Methods and Systems for Predicting the Life and Health of Rolls in Five-Stand Cold Rolling Mills

Abstract

This invention provides a method and system for predicting the lifespan and conducting online health diagnosis of rolls in a five-stand cold rolling mill, relating to the field of metallurgical engineering technology. The method involves collecting and preprocessing multi-source heterogeneous data covering the entire lifespan of the rolls to construct a correlated time-series database; establishing a refined finite element model coupled with an Archard wear model using ABAQUS software; acquiring roll stress, strain, and theoretical wear data through multi-condition simulation; constructing a high-precision surrogate model based on a Gaussian process multi-objective regression algorithm; linking the surrogate model with the online production system to dynamically predict the remaining lifespan and health index of the rolls; and achieving precise location of abnormal rolls through defect feature extraction, wavelength matching, and similarity calculation. This invention solves the problems of inaccurate roll life prediction and difficult fault location in existing technologies, enabling intelligent roll management, improving product quality, and reducing production costs, demonstrating significant practicality and economic benefits.

Description

Technical Field

[0001] This invention relates to the field of metallurgical engineering technology, and in particular to a method and system for predicting the life and health of rolls in a five-stand cold rolling mill. Background Technology

[0002] In the five-stand cold continuous rolling mill process, the rolls (including work rolls, intermediate rolls, and support rolls) are key consumables that directly participate in metal deformation and determine the surface quality of the strip. Each stand is typically equipped with six rolls. Under harsh conditions of high speed and high pressure, the rolls will experience both uniform and uneven wear (such as localized spalling and cracking). Uneven wear will be directly replicated on the strip surface, forming periodic defects that lead to product downgrading or scrapping.

[0003] Currently, the management and maintenance of rolling mill rolls mainly rely on empirically-based periodic roll replacements based on fixed rolling tonnage or operating time, or on forced shutdowns for troubleshooting when serious quality problems occur. This approach has significant drawbacks: 1) Inaccurate prediction: Empirical cycles cannot accurately reflect the actual wear state of rolling mill rolls under different specifications, steel grades, and process parameters, potentially leading to "over-maintenance" or "under-maintenance"; 2) Difficulty in locating defects: When defects appear on the strip surface, due to the interconnected operation of the five stands, it is difficult to quickly and accurately determine which roll on which stand the problem originates. Inspections must be conducted on each stand individually, which is time-consuming and labor-intensive, severely impacting production efficiency and costs.

[0004] In existing technologies, although some studies predict roll wear through theoretical formulas or simple models, the accuracy is limited and they generally lack deep integration with online production data, making it impossible to achieve true dynamic prediction and accurate diagnosis. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for predicting the lifespan and online health diagnosis of rolls in a five-stand cold rolling mill. By dynamically and accurately predicting the remaining service life of rolls in each stand, and when quality problems occur on the surface of the strip, it can quickly and accurately diagnose and locate the specific stand and roll causing the problem, thereby achieving scientific and intelligent management of rolls and solving the problems of inaccurate roll life prediction and difficulty in fault location in the prior art.

[0006] To achieve the above objectives, the present invention provides a method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill, the method comprising the following steps:

[0007] Collect production process data, roll body data, and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key;

[0008] Based on the roll body data, a fine finite element model including the work roll, intermediate roll and support roll was established using ABAQUS software. The model was coupled with the Archard wear model, and the roll contact stress distribution, strain energy density and theoretical wear data were output through multi-condition simulation.

[0009] The preprocessed production process data is used as the input feature vector, and the roll contact stress distribution, strain energy density and theoretical wear data are used as the multi-objective output vector to construct a training sample set. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed.

[0010] Based on a high-precision surrogate model, the remaining service life and health index of the rolls are dynamically predicted.

[0011] Preferably, the production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage of each stand;

[0012] The data of the roll body includes the chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount and surface morphology of the work roll, intermediate roll and support roll;

[0013] The strip quality data includes defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics.

[0014] Preferably, the refined finite element model is a dynamic explicit analysis model that considers the elastic deformation of the rolls, used to simulate the transient processes of strip biting, stable rolling, and strip ejection.

[0015] Preferably, the formula for the Archard wear model is:

[0016] Where V is the wear volume and K is the wear coefficient. Where L is the normal load on the contact surface, L is the sliding distance, and H is the hardness of the roll material;

[0017] In the refined finite element model simulation, the Archard wear model is discretized into incremental form, and the incremental form formula is as follows:

[0018] in, This refers to the depth of localized wear. To contact pressure, The sliding distance within the increment step. The hardness of the roll material.

[0019] Preferably, the dynamic prediction of the remaining service life of the roll includes: dynamically predicting the remaining service life of the roll by integrating the instantaneous wear rate predicted by the high-precision surrogate model over time, and the core formula for calculating the total wear amount is:

[0020]

[0021] in, The instantaneous wear rate predicted by the high-precision surrogate model. Total wear amount;

[0022] The formula for estimating the remaining service life of the rolls is:

[0023]

[0024] in, This represents the maximum permissible wear limit of the rolls. This is the predicted average wear rate over a future period of time. For the remaining service life of the rolls, This represents the current usage time of the roll.

[0025] Preferably, the formula for dynamically predicting the roll health index is:

[0026]

[0027] Wherein, HI is the health index, ranging from 0 to 1, and α, β, γ are weighting coefficients, with α + β + γ = 1. The maximum equivalent stress of the rolls predicted by the surrogate model. The yield strength of the roll material. This is a scoring index based on the surface quality data of strip steel produced recently by this roll.

[0028] Preferably, the method for predicting the life and online health of rolls in a five-stand cold rolling mill further includes,

[0029] When a defect is detected on the surface of the strip, the defect image of the strip is transformed in the frequency domain to extract the dominant spatial wavelength;

[0030] The theoretical defect cycle is calculated by combining the real-time rotation speed of the rolls in each stand, and the set of suspected stands is screened by wavelength matching.

[0031] The theoretical wear morphology of the suspected stand rolls under the current working conditions is inverted by calling a high-precision proxy model, and periodic features are extracted;

[0032] Modal matching or cross-correlation coefficient algorithms are used to calculate the similarity between theoretical wear morphology features and measured defect features. Based on the similarity, the faulty frame and roll type are identified and maintenance suggestions are generated.

[0033] Compared with existing technologies, the present invention provides a method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill, which has the following advantages: The present invention collects production process data, roll body data, and strip quality data, and preprocesses the data to construct a relational time-series database with coil number as the primary key and roll ID as the foreign key. Based on the roll body data, a refined finite element model including work rolls, intermediate rolls, and support rolls is established using ABAQUS software, coupled with an Archard wear model. Multi-condition simulation outputs roll contact stress distribution, strain energy density, and theoretical wear data. The preprocessed production process data is used as the input feature vector, and the roll contact stress distribution, strain energy density, and theoretical wear data are used as multi-objective output vectors to construct a training sample set. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed. Based on the high-precision surrogate model, the remaining roll life and health index are dynamically predicted. By integrating real production data with high-fidelity physical simulation of the Archard model, combined with the Gaussian process regression algorithm, dynamic and accurate prediction of the remaining roll life is achieved, avoiding over-maintenance or under-maintenance.

[0034] This invention also provides an online system for predicting the life and health of rolls in a five-stand cold rolling mill, the system comprising:

[0035] The data acquisition module is used to collect production process data, roll body data and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key.

[0036] The finite element simulation module is used to build a fine finite element model of the work roll, intermediate roll and support roll based on the roll body data using ABAQUS software, coupled with the Archard wear model, and output the roll contact stress distribution, strain energy density and theoretical wear data through multi-condition simulation.

[0037] The surrogate model construction module is used to construct a training sample set by taking the preprocessed production process data as the input feature vector and the roll contact stress distribution, strain energy density and theoretical wear data as the multi-objective output vector. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed.

[0038] The online prediction module is used to dynamically predict the remaining service life and health index of the rolls based on a high-precision surrogate model.

[0039] Preferably, the production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage of each stand;

[0040] The data of the roll body includes the chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount and surface morphology of the work roll, intermediate roll and support roll;

[0041] The strip quality data includes defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics.

[0042] Preferably, the five-stand cold rolling mill roll life prediction and health online diagnosis system further includes an anomaly diagnosis module, which is used to perform frequency domain transformation on the strip defect image and extract the dominant spatial wavelength when a defect is detected on the strip surface;

[0043] The theoretical defect cycle is calculated by combining the real-time rotation speed of the rolls in each stand, and the set of suspected stands is screened by wavelength matching.

[0044] The theoretical wear morphology of the suspected stand rolls under the current working conditions is inverted by calling a high-precision proxy model, and periodic features are extracted;

[0045] Modal matching or cross-correlation coefficient algorithms are used to calculate the similarity between theoretical wear morphology features and measured defect features. Based on the similarity, the faulty frame and roll type are identified and maintenance suggestions are generated.

[0046] Compared with the prior art, the beneficial effects of the five-stand cold rolling mill roll life prediction and health online diagnosis system provided by the present invention are the same as those of the five-stand cold rolling mill roll life prediction and health online diagnosis method provided by the above-mentioned technical solutions, and will not be repeated here.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 The diagram shows a flowchart of a method for predicting the life of rolls and conducting online health diagnosis for a five-stand cold rolling mill, as provided in an embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] In this embodiment, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations, intended to present related concepts in a specific manner, and should not be construed as superior or more advantageous than other embodiments or designs.

[0052] This invention provides a method for predicting the life of rolls and conducting online health diagnosis for a five-stand cold rolling mill, and implements it on a cold rolling mill production line of a steel plant. This embodiment involves two types of strip steel with different strengths and widths: steel grade 1 is SGCC with a width of 1225 mm and a yield strength of 320 MPa; steel grade 2 is SPCC with a width of 1288 mm and a yield strength of 360 MPa. Figure 1 The diagram illustrates a flowchart of an online method for predicting the life and health of rolls in a five-stand cold rolling mill, as provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0053] Step S1: Collect production process data, roll body data and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key.

[0054] It should be noted that the steel plant collected production process data and multi-source heterogeneous data on the entire life cycle of the rolls from the basic automation system, process control system and manufacturing execution system of the cold rolling mill over the past year. After preprocessing, a training sample library containing more than 100,000 valid data records was formed.

[0055] The production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage for each stand.

[0056] The multi-source heterogeneous data for the entire life cycle of the rolls includes roll body data and strip quality data. Specifically, the roll body data includes the material chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount after removal from the mill, and surface morphology of the work roll, intermediate roll, and support roll.

[0057] The strip quality data includes the defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics recorded by the online strip surface inspection system that strictly corresponds to the production batch.

[0058] The collected production process data, roll body data, and strip quality data are cleaned, timestamp aligned, outlier handled, and normalized preprocessed to eliminate data noise and redundancy, providing high-quality data support for subsequent simulation and modeling.

[0059] Step S2: Based on the roll body data, use ABAQUS software to establish a fine finite element model including the work roll, intermediate roll and support roll, couple the Archard wear model, and output the roll contact stress distribution, strain energy density and theoretical wear data through multi-condition simulation.

[0060] It should be noted that, based on the roll body data collected in step S1, and considering the commonly used high-chromium steel work roll material and specifications of this rolling line, a refined finite element model including the work roll, intermediate roll, and support roll is established using ABAQUS finite element analysis software. Specifically, the refined finite element model is a dynamic explicit analysis model that considers the elastic deformation of the roll, used to accurately simulate the transient processes of strip biting, stable rolling, and strip ejection.

[0061] The data collected in step S1 is used as loads and boundary conditions, which are applied to the refined finite element model for simulation calculation. The initial wear coefficient is set to K = 2.5 × 10⁻⁻⁻⁶ in the simulation. 7 (Determined based on material testing), material hardness H=650HB, coupled with the Arcard wear model to output contact stress distribution, strain energy density and theoretical wear data of the roll under various working conditions.

[0062] Furthermore, the basic formula of the Archard wear model is:

[0063]

[0064] Where V is the wear volume and K is the wear coefficient. Where L is the normal load on the contact surface, L is the sliding distance, and H is the hardness of the roll material.

[0065] In the refined finite element model simulation, the Archard wear model is discretized into incremental form, and the incremental form formula is as follows:

[0066] in, This refers to the depth of localized wear. To contact pressure, The sliding distance within the increment step. The hardness of the roll material.

[0067] However, the wear coefficient K is not a constant, but is calibrated by inverse method. That is, by comparing the actual grinding amount of the roll after it comes off the mill with the wear amount predicted by simulation, a mapping relationship between the K value and the lubrication conditions (emulsion flow rate and concentration) is established, so as to realize the online adaptive correction of the Archard wear model.

[0068] Step S3: Using the preprocessed production process data as the input feature vector, and the roll contact stress distribution, strain energy density, and theoretical wear data as the multi-objective output vector, a training sample set is constructed. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is built.

[0069] It should be noted that the production process data, roll body data and strip quality data after preprocessing in step S1 are used as input feature vector X, and the roll contact stress distribution, strain energy density and theoretical wear data obtained from simulation in step S2 are used as multi-objective output vector Y to construct a training sample set.

[0070] A Gaussian process multi-objective regression algorithm is employed to establish an independent regression model for each objective variable in the output vector Y. The kernel function uses a combination of radial basis functions and white noise functions to simultaneously capture global trends and local noise. Maximum likelihood estimation is used to optimize the kernel function hyperparameters, and K-fold cross-validation is employed to evaluate the model's generalization ability. Ultimately, a high-precision surrogate model is obtained that can quickly map production process data to roll contact stress distribution, strain energy density, and theoretical wear data. This high-precision surrogate model can rapidly calculate the roll state under given production process data.

[0071] Furthermore, a Gaussian process multi-objective regression algorithm is employed to establish an independent regression model for each objective variable in the output vector Y. The prediction results of the established Gaussian process regression model include not only the predicted value y of each objective variable, but also the uncertainty variance of the predicted value. The system marks regions where the prediction uncertainty exceeds a set threshold as high uncertainty regions. When online prediction enters this region, the system automatically alerts to a decrease in reliability and suggests adopting a conservative maintenance strategy or initiating a recalculation based on finite element simulation for verification.

[0072] Step S4: Based on a high-precision surrogate model, dynamically predict the remaining service life and health index of the rolls.

[0073] It should be noted that the well-trained high-precision surrogate model is integrated into the online monitoring system of the cold continuous rolling production line. Real-time data on rolling force, bending force, and rolling speed of each stand are collected and input into the high-precision surrogate model. The model dynamically predicts and displays the remaining service life (RUL) or health index (HI) of the rolls on each stand in real time on the monitoring interface.

[0074] Furthermore, the remaining useful life (RUL) prediction is achieved by integrating the instantaneous wear rate predicted by the high-precision surrogate model over time. The core formula for calculating the total wear amount is as follows:

[0075]

[0076] in, The instantaneous wear rate predicted by the high-precision surrogate model. Total wear amount;

[0077] The remaining service life of the rolls is estimated using the following formula:

[0078]

[0079] in, This represents the maximum permissible wear limit of the rolls. This is the predicted average wear rate over a future period of time. For the remaining service life of the rolls, This represents the current usage time of the roll.

[0080] Furthermore, the roll health index is a comprehensive evaluation index that integrates multiple factors, and its calculation formula is as follows:

[0081]

[0082] Wherein, HI is the health index, ranging from 0 to 1, and α, β, γ are weighting coefficients, with α + β + γ = 1. The maximum equivalent stress of the rolls predicted by the surrogate model. The yield strength of the roll material. This is a scoring index based on the surface quality data of strip steel produced recently by this roll.

[0083] As one possible approach, when a defect is detected on the strip surface, the following steps are performed to diagnose and locate it:

[0084] Frequency domain transformation is performed on the defect image of the strip steel to extract the dominant spatial wavelength λ.

[0085] The theoretical defect cycle is calculated by combining the real-time rotational speed of each stand's rolls. The suspected stand set {Suspicious_Stand} is initially screened by wavelength matching. Specifically, wavelength matching involves calculating the circumference of the work roll or intermediate roll of each stand using the formula Ci=π×Di, where Di is the current roll diameter and Ci is the circumference of the work roll or intermediate roll of the stand. The dominant defect wavelength λ is compared with Ci / n, where n is the harmonic order, usually taken as 1 or 2. If |λ−Ci / n|<ϵ (ϵ is the allowable error threshold), then stand i is included in the suspected stand set.

[0086] A high-precision proxy model is invoked to inversely calculate the theoretical wear morphology of the suspected stand rolls under the current working conditions, and its periodic characteristics are extracted.

[0087] By employing modal matching or cross-correlation coefficient algorithms, the similarity between theoretical wear morphology features and measured defect features is calculated. The specific stand and roll type that cause quality problems is identified based on the highest similarity, and targeted roll replacement and maintenance suggestions are generated.

[0088] For example, during production, the strip surface inspection instrument alarmed, detecting a bright and dark stripe defect with a spacing of approximately 15mm. The defect image was frequency-domain transformed to extract the dominant spatial wavelength λ≈15mm. The circumference of the work rolls and intermediate rolls of each stand was calculated, and λ was compared with Ci / n, initially identifying the F3 stand as a suspect set. A high-precision proxy model was invoked, and the theoretical wear cycle of the work rolls of the F3 stand was calculated based on the current production process data, approximately 14.8mm. The cross-correlation coefficient algorithm was used to calculate the similarity between the theoretical wear morphology features and the measured defect features, achieving a similarity of over 98%, thus pinpointing the fault source. The diagnostic interface concluded that uneven wear of the work rolls under the F3 stand was suspected, recommending priority inspection and replacement, and guiding maintenance personnel to perform precise operations. Implementation verification showed that this method significantly improved roll management efficiency, reduced product defect rate and production costs, achieving good economic benefits.

[0089] Compared with the prior art, the method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill provided by the embodiments of the present invention has the following beneficial effects:

[0090] 1. By integrating real production data with high-fidelity physical simulation of the Archard wear model, and combining Gaussian process regression algorithm and integral calculation, dynamic and accurate prediction of the remaining life of the rolls is achieved. Relying on authoritative wear models and mathematical integration methods, solid physical and mathematical support is provided for the prediction process, enabling the roll replacement plan to be upgraded from timely roll replacement to on-demand roll replacement, effectively avoiding over-maintenance or under-maintenance, and significantly improving the overall efficiency of the equipment.

[0091] 2. To address strip surface quality issues, a coherent technical approach involving defect feature extraction, wavelength matching, and similarity calculation is employed to quickly and accurately pinpoint faulty stands and corresponding rolls. This significantly reduces the time required for traditional piecemeal inspections, substantially improves production efficiency, and provides core support for the scientific and intelligent management of rolls.

[0092] 3. An innovative inverse method is adopted to establish a correlation between the wear coefficient K and the lubrication conditions, realizing online adaptive correction of the Archard wear model. This enables it to flexibly adapt to production scenarios with different strip steel specifications, steel grades and process parameters, breaking through the applicability limitations of traditional models.

[0093] 4. The technology is highly portable and can be adapted to cold rolling mills with various parameters. It can be implemented with only minor modifications to existing technologies. It effectively reduces product defects caused by untimely roll replacement, improves the surface quality of strip steel while reducing production costs, and has significant practical value and economic benefits.

[0094] This invention also provides a five-stand cold rolling mill roll life prediction and online health diagnosis system, the system comprising:

[0095] The data acquisition module is used to collect production process data, roll body data and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key.

[0096] The finite element simulation module is used to build a fine finite element model of the work roll, intermediate roll and support roll based on the roll body data using ABAQUS software, coupled with the Archard wear model, and output the roll contact stress distribution, strain energy density and theoretical wear data through multi-condition simulation.

[0097] The surrogate model construction module is used to construct a training sample set by taking the preprocessed production process data as the input feature vector and the roll contact stress distribution, strain energy density and theoretical wear data as the multi-objective output vector. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed.

[0098] The online prediction module is used to dynamically predict the remaining service life and health index of the rolls based on a high-precision surrogate model.

[0099] Preferably, the production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage of each stand;

[0100] The data of the roll body includes the chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount and surface morphology of the work roll, intermediate roll and support roll;

[0101] The strip quality data includes defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics.

[0102] Preferably, the five-stand cold rolling mill roll life prediction and health online diagnosis system further includes an anomaly diagnosis module, which is used to perform frequency domain transformation on the strip defect image and extract the dominant spatial wavelength when a defect is detected on the strip surface;

[0103] The theoretical defect cycle is calculated by combining the real-time rotation speed of the rolls in each stand, and the set of suspected stands is screened by wavelength matching.

[0104] The theoretical wear morphology of the suspected stand rolls under the current working conditions is inverted by calling a high-precision proxy model, and periodic features are extracted;

[0105] Modal matching or cross-correlation coefficient algorithms are used to calculate the similarity between theoretical wear morphology features and measured defect features. Based on the similarity, the faulty frame and roll type are identified and maintenance suggestions are generated.

[0106] Compared with the prior art, the beneficial effects of the five-stand cold rolling mill roll life prediction and health online diagnosis system provided in this embodiment of the invention are the same as those of the five-stand cold rolling mill roll life prediction and health online diagnosis method provided in the above technical solution, and will not be repeated here.

[0107] Furthermore, this invention also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are connected via a bus. When the computer program is executed by the processor, it implements the various processes of the above-described embodiment of the method for predicting the life of rolls and conducting online health diagnosis of a five-stand cold rolling mill, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0108] Furthermore, this embodiment of the invention also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the various processes of the above-described embodiment of the method for predicting the life of rolls and conducting online health diagnosis of a five-stand cold rolling mill, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill, wherein... The feature is that it includes the following steps: Collect production process data, roll body data, and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key; Based on the roll body data, a fine finite element model including the work roll, intermediate roll and support roll was established using ABAQUS software. The model was coupled with the Archard wear model, and the roll contact stress distribution, strain energy density and theoretical wear data were output through multi-condition simulation. The preprocessed production process data is used as the input feature vector, and the roll contact stress distribution, strain energy density and theoretical wear data are used as the multi-objective output vector to construct a training sample set. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed. Based on a high-precision surrogate model, the remaining service life and health index of the rolls are dynamically predicted.

2. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage for each stand. The data of the roll body includes the chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount and surface morphology of the work roll, intermediate roll and support roll; The strip quality data includes defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics.

3. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The refined finite element model is a dynamic explicit analysis model that considers the elastic deformation of the rolls, used to simulate the transient processes of strip biting, stable rolling, and strip ejection.

4. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The formula for the Archard wear model is: Where V is the wear volume and K is the wear coefficient. Where L is the normal load on the contact surface, L is the sliding distance, and H is the hardness of the roll material; In the refined finite element model simulation, the Archard wear model is discretized into incremental form, and the incremental form formula is as follows: in, This refers to the depth of localized wear. To contact pressure, The sliding distance within the increment step. The hardness of the roll material.

5. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The dynamic prediction of the remaining service life of the rolls includes: dynamically predicting the remaining service life of the rolls by integrating the instantaneous wear rate predicted by the high-precision surrogate model over time. The core formula for calculating the total wear amount is: in, The instantaneous wear rate predicted by the high-precision surrogate model. Total wear amount; The formula for estimating the remaining service life of the rolls is: in, This represents the maximum permissible wear limit of the rolls. This is the predicted average wear rate over a future period of time. For the remaining service life of the rolls, This represents the current usage time of the roll.

6. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The formula for dynamically predicting the health index of the rolls is: Wherein, HI is the health index, ranging from 0 to 1, and α, β, γ are weighting coefficients, with α + β + γ = 1. The maximum equivalent stress of the rolls is predicted by the surrogate model. The yield strength of the roll material. This is a scoring index based on the surface quality data of strip steel produced recently by this roll.

7. The method for predicting the life and online health diagnosis of rolls in a five-stand cold rolling mill according to claim 1, characterized in that, The method for predicting the life and online health of rolls in a five-stand cold rolling mill also includes... When a defect is detected on the surface of the strip, the defect image of the strip is transformed in the frequency domain to extract the dominant spatial wavelength; The theoretical defect cycle is calculated by combining the real-time rotation speed of the rolls in each stand, and the set of suspected stands is screened by wavelength matching. The theoretical wear morphology of the suspected stand rolls under the current working conditions is inverted by calling a high-precision proxy model, and periodic features are extracted; Modal matching or cross-correlation coefficient algorithms are used to calculate the similarity between theoretical wear morphology features and measured defect features. Based on the similarity, the faulty frame and roll type are identified and maintenance suggestions are generated.

8. A five-stand cold rolling mill roll life prediction and health online diagnosis system, characterized in that, include: The data acquisition module is used to collect production process data, roll body data and strip quality data, and preprocess the data to construct a relational time-series database with the coil number as the primary key and the roll ID as the foreign key. The finite element simulation module is used to build a fine finite element model of the work roll, intermediate roll and support roll based on the roll body data using ABAQUS software, coupled with the Archard wear model, and output the roll contact stress distribution, strain energy density and theoretical wear data through multi-condition simulation. The surrogate model construction module is used to construct a training sample set by taking the preprocessed production process data as the input feature vector and the roll contact stress distribution, strain energy density and theoretical wear data as the multi-objective output vector. Based on the Gaussian process multi-objective regression algorithm, a high-precision surrogate model is constructed. The online prediction module is used to dynamically predict the remaining service life and health index of the rolls based on a high-precision surrogate model.

9. The online roll life prediction and health diagnosis system for a five-stand cold rolling mill according to claim 8, characterized in that, The production process data includes the real-time rolling force, bending roll force, rolling speed, inter-stand tension, rolling kilometers, and rolling tonnage for each stand. The data of the roll body includes the chemical composition, microstructure, initial geometric dimensions, initial surface roughness, measured grinding amount and surface morphology of the work roll, intermediate roll and support roll; The strip quality data includes defect type, defect image grayscale distribution, defect location coordinates, and defect periodicity characteristics.

10. The online roll life prediction and health diagnosis system for a five-stand cold rolling mill according to claim 8, characterized in that, The five-stand cold rolling mill roll life prediction and health online diagnosis system also includes an anomaly diagnosis module, which is used to perform frequency domain transformation on the strip defect image and extract the dominant spatial wavelength when a defect is detected on the strip surface. The theoretical defect cycle is calculated by combining the real-time rotation speed of the rolls in each stand, and the set of suspected stands is screened by wavelength matching. The theoretical wear morphology of the suspected stand rolls under the current working conditions is inverted by calling a high-precision proxy model, and periodic features are extracted; Modal matching or cross-correlation coefficient algorithms are used to calculate the similarity between theoretical wear morphology features and measured defect features. Based on the similarity, the faulty frame and roll type are identified and maintenance suggestions are generated.

Images