Self-adaptive setting method and system for bar straightening process parameters
By constructing a database of bar straightening process parameters and using machine learning algorithms to train a prediction model, the automatic and intelligent setting of bar straightening process parameters is achieved. This solves the problem of relying on manual experience in existing technologies, improves production efficiency and product quality stability, and adapts to changes in complex working conditions.
Patent Information
- Application Number
- CN202511686915.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-18
- Publication Date
- 2026-02-27
AI Technical Summary
The existing bar straightening process parameters are highly dependent on human experience, resulting in unstable product quality, low production efficiency, difficulty in inheritance and optimization, and inability to adapt to complex working conditions. The existing models have not fully considered the influence of material properties and geometry.
A database of bar straightening process parameters is constructed, and a prediction model is trained using machine learning algorithms. The optimal straightening process parameters are automatically set based on real-time input feature parameters. The system includes modules for data management, model training and prediction, real-time data acquisition, and control command issuance, thereby achieving automated and intelligent parameter setting.
It reduces reliance on human experience, improves the stability and consistency of straightening quality, reduces setup time, lowers scrap rate and production costs, enhances the flexibility and adaptability of the production line, and forms corporate knowledge assets.
Smart Images
Figure CN121580466A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of metal material processing, and particularly to a method and system for self-adaptive setting of straightening process parameters of a bar based on material performance and cross-sectional shape. BACKGROUND
[0002] As a basic industrial raw material, bars are widely used in mechanical manufacturing, construction, automobile industry, etc. After rolling or heat treatment, bars often bend due to internal stress, uneven cooling, etc., and need to be straightened to ensure that their straightness meets the standard. The straightening process is a complex elastic-plastic deformation process, and the setting of its process parameters (such as the reduction of each straightening roller, the inclination angle, etc.) directly affects the straightening quality and efficiency.
[0003] At present, the bar straightening operation of most domestic steel enterprises still highly depends on the experience of operators. Operators set parameters on the straightening machine according to the basic information of the bar material, diameter, etc. and their personal experience.
[0004] This method has the following significant disadvantages:
[0005] 1. High dependence and poor stability: The product quality is subject to the technical level and working state of the workers, and the parameters set by different workers differ greatly, leading to fluctuations in product quality.
[0006] 2. Long machine adjustment time and low efficiency: When a new material grade or specification is changed, a long time of trial straightening and parameter adjustment is needed, which reduces the production efficiency and increases the scrap rate.
[0007] 3. Difficult to pass on and optimize knowledge: The experience of senior workers is a kind of "implicit knowledge" that is difficult to quantify, record and standardize, which is not conducive to the accumulation and inheritance of technology.
[0008] 4. Unable to adapt to complex working conditions: For new materials or non-standard specifications of bars, as well as dynamic changes in bar temperature, ovality, etc., manual experience often cannot quickly and accurately respond.
[0009] In the prior art, there are also some theoretical formulas or expert systems for calculating straightening process parameters, but these methods usually only consider a few factors such as diameter, the model is too simplified, and key variables such as the change of material yield strength with temperature and the influence of cross-sectional ovality on contact state are not fully considered, resulting in limited prediction accuracy and practicality.
[0010] Therefore, developing an intelligent method that can comprehensively consider multiple variables such as material performance, geometric shape and working conditions, and automatically and accurately set straightening process parameters, has become a technical problem to be solved in the field. SUMMARY
[0011] The technical problem solved by the present application is to provide a bar straightening process parameter adaptive setting method and system to reduce the dependence on manual experience and realize intelligent, precise and standardized setting of straightening process parameters.
[0012] The technical solution adopted by the present application to solve the technical problem is: a bar straightening process parameter adaptive setting method, comprising the following steps,
[0013] S1: constructing a bar straightening process parameter database, the database at least stores historical production data, each data record includes: input characteristic parameters of the bar and corresponding optimal straightening process parameters; wherein the input characteristic parameters include at least four of material grade, real-time yield strength, bar temperature, nominal diameter and measured ovality; the optimal straightening process parameters include the reduction, inclination and driving speed of each straightening roll of the straightening machine;
[0014] S2: based on the database, a straightening process parameter prediction model is trained by using a machine learning algorithm, the model takes the input characteristic parameters as independent variables and the optimal straightening process parameters as dependent variables;
[0015] S3: when straightening a new bar, the real-time input characteristic parameters of the bar are collected and input into the straightening process parameter prediction model;
[0016] S4: the model outputs the predicted optimal straightening process parameter combination, and automatically issues the parameter combination to the control system of the straightening machine to complete the straightening operation.
[0017] Further, in step S1 of the present application, the bar straightening process parameter database also stores a straightening quality evaluation index corresponding to each data record, and the index includes straightening straightness and surface residual stress; in step S2, the training target of the machine learning algorithm is to optimize the predicted process parameters under the premise of meeting the preset straightening quality evaluation index threshold.
[0018] Further, in step S2 of the present application, the machine learning algorithm is one or a combination of random forest, gradient boosting decision tree, support vector regression or artificial neural network.
[0019] Further, in step S3 of the present application, the real-time yield strength is obtained by querying a material performance database through the material grade and bar temperature; or is obtained by converting the data measured by an online hardness instrument installed on the production line.
[0020] Further, the application further comprises a parameter verification step before the parameter combination is issued to the straightening machine in step S4: comparing the model output parameters with the preset safety threshold range, if it is out of range, the boundary value within the safety threshold range is replaced, and an alarm is issued.
[0021] Meanwhile, the application also provides a system for realizing adaptive setting of bar straightening process parameters, comprising:
[0022] a data management module for constructing, storing, updating and querying the bar straightening process parameter database;
[0023] a model training and prediction module integrated with the machine learning algorithm, for training the straightening process parameter prediction model and using the model for real-time prediction;
[0024] a real-time data acquisition module for acquiring real-time input characteristic parameters of new bars and transmitting them to the model training and prediction module;
[0025] a control instruction issuing module for receiving the optimal straightening process parameter combination predicted by the model and converting it into control instructions executable by the straightening machine and issuing them to the straightening machine.
[0026] Further, the real-time data acquisition module of the application comprises an on-line diameter and ovality measuring instrument, an infrared temperature measuring instrument and / or a data interface for communication with the production manufacturing execution system of the upstream process.
[0027] Still further, the system of the application further comprises a man-machine interface for displaying the model-predicted parameters and the expected straightening quality results and allowing the operator to confirm or revise them.
[0028] The application has the advantages of solving the defects in the background art,
[0029] 1. The automatic setting of straightening process parameters is realized by constructing a prediction model, greatly reducing the dependence on the personal experience of the operator and reducing the influence of human factors on the straightening quality, and realizing the automatic process from parameter prediction to instruction issuing, thereby improving the production automation level.
[0030] 2. The process database covers the key variables affecting the straightening quality, and the prediction model trained based on the machine learning algorithm can accurately mine the complex nonlinear relationship between the input characteristics and the process parameters, and the output optimal process parameters can effectively ensure the straightening quality. In actual application, the straightening bar has a reduced straightness fluctuation range, the surface residual stress is controlled within a reasonable range, and the product quality stability and consistency are significantly improved.
[0031] 3. No lengthy manual trial straightening and parameter adjustments are required. New bars can be quickly set up and straightened upon entering the straightening area, significantly reducing setup time and improving production efficiency. Simultaneously, reasonable process parameter settings reduce scrap rates caused by improper parameters, minimizing material waste; automated operation also reduces labor costs and energy consumption, resulting in an overall reduction in production costs.
[0032] 4. Store the straightening operation experience in the process database in the form of data. As production continues, the database will continuously accumulate new data. By retraining the model, the model performance can be continuously optimized, enabling the system to have the ability to continuously learn and improve itself, forming valuable corporate knowledge assets and adapting to the needs of long-term production development.
[0033] 5. The system can collect dynamic parameters such as temperature and ellipticity of the bar in real time, and set parameters in combination with static parameters such as material grade and yield strength. It can quickly adapt to the straightening needs of new materials and new specifications of bars, and at the same time, it can cope with dynamic changes in production conditions, enhance the flexibility of the production line, and meet diverse production requirements. Attached Figure Description
[0034] Figure 1 This is a schematic diagram of the overall process of the method of the present invention. Detailed Implementation
[0035] The present invention will now be described in further detail with reference to the accompanying drawings and preferred embodiments. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.
[0036] like Figure 1 The invention presents an adaptive setting method and system for bar straightening process parameters. Its main innovation lies in constructing a multivariate mathematical model of the straightening process that integrates material properties, cross-sectional shape, and process parameters, and establishing an intelligent parameter setting system based on this model. The system first establishes and continuously updates a process database containing key parameters such as material grade, yield strength, temperature, diameter, and ellipticity through a data acquisition and preprocessing module. Then, it uses machine learning algorithms to train the database, forming a mathematical model that accurately reflects the complex nonlinear relationship between process parameters and straightening quality (such as straightness and residual stress). In practical applications, the system automatically matches or calculates the optimal straightening parameters (such as the reduction amount and rotational speed of each straightening roller) from the database or through the model in real time based on the input real-time bar information, and then sends these parameters to the straightening machine for execution. This invention effectively reduces reliance on operator experience, improves the stability, accuracy, and intelligence of the straightening process, and is suitable for straightening bar production of various materials and specifications.
[0037] The implementation process of this method is as follows:
[0038] First stage: system construction and model training
[0039] S101: Data collection and database construction. Extract the data of material grade, production batch, historical straightening process parameters of the bar from the existing MES (Manufacturing Execution System) of the enterprise; obtain the actual roll reduction, inclination angle, driving speed and other operating parameters of each straightening roll from the PLC of the straightening machine; at the same time, combine the recorded post-straightening bar straightness, surface residual stress and other quality data to form an original data set. Clean the original data, such as removing abnormal temperature data caused by sensor failure, and reducing the roll reduction data beyond the limit of the equipment due to operation errors; for a small amount of missing yield strength data, according to the material grade and corresponding temperature, the material performance database is queried for filling. After cleaning, the data is stored in the bar straightening process parameter database according to the preset data format. A typical data record is: {material grade: 45# steel, temperature: 50℃, yield strength: 320MPa, nominal diameter: 50mm, ovality: 0.5mm, optimal parameters: [roll 1 reduction: 8.5mm, roll 2 reduction: 7.2mm, roll 3 reduction: 6.8mm, roll 1 inclination angle: 1.5°, roll 2 inclination angle: 1.2°, roll 3 inclination angle: 1.0°, driving speed: 1.2m / s], post-straightening straightness: 1.2mm / m, surface residual stress: 50MPa}.
[0040] S102: Model training. Select random forest regression algorithm as machine learning model. Select 80% of the data from the database as the training set and 20% of the data as the test set. The "material grade, temperature, yield strength, nominal diameter, measured ovality" of each record in the training set are used as feature variables X, and the "roll reduction, inclination angle, driving speed" are used as target variables Y. During model training, set the number of decision trees to 100, the maximum tree depth to 15, and use 5-fold cross-validation method to evaluate the model. By continuously adjusting the hyperparameters, such as when the number of decision trees increases from 80 to 100, the root mean square error of the model on the validation set decreases from 0.8 to 0.6 (preset threshold is 0.7), at this time the model accuracy meets the requirements. Then, use the test set to verify the trained model, the root mean square error on the test set is 0.55, which is lower than the preset threshold, the model training is completed. The trained model is deployed to the model training and prediction module in the production environment.
[0041] Second stage: online intelligent setting
[0042] S201: Real-time data acquisition. A 40Cr alloy steel bar to be straightened enters the straightening area. The infrared thermometer measures its surface temperature in real time as 58°C; the diameter and ovality online measurement instrument measures its average diameter as 60.1 mm and ovality as 0.4 mm; through communication with the upstream MES system, it is obtained that the material grade of the bar is 40Cr; the system queries the pre-stored 40Cr steel material performance-temperature relationship table (which is plotted through a large number of experimental data, recording the yield strength value of 40Cr steel at different temperatures) according to the material grade “40Cr” and the temperature “58°C”, and the yield strength at this temperature is calculated as 450 MPa by using linear interpolation method. The collected set of real-time input characteristic parameters [material grade: 40Cr, temperature: 58°C, yield strength: 450 MPa, nominal diameter: 60.1 mm, ovality: 0.4 mm] are preliminarily checked for no abnormalities, and then transmitted to the model training and prediction module.
[0043] S202: Model prediction. The model training and prediction module calls the deployed random forest regression model and inputs the received real-time input characteristic parameters into the model. The model quickly outputs the predicted optimal straightening process parameter combination through internal decision tree voting calculation, which is specifically: [roll 1 reduction: 10.2 mm, roll 2 reduction: 8.8 mm, roll 3 reduction: 7.5 mm, roll 1 inclination: 1.8°, roll 2 inclination: 1.5°, roll 3 inclination: 1.2°, driving speed: 1.0 m / s].
[0044] S203: Parameter verification and issuance. The control instruction issuance module receives the optimal straightening process parameter combination output by the model, and compares each parameter with the preset safety threshold range. Among them, the maximum reduction of each straightening roll of the straightening machine is 15 mm, the minimum driving speed is 0.5 m / s, and the adjustment range of each inclination is 0.5°-3.0°. After comparison, all parameters output by the model are within the safety threshold range. Then, the control instruction issuance module converts the parameter combination into a control instruction conforming to the OPC UA communication protocol and issues it to the PLC of the straightening machine through industrial Ethernet.
[0045] S204: execution and feedback. After the straightening machine PLC receives the control instruction, the reduction amount and the inclination angle of each straightening roll are automatically adjusted, and the driving speed is adjusted to 1.0 m / s. After the bar enters the straightening machine, the straightening operation is carried out according to the set process parameters. After straightening, the online straightness measuring instrument measures the straightening straightness to be 1.0 mm / m, and the residual stress tester measures the surface residual stress to be 45 MPa, which all meet the preset quality evaluation index threshold (straightening straightness ≤1.5 mm / m, surface residual stress ≤60 MPa). The real-time input feature parameters, optimal process parameters and post-straightening quality data of this straightening operation are fed back to the data management module, and the data management module updates the bar straightening process parameter database to provide data support for subsequent model retraining and optimization.
[0046] The above description is only a specific embodiment of the present application, and various examples do not limit the essential content of the present application. Those skilled in the art can modify or deform the previously described specific embodiments without departing from the essence and scope of the application.
Claims
1. A method for adaptive setting of process parameters for straightening of a bar, characterized in that: Comprising the following steps, S1: constructing a bar straightening process parameter database, the database at least stores historical production data, each data record includes: input characteristic parameters of the bar and corresponding optimal straightening process parameters; wherein the input characteristic parameters include at least four of material grade, real-time yield strength, bar temperature, nominal diameter, and measured ovality; the optimal straightening process parameters include the reduction, inclination and driving speed of each straightening roll of the straightening machine; S2: based on the database, a straightening process parameter prediction model is trained using a machine learning algorithm, which takes the input characteristic parameters as the independent variable and the optimal straightening process parameters as the dependent variable; S3: when straightening a new bar, the real-time input characteristic parameters of the bar are collected and input into the straightening process parameter prediction model; S4: the model outputs the predicted optimal straightening process parameter combination, which is automatically issued to the control system of the straightening machine to complete the straightening operation.
2. A method of self-adapting setting of process parameters for straightening of a bar, according to claim 1, characterized in that: In step S1, the bar straightening process parameter database also stores a straightening quality evaluation index corresponding to each data record, which includes the straightening straightness and surface residual stress; in step S2, the training target of the machine learning algorithm is to optimize the predicted process parameters while meeting the preset straightening quality evaluation index threshold.
3. A method of self-adapting setting of bar straightening process parameters according to claim 1, characterized in that: In step S2, the machine learning algorithm is one or a combination of random forest, gradient boosting decision tree, support vector regression, or artificial neural network.
4. A method of self-adapting setting of bar straightening process parameters according to claim 1, characterized in that: In step S3, the real-time yield strength is obtained by querying the material performance database based on the material grade and bar temperature; Or, it is obtained by converting the data measured by the online hardness instrument installed on the production line.
5. A method of self-adapting setting of process parameters for straightening of a bar, according to claim 1, characterized in that: In step S4, before issuing the parameter combination to the straightening machine, a parameter verification step is included: comparing the model output parameters with the preset safety threshold range, if it is out of range, replace it with the boundary value within the safety threshold range, and issue an alarm.
6. A system for implementing the method of any one of claims 1-5, characterized by: Comprising: A data management module for constructing, storing, updating and querying the bar straightening process parameter database; A model training and prediction module integrated with the machine learning algorithm for training the straightening process parameter prediction model and using the model for real-time prediction; A real-time data acquisition module for obtaining real-time input characteristic parameters of a new bar and transmitting them to the model training and prediction module; A control instruction issuing module for receiving the optimal straightening process parameter combination predicted by the model and converting it into control instructions executable by the straightening machine and issuing them to the straightening machine.
7. A system for adaptive setting of rod straightening process parameters according to claim 6, characterized in that: The real-time data acquisition module includes a diameter and ovality online measurement instrument, an infrared temperature measurement instrument, and / or a data interface for communication with the production and manufacturing execution system of the upstream process.
8. A system for adaptive setting of rod straightening process parameters according to claim 6, characterized in that: The system also includes a human-machine interface for displaying the model predicted parameters and the expected results of the straightening quality, and allowing the operator to confirm or make limited revisions.