A method for model-based adaptive learning for protecting a main shaft of a rolling mill
By establishing a data relationship model between rolling torque and workpiece reduction, the rolling roll gap can be adjusted in real time, solving the problem of easy damage to the mill spindle, realizing reliable equipment protection and production continuity, and reducing failure risk and energy waste.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING IRON & STEEL CO LTD
- Filing Date
- 2023-05-31
- Publication Date
- 2026-07-31
AI Technical Summary
In the current technology for rolling high-strength steel plates, pipeline steel, bridge steel and other steel materials, the long-term high-load operation of the equipment can easily cause structural damage to the mill spindle, leading to sudden failures and rolling interruptions, which affects production organization.
By establishing a data relationship model between rolling torque and workpiece reduction, the rolling roll gap is adjusted in real time. The model's adaptive learning is used to protect the mill spindle and prevent the rolling torque from exceeding the limit. Combined with the secondary mill control model, the workpiece thickness is adjusted to ensure that the workpiece dimensions meet the requirements.
It effectively reduces unexpected failures, ensures production quality and equipment integrity, reduces energy waste, and improves production efficiency and safety.
Smart Images
Figure CN116586436B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of rolling process technology, and in particular to a method for protecting the main shaft of a rolling mill based on model adaptive learning. Background Technology
[0002] When rolling high-strength steel plates, pipeline steel, bridge steel, crack-arresting steel, and other steel products, the mill's main transmission torque becomes excessive and the load unbalanced due to prolonged high-load operation. This can easily lead to structural damage in some parts of the equipment, causing sudden failures. For example, the mill spindle is highly susceptible to internal damage under long-term impact loads. Currently, the only way to protect the equipment from abnormal damage is to roughly interrupt the rolling process. This can result in abnormal rolling interruptions, causing unplanned rolling and negatively impacting production organization. Summary of the Invention
[0003] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method for protecting the main shaft of a rolling mill based on model adaptive learning.
[0004] To solve the above technical problems, the technical solution of the present invention is as follows:
[0005] A method for protecting the main shaft of a rolling mill based on model adaptive learning includes:
[0006] Obtain the parameters of the rolled piece;
[0007] Establish and train a data relationship model between rolling torque and workpiece reduction;
[0008] When the instantaneous rolling torque exceeds a preset threshold, the difference between the instantaneous rolling torque and the preset threshold is calculated. Based on the data relationship model between the rolling torque and the reduction of the rolled piece, the adjustment value of the reduction of the rolled piece is obtained. The rolling roll gap is adjusted according to the adjustment value, and the rolling adjustment parameters are recorded.
[0009] The thickness parameters of the rolled piece are acquired in real time and transmitted to the secondary rolling mill control model;
[0010] The secondary rolling mill control model adjusts the rolling roll gap in real time according to the thickness curve of the rolled piece to ensure that the dimensions of the rolled piece meet the requirements, and records the rolling correction parameters.
[0011] The secondary mill control model is trained based on the parameters of the rolled piece, the rolling adjustment parameters, and the rolling correction parameters to obtain the rolling process parameters of the corresponding rolled piece.
[0012] The secondary rolling mill control model obtains the corresponding rolling process parameters based on the parameters of the workpiece to be rolled, and rolls the workpiece accordingly.
[0013] As a preferred embodiment of the model-adaptive learning-based method for protecting the mill spindle described in this invention, the parameters of the rolled piece include the type, width, and thickness of the rolled piece.
[0014] As a preferred embodiment of the model-adaptive learning-based method for protecting the mill spindle described in this invention, the method for establishing a data relationship model between rolling torque and workpiece reduction includes:
[0015] The workpiece is rolled in several passes, and the rolling torque and the amount of reduction of the workpiece in each pass are recorded.
[0016] The rolling torque variation and workpiece reduction variation of two adjacent passes are obtained, and a data model relating rolling torque and workpiece reduction is established based on this.
[0017] As a preferred embodiment of the model-adaptive learning-based method for protecting the mill spindle described in this invention, the training model for the data relationship between rolling torque and workpiece reduction includes:
[0018] Based on the data model relating rolling torque and workpiece reduction, the preset change value of workpiece reduction and its corresponding preset change value of rolling torque are obtained.
[0019] Adjust the rolling roll gap based on the preset change value of the workpiece reduction to obtain the actual change value of the rolling torque;
[0020] Determine whether the deviation between the actual change value of the rolling torque and the preset change value of the rolling torque exceeds 5% of the preset change value of the rolling torque. If so, use it as sample data to self-correct the data model of the relationship between the rolling torque and the reduction of the rolled piece. Otherwise, keep the original model parameters.
[0021] As a preferred embodiment of the model-adaptive learning-based method for protecting the mill spindle described in this invention, the real-time acquisition of the workpiece thickness parameters includes:
[0022] The thickness variation curve of the rolled piece is obtained in real time through the primary rolling mill control system.
[0023] As a preferred embodiment of the model-adaptive learning-based method for protecting the main shaft of a rolling mill according to the present invention, the training times of the secondary rolling mill control model are greater than or equal to 100 times.
[0024] The beneficial effects of this invention are:
[0025] (1) The present invention adjusts the rolling roll gap in real time according to the rolling torque to avoid the main drive rolling torque from exceeding the limit, effectively reducing sudden failures and lowering the safety risks brought about by emergency repairs.
[0026] (2) The present invention can guarantee rolling accuracy and avoid rolling dimensions not meeting requirements due to equipment protection, thereby ensuring reliable production quality while ensuring the equipment is intact.
[0027] (3) This invention can reduce energy waste caused by equipment failure and increase the amount of carbon neutrality tasks that enterprises can accomplish. Attached Figure Description
[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.
[0029] Figure 1 This is a flowchart illustrating the method for protecting the mill spindle based on model adaptive learning provided by the present invention. Detailed Implementation
[0030] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.
[0031] Example 1: Figure 1 This is a flowchart illustrating a model-adaptive learning-based method for protecting the mill spindle, as provided in an embodiment of this application. The method includes steps S101 to S107, which are described in detail below:
[0032] Step S101: Obtain the parameters of the rolled piece.
[0033] Specifically, before rolling the workpiece, the primary rolling mill control system acquires and stores the workpiece parameters. These parameters include, but are not limited to, the workpiece type (material type), width, and thickness.
[0034] Step S102: Establish and train a data relationship model between rolling torque and workpiece reduction.
[0035] Specifically, the first step is to establish a data relationship model between rolling torque and workpiece reduction. The method for establishing this model is as follows:
[0036] Step S102a: Perform several passes of rolling on the workpiece and record the rolling torque and workpiece reduction for each pass.
[0037] Step S102b: Obtain the changes in rolling torque and workpiece reduction between two adjacent passes, and establish a data model relating rolling torque and workpiece reduction based on this data.
[0038] The established data relationship model between rolling torque and workpiece reduction was then trained to improve its accuracy. The specific training method is as follows:
[0039] Step S102c: Obtain the preset change value of the rolling reduction and the corresponding preset change value of the rolling torque based on the data model of the relationship between rolling torque and rolling reduction.
[0040] Step S102d: Adjust the rolling roll gap based on the preset change value of the workpiece reduction to obtain the actual change value of the rolling torque.
[0041] Step S102e: Determine whether the deviation between the actual change value of the rolling torque and the preset change value of the rolling torque exceeds 5% of the preset change value of the rolling torque. If so, use it as sample data to self-correct the data model of the relationship between the rolling torque and the reduction of the rolled piece. Otherwise, keep the original model parameters.
[0042] Step S103: Roll the workpiece. When the instantaneous rolling torque exceeds the preset threshold, calculate the difference between the instantaneous rolling torque and the preset threshold. Based on the data relationship model between the rolling torque and the workpiece reduction, obtain the adjustment value of the workpiece reduction. Adjust the rolling roll gap according to the adjustment value and record the rolling adjustment parameters.
[0043] Specifically, during the rolling process, when the instantaneous torque exceeds a preset threshold, based on the magnitude of the excess and the data model relating rolling torque to workpiece reduction in step 102, the hydraulic roll gap control system can be used to adjust the roll gap to unload the load, thereby reducing the rolling load to the level allowed by the equipment. In this embodiment, the maximum increase in roll gap is 1 mm. In this embodiment, the preset threshold for instantaneous torque is 2.3 times the main transmission torque of the rolling mill.
[0044] The recorded rolling adjustment parameters include the adjustment amount of the rolling roll gap and the adjustment time of the rolling roll gap.
[0045] It should be noted that the adjustment value for the reduction of the rolled piece is equal to the adjustment value for the roll gap.
[0046] Step S104: Obtain the thickness parameters of the rolled piece in real time and transmit them to the secondary rolling mill control model.
[0047] Specifically, a communication channel is established between the primary rolling mill control system and the secondary rolling mill control model. Because the torque exceeding the limit in step S103 caused the roll gap to be adjusted, it will result in uneven longitudinal thickness of the rolled piece. Through this communication channel, the thickness curve of the rolled piece can be fed back to the secondary rolling mill control model in real time via the primary rolling mill control system.
[0048] Step S105: The secondary rolling mill control model adjusts the rolling roll gap in real time according to the thickness curve of the rolled piece to ensure that the dimensions of the rolled piece meet the requirements, and records the rolling correction parameters.
[0049] Specifically, based on the thickness variation curve fed back by the primary rolling mill control system, the secondary rolling mill control model will, within the allowable range of equipment and process, increase the allowable reduction per pass by means of load reduction, i.e., adjust the rolling roll gap, in order to compensate for the impact of unloading in the previous pass on the thickness of the rolled piece, and ensure the controllability of rolling dimensions and rolling process.
[0050] The rolling correction parameters include the adjustment amount of the rolling roll gap and the adjustment time of the rolling roll gap in each pass.
[0051] Step S106: Train the secondary mill control model based on the parameters of the workpiece, rolling adjustment parameters, and rolling correction parameters to obtain the rolling process parameters of the corresponding workpiece.
[0052] Specifically, the rolling parameters, rolling adjustment parameters, and rolling correction parameters of each rolled piece are used as a set of sample data to train the secondary rolling mill control model. The training iterations are greater than or equal to 100 to ensure the accuracy of the model control.
[0053] Step S107: The secondary rolling mill control model obtains the corresponding rolling process parameters based on the parameters of the workpiece to be rolled, and rolls the workpiece accordingly.
[0054] Example 2: This example provides a model-based adaptive learning method for protecting the mill spindle. The difference from Example 1 is that the maximum increase in roll gap when torque exceeds the limit is 2mm. After implementation, the rolling process of 100 steel plates of the same specification was continuously monitored.
[0055] Example 3: This example provides a method for protecting the mill spindle based on model adaptive learning. The difference from Example 1 is that the preset threshold for instantaneous torque is 2.0 times the main transmission torque of the mill. After implementation, the rolling process of 100 steel plates of the same specification was continuously tracked.
[0056] Therefore, the technical solution of this application automatically opens the roll gap when the torque exceeds the limit, and then feeds back the automatic opening status of the roll gap to the model for self-learning modification and compensation, thus ensuring reliable production quality while ensuring the equipment is in good working order.
[0057] In addition to the above embodiments, the present invention may have other implementation methods; all technical solutions formed by equivalent substitution or equivalent transformation fall within the protection scope claimed by the present invention.
Claims
1. A method for protecting the main shaft of a rolling mill based on model adaptive learning, characterized in that: include: Obtain the parameters of the rolled piece; Establish and train a data relationship model between rolling torque and workpiece reduction; wherein, establishing a data relationship model between rolling torque and workpiece reduction includes: rolling the workpiece in several passes and recording the rolling torque and workpiece reduction in each pass; obtaining the change in rolling torque and the change in workpiece reduction between two adjacent passes, and using this to establish a data relationship model between rolling torque and workpiece reduction; When the instantaneous rolling torque exceeds a preset threshold, the difference between the instantaneous rolling torque and the preset threshold is calculated. Based on the data relationship model between the rolling torque and the reduction of the rolled piece, the adjustment value of the reduction of the rolled piece is obtained. The rolling roll gap is adjusted according to the adjustment value, and the rolling adjustment parameters are recorded. The thickness parameters of the rolled piece are acquired in real time, which includes: acquiring the thickness change curve of the rolled piece in real time through the primary rolling mill control system and transmitting it to the secondary rolling mill control model; The secondary rolling mill control model adjusts the rolling roll gap in real time according to the thickness curve of the rolled piece to ensure that the dimensions of the rolled piece meet the requirements, and records the rolling correction parameters; among them, the real-time adjustment of the rolling roll gap by the secondary rolling mill control model according to the thickness curve of the rolled piece includes: adjusting the rolling roll gap by means of load reduction; The secondary mill control model is trained based on the parameters of the rolled piece, the rolling adjustment parameters, and the rolling correction parameters to obtain the rolling process parameters of the corresponding rolled piece. The secondary rolling mill control model obtains the corresponding rolling process parameters based on the parameters of the workpiece to be rolled, and rolls the workpiece accordingly.
2. The method of model based adaptive learning for protection of rolling mill spindle as claimed in claim 1 wherein: The parameters of the rolled piece include the type, width, and thickness of the rolled piece.
3. The method of model based adaptive learning for protection of rolling mill spindle as claimed in claim 1 wherein: The training model for the data relationship between rolling torque and workpiece reduction includes: Based on the data model relating rolling torque and workpiece reduction, the preset change value of workpiece reduction and its corresponding preset change value of rolling torque are obtained. Adjust the rolling roll gap based on the preset change value of the workpiece reduction to obtain the actual change value of the rolling torque; Determine whether the deviation between the actual change value of the rolling torque and the preset change value of the rolling torque exceeds 5% of the preset change value of the rolling torque. If so, use it as sample data to self-correct the data model of the relationship between the rolling torque and the reduction of the rolled piece. Otherwise, keep the original model parameters.
4. The method of model based adaptive learning for protection of rolling mill spindle as claimed in claim 1 wherein: The training times for the secondary rolling mill control model are greater than or equal to 100.