with downstream subsystems I 下 ; As shown below:

[0084]

[0085] S34, considering the actual situation of communication between subsystems, construct a mechanism structure matrix S that conforms to the subsystem interconnection, that is, any element s in the matrix ij Indicates whether subsystems i and j are connected. If so, it is 1, otherwise it is 0. It should be noted that the connection between subsystems is directional, that is, there is a s ij ≠s ji Taking the whole process of finishing rolling as an example, the corresponding mechanism structure matrix is:

[0086]

[0087] In the formula, s 12 =1 means that the upstream and midstream are directly related, s 13 =0 means that the upstream and downstream are not directly connected.

[0088] S35, according to the mechanism structure matrix S, fusion data set I b It can be updated according to the following rules:

[0089]

[0090] In the formula, s ib represents the element corresponding to the i-th row and the b-th column in the mechanism structure matrix S, P i Represents X i The corresponding compression matrix, i,b∈{upper, middle, lower}.

[0091] S36, using a sliding window to 上 ,I 中 and I 下 They are constructed as time series data respectively;

[0092] S37, using I 上 ,I 中 and I下 The corresponding time series data are used to train the operation status evaluation models corresponding to the upstream, midstream and downstream subsystems respectively, and the trained evaluation models corresponding to each subsystem are obtained.

[0093] Among them, the operating status evaluation model adopted in this embodiment is a Siamese gated recurrent unit network (SGRU) formed by combining a Siamese neural network (SNN) with a gated recurrent unit (GRU).

[0094] Among them, the specific combination of GRU and SNN is as follows Figure 5 As shown, SNN has two inputs (time series s in and time series s re ), the two inputs are fed into two neural networks (GRU in and GRU re ), the two neural networks map the input to the new space respectively, forming the representation of the input in the new space. The similarity of the two inputs is evaluated by calculating the loss function.

[0095] As a variant of LSTM, the cell state in the gated recurrent unit (GRU) network structure is removed, and hidden states are used to transmit information. Its network performance is similar to that of LSTM, but with lower computational complexity. From the perspective of classification, running state evaluation is a multi-classification problem. Siamese neural network (SNN) is a widely used method for comparing sample similarities. It is a symmetric network composed of two identical networks in parallel, in which weights and thresholds are shared. The main purpose of SNN is to minimize the distance between samples of the same type and maximize the distance between samples of different types.

[0096] The SGRU network structure is as follows Figure 5 As shown in Figure 2, in order to widen the gap between the superior samples and other grade samples (non-superior samples and faulty samples) and improve the classification accuracy of similar samples, SGRU has two inputs: in (samples of the training set), s re (A reference sample randomly drawn from the training set is used to calculate the similarity).

[0097] The GRU encoder-decoder is used to extract s in and re Dynamic nonlinear characteristics of and

[0098]

[0099] In the formula, and represents the update gate of GRU, and represents a candidate state; further, and is transferred to the decoder GRU, output and It can be regarded as reconstructing sequence data; It represents the dynamic nonlinear characteristics at time T-1, and T represents the length of the time series.

[0100] Then, the SGRU is divided into a similarity comparison unit and a classification unit (fault identification and operating status evaluation).

[0101] In the similarity comparison unit, by calculating and The Euclidean distance of is used to compare the similarity S of dynamic nonlinear features:

[0102]

[0103] In the classification unit, the fully connected layer uses the softmax function for fault identification and operation status evaluation; Represents input The probability of being identified as class k, for a class k classifier, the output is:

[0104]

[0105] Where K represents the number of categories (such as different types of faults, different operating status levels), Represents the softmax model parameters Represents the dynamic nonlinear features extracted by the GRU encoder-decoder.

[0106] S4, integrate the operating status evaluation results of each subsystem to realize the operating status evaluation of the entire finishing rolling process.

[0107] Specifically, in this embodiment, the above steps implement a method such as Figure 6 The distributed comprehensive evaluation system shown in the figure, based on which, the process of implementing the evaluation of the running status of the whole process of finishing rolling in this embodiment is as follows:

[0108] S41, according to the thickness deviation of the strip steel product (as shown in Table 1), the operating status is divided into four levels: excellent, good, medium and poor, and represented by serial numbers 1, 2, 3 and 4 respectively. The fault types are fault 1, fault 2...fault F, and for simplicity, they are numbered 5, 6...F+4.

[0109] Table 1 Thickness deviation and operating status level

[0110] Thickness(mm) excellent good middle Difference 0.80～2.00 ±0.04 ±0.08 ±0.12 ±0.15 2.00～3.00 ±0.05 ±0.10 ±0.14 ±0.18 3.00～4.00 ±0.06 ±0.12 ±0.17 ±0.22

[0111] S42, at time t, the upstream subsystem evaluation result G 上 (t), midstream subsystem evaluation results G 中 (t), downstream subsystem evaluation results G 下 (t) Perform the following fusion to obtain the global evaluation result:

[0112] G(t)=max{G 上 (t),G 中 (t),G 下 (t)}

[0113] Specifically, Figure 4 As shown:

[0114] If G(t)=1, the process is in the optimal operating state.

[0115] If 1<G(t)≤4, the process is in a non-optimal operating state.

[0116] If G(t)>4, the process is in an abnormal state. Due to the automatic control system, the abnormality of the upstream subsystem can be compensated by the subsequent subsystem. However, when the abnormality occurs in the midstream or downstream subsystem, it is difficult to compensate, and the degree of failure should be evaluated according to the following procedure:

[0117] Serious failure (G 下 (t)>4): The downstream rack fails and the system cannot meet normal production requirements. In this case, the entire production line needs to be stopped and the faulty equipment needs to be overhauled to ensure production safety and product quality.

[0118] Medium fault (G 中 (t)>4): The intermediate rack fails, reducing the operating performance of non-critical processes. Since the slab rolled on the intermediate rack will be further rolled on the downstream rack, the probability of failure of the downstream rack is greatly increased. At this time, the system will receive a warning. If G 下 If (t)≤4, process maintenance should be performed after the current batch is completed. Otherwise, it is necessary to stop operation and perform overhaul.

[0119] Minor fault (G 上(t)>4): The fault occurs in the upstream stand, affecting the input thickness of the midstream lower stand. Due to the control system of the finishing process, this type of fault can be compensated or even eliminated. Therefore, it is only necessary to record the evaluation results and perform maintenance during downtime that does not affect normal production.

[0120] Evaluation results such as Figure 7 shown.

[0121] In summary, in order to effectively monitor the normal operation status of the entire process of finishing rolling and avoid production interruption due to faults, this embodiment draws on the idea of ​​distributed process monitoring and combines the advantages of GRU and SNN. The advantages of SNN and GRU are combined. The SGRU model has the ability to simultaneously process process dynamics and nonlinearity, and can better measure the similarity between samples, thereby improving the accuracy of the network model. Based on this, this embodiment provides a distributed operation status comprehensive evaluation method, which combines process technology knowledge with the circulation of three flows (material flow, energy flow, and information flow), divides and fuses the subsystems of the entire process of finishing rolling, and considers the correlation between different subsystems; different from the traditional operation status evaluation (normal working condition), the comprehensive evaluation adds the evaluation of abnormal working conditions on this basis, and can effectively judge the degree of failure in the process. Thereby solving the operation status evaluation problem of normal working conditions and abnormal working conditions in the process industry, and achieving the effect of timely and accurate effective monitoring and judgment of the operation status of the production process.

[0122] Second embodiment

[0123] This embodiment provides an electronic device, which includes a processor and a memory; wherein 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.

[0124] The electronic device may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) and one or more memories, wherein the memory stores at least one instruction, and the instruction is loaded by the processor to execute the above method.

[0125] Third embodiment

[0126] This embodiment provides a computer-readable storage medium, which stores at least one instruction, and the instruction is loaded and executed by a processor to implement the method of the first embodiment. The computer-readable storage medium may be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc. The instructions stored therein may be loaded by a processor in a terminal to execute the method.

[0127] 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 embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0128] 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.

[0129] 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 1 These 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 A step that specifies a function in one or more boxes.

[0130] 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", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a 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 further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0131] 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 comprehensive evaluation method for the operation status of the entire finishing rolling process, characterized in that: include: Obtain historical sample data generated in the entire process of finishing rolling and establish a historical sample database; The whole process of finishing rolling is divided into three subsystems: upstream subsystem, midstream subsystem and downstream subsystem; Constructing operation status evaluation models corresponding to each subsystem respectively, and training the constructed operation status evaluation models corresponding to each subsystem based on the historical sample database, and using the trained operation status evaluation models corresponding to each subsystem to respectively implement operation status evaluation of each subsystem; The operating status evaluation results of each subsystem are integrated to realize the operating status evaluation of the whole process of finishing rolling; The running state evaluation model is a Siamese gated recurrent unit network SGRU formed by combining a Siamese neural network SNN with a gated recurrent unit GRU; SGRU has two inputs: in and re ; Among them, s in is the sample of the training set, s re is a reference sample randomly selected from the training set and used to calculate the similarity; The GRU encoder-decoder is used to extract s in and re Dynamic nonlinear characteristics of and In the formula, and represents the update gate of GRU, and Indicates candidate status; and is transferred to the decoder GRU, output and To reconstruct sequence data; It represents the dynamic nonlinear characteristics at time T-1, and T represents the length of the time series; SGRU is divided into similarity comparison unit and classification unit; among them, In the similarity comparison unit, by calculating and The Euclidean distance of is used to compare the similarity S of dynamic nonlinear features: In the classification unit, the fully connected layer uses a softmax function to perform fault identification and operation status evaluation; Represents input The probability of being identified as class k, the output of the k-class classifier for: In the formula, K represents the number of categories, Represents the softmax model parameters; Represents the dynamic nonlinear features extracted by the GRU encoder-decoder.

2. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 1, wherein the entire finishing rolling process includes a first frame, a second frame, a third frame, a fourth frame, a fifth frame, a sixth frame and a seventh frame; characterized in that: The historical sample data include the roll gap, rolling force, rolling speed, current, voltage corresponding to the first frame, the second frame, the third frame, the fourth frame, the fifth frame, the sixth frame and the seventh frame when steel plates of different steel grades and different plate thicknesses were generated in the past, and the bending roll force corresponding to the second frame, the third frame, the fourth frame, the fifth frame, the sixth frame and the seventh frame.

3. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 2 is characterized in that: The upstream subsystem includes a first rack, a second rack and a third rack; the midstream subsystem includes a fourth rack and a fifth rack; and the downstream subsystem includes a sixth rack and a seventh rack.

4. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 1 is characterized in that: Based on the historical sample database, the operation status assessment model corresponding to each subsystem is trained, including: According to the subsystems corresponding to the historical sample data, the data in the historical sample database is divided into an upstream subsystem data set, a midstream subsystem data set and a downstream subsystem data set; The principal component analysis algorithm is used to compress and transform the data sets corresponding to each subsystem, and the compressed data set X corresponding to the upstream subsystem is obtained. 上 , the compressed transformed data set X corresponding to the midstream subsystem 中 , and the compressed transformed dataset X corresponding to the downstream subsystem 下 ; Construct the fusion data set corresponding to each subsystem and obtain the upstream subsystem fusion data set I 上 , Midstream Subsystem Fusion Dataset I 中 Fusion of datasets with downstream subsystems I 下 ;in, Where P 中 For X 中 The corresponding compression matrix, P 下 For X 下 The corresponding compression matrix, P 上 For X 上 The corresponding compression matrix; Use sliding window to 上 ,I 中 and I 下 They are constructed as time series data respectively; Utilize I 上 ,I 中 and I 下 The corresponding time series data are used to train the operation status assessment models corresponding to the upstream subsystem, the midstream subsystem and the downstream subsystem respectively to obtain the trained operation status assessment models.

5. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 4 is characterized in that: After constructing the fusion data set corresponding to each subsystem, the method further includes: Construct a mechanism structure matrix S that conforms to the interconnection of subsystems; According to the mechanism structure matrix S, the fusion data set corresponding to each subsystem is updated according to the following rules: In the formula, s ib represents the element corresponding to the i-th row and the b-th column in the mechanism structure matrix S, P i Represents X i The corresponding compression matrix,i,b∈{upper,middle,lower}.

6. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 1 is characterized in that: The operating status evaluation results of each subsystem are integrated to realize the operating status evaluation of the entire finishing rolling process, including: According to the thickness deviation of the strip steel product, the operating status is divided into four levels: excellent, good, medium and poor, which are represented by serial numbers 1, 2, 3 and 4 respectively; the fault types are fault 1, fault 2...fault F, and each fault type is numbered 5, 6,..., F+4; At time t, the operating status evaluation result G of the upstream subsystem is 上 (t), the operating status evaluation result of the midstream subsystem G 中 (t) and the operating status evaluation result G of the downstream subsystem 下 (t) is fused as follows to obtain: G(t)=max{G 上 (t),G 中 (t),G 下 (t)} Among them, G(t) represents the global evaluation result; Based on G(t), the operating status and fault type are determined.

7. The comprehensive evaluation method for the operation status of the entire finishing rolling process according to claim 6 is characterized in that: The determining of the operating state and the fault type based on G(t) includes: If G(t) = 1, the process is determined to be in the optimal operating state; If 1<G(t)≤4, it is determined that the process is in a non-optimal operating state; If G(t)>4, it is determined that the process is in an abnormal state. At this time, the fault degree is evaluated according to the following rules: If G 下 (t)>4 indicates a serious fault. In this case, the entire production line needs to be stopped to repair the faulty equipment to ensure production safety and product quality; If G 中 (t)>4, indicating a medium fault, in which case a fault warning is issued; If G 下 (t)≤4, maintenance will be performed after the current batch is completed; If G 上 (t)>4 indicates a minor fault, in which case the evaluation results are recorded and maintenance is performed during downtime.

Images