Shearing process and shearing device for automobile steel plate spring production

Abstract

The invention relates to the technical field of automobile steel plate spring production, and provides a shearing process and a shearing device for automobile steel plate spring production, and the device integrates a front-end visual positioning unit, a rear-end reinspection visual unit, a sensing unit comprising a pressure and temperature sensor and a central controller on the basis of a rack, a feeding mechanism, a pressing mechanism and a shearing mechanism. And the central controller calculates the actual springback value on line through a feedback system formed by the double-vision unit and the sensing unit, and drives the self-learning springback value prediction model to be continuously optimized, so that the stroke of the feeding mechanism is dynamically adjusted to carry out feed-forward compensation. According to the method, the problem of springback fixed-length errors in the steel plate spring shearing process is effectively solved, and intelligent, high-precision and high-stability shearing production is achieved.

Description

Technical Field

[0001] This invention relates to the field of automotive leaf spring manufacturing technology, specifically a shearing process and shearing device for automotive leaf spring manufacturing. Background Technology

[0002] Automotive leaf springs are important elastic components in suspension systems, primarily used in commercial vehicles such as trucks and buses. They consist of several alloy spring steel plates of varying lengths but with the same curvature, stacked together and fastened to the axle at the center with U-bolts, with both ends connected to the vehicle frame via lugs or sliders.

[0003] In the processing of automotive leaf springs, high-precision fixed-length shearing is crucial for ensuring product performance. Traditional methods often rely on mechanical stops or single-point vision positioning, which are open-loop or simple feedback control methods. Due to the high elasticity of leaf springs, stress release during shearing causes springback, resulting in a final length exceeding the set value. Since the springback is affected by the nonlinear coupling of multiple variables such as material batch, clamping force, blade wear, and temperature, its variation is difficult to compensate for with a fixed formula, leading to unstable accuracy in existing automated solutions. Current technology struggles to measure and compensate for this variable online, resulting in poor product length consistency.

[0004] Therefore, it is necessary to develop a shearing process and shearing device for the production of automotive leaf springs to solve the above problems. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] A shearing device for producing automotive leaf springs includes a frame, a feeding mechanism, a clamping mechanism, a shearing mechanism, and a drive mechanism for feeding the feeding mechanism. It also includes: A front-end vision positioning unit is installed at the feeding end of the shearing mechanism to identify the initial end position of the sheet metal. A back-end re-inspection vision unit is set on the discharge side of the shearing mechanism to measure the actual length of the cut-out sheet after shearing is completed. The sensing unit is used to acquire at least one physical parameter related to the shearing process in real time; The central controller is connected to the front-end visual positioning unit, the back-end re-inspection visual unit, the sensing unit, and the drive mechanism via signals, respectively. The central controller is configured to dynamically adjust the feeding length of the drive mechanism based on the feedback signals from the front-end vision positioning unit and the back-end re-inspection vision unit, as well as the real-time data from the sensing unit, in order to compensate for the fixed-length error caused by shearing springback.

[0007] Preferably, the sensing unit includes: A pressure sensor is installed on the clamping mechanism to detect the clamping force on the plate during shearing in real time; And / or a temperature sensor, located near the blade of the shearing mechanism, for real-time detection of the blade's operating temperature.

[0008] Preferably, the clamping mechanism includes the hydraulic cylinder, which is fixedly connected to the frame via a bracket. A lifting plate is fixedly connected to the output end of the hydraulic cylinder. The lower end of the lifting plate is fixedly connected to a pressure plate via a spring rod. The pressure sensor is located at the connection position between the spring rod and the lifting plate and is used to measure the clamping force on the sheet metal during shearing.

[0009] Preferably, the central controller is configured to have: The springback calculation module is used to calculate the actual springback based on the actual length of the plate measured by the back-end re-inspection vision unit and the corresponding actual positioning length. The prediction model update module is used to update the rebound amount prediction model based on the actual rebound amount and the corresponding sensor unit data. The feeding compensation module is used to predict the springback amount of the next shear based on the updated model, and adjust the feeding target value of the drive mechanism accordingly.

[0010] Preferably, the central controller is also connected to the clamping mechanism and is configured to: dynamically adjust the output force of the clamping mechanism based on the feedback from the pressure sensor, so that the clamping force during shearing is stabilized within a set range.

[0011] Preferably, the working sequence of the front-end visual positioning unit and the back-end re-inspection visual unit is uniformly scheduled by the central controller, and the image acquisition trigger signals of both are synchronized with the encoder position signal of the drive mechanism or the action phase signal of the shearing mechanism.

[0012] Preferably, the springback prediction model is a multiple linear regression model or a neural network model; the input variables of the model include at least: the clamping force F and / or blade temperature T collected by the sensing unit, and the specification parameters of the sheet material; the output variable of the model is the predicted springback amount ΔL_ct.

[0013] Preferably, it also includes an identification module connected to the central controller for identifying batch or material codes on the sheet material; the central controller stores and manages the rebound prediction model parameters and historical data under different identifiers in partitions.

[0014] Preferably, when the rebound amount prediction model is a neural network model, the central controller is configured to use transfer learning in the initial stage, based on a pre-trained base network model, and to perform rapid fine-tuning in combination with the initial sampling data of the current production line to accelerate model convergence.

[0015] A shearing process for a shearing device used in the production of automotive leaf springs includes the following steps: S1: Positioning step, the initial end position of the board is identified by the front-end vision positioning unit; S2: Feeding step, the central controller controls the drive mechanism to drive the feeding mechanism to send the sheet to the shearing station based on the target length L_t and the predicted springback compensation amount. S3: Shearing step, control the clamping mechanism to clamp the plate, and then control the shearing mechanism to perform shearing; S4: Measurement step, the actual length L_a of the blanked plate is measured by the back-end re-inspection vision unit; S5: Update step: The central controller calculates the actual rebound amount and updates the rebound amount prediction model based on the actual length L_a, the actual positioning length L_p and the current process parameters collected by the sensing unit. S6: Prediction step, based on the updated model and real-time process parameters, predicts the springback amount of the next shearing, which is used to compensate for the next feeding step S2.

[0016] The beneficial effects of this invention are: This invention employs a dual-verification closed loop comprised of front-end and back-end vision units, combined with real-time sensing of multiple process parameters such as clamping force and blade temperature, enabling the system to accurately capture the actual springback amount of each shearing operation. The self-learning predictive model running in the central controller dynamically optimizes based on historical data, thus providing accurate feedforward compensation for the next shearing operation. This overcomes the fixed-length error caused by material springback, improves shearing accuracy, and ensures stability in long-term production. Simultaneously, the system's material identification and model partitioning management functions allow it to quickly adapt to the production of different batches of sheet metal, and, in conjunction with transfer learning technology, shorten changeover and debugging time. Attached Figure Description

[0017] 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.

[0018] in: Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2 This is a schematic diagram of the pressing mechanism and the feeding mechanism; Figure 3 This is a schematic diagram of the clamping mechanism and the shearing mechanism; Figure 4 for Figure 3 Enlarged view of the structure at point A in the middle; Figure 5 A schematic diagram of the clamping mechanism, sensing unit, and front-end vision positioning unit; Figure 6 This is the main control flowchart of the system of the present invention; Figure 7 This is a flowchart illustrating the core process of model learning and prediction in this invention. In the picture: 1. Frame; 2. Feeding mechanism; 3. Clamping mechanism; 31. Hydraulic cylinder; 32. Lifting plate; 33. Pressure plate; 34. Spring rod; 4. Shearing mechanism; 51. Front-end visual positioning unit; 52. Back-end re-inspection visual unit; 6. Drive mechanism; 7. Central controller; 8. Sensing unit; 81. Pressure sensor; 82. Temperature sensor; 9. Identification module. Detailed Implementation

[0019] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0020] Example: like Figures 1-7As shown, a shearing device for producing automotive leaf springs includes a frame 1, a feeding mechanism 2 (usually one or more sets of conveyor rollers driven by a motor), a clamping mechanism 3, a shearing mechanism 4 (the relative movement of blades driven by hydraulic or mechanical means to complete the shearing), and a drive mechanism 6 (usually driven by a servo motor or stepper motor to achieve positioning) that drives the feeding mechanism 2 to feed material. It also includes: The front-end vision positioning unit 51 (usually composed of an industrial camera, lens and matching light source, which identifies the edge of the board through image processing) is set at the feeding end of the shearing mechanism 4 and is used to identify the initial end position of the board. The back-end inspection vision unit 52 (usually composed of an industrial camera, lens and matching light source, which identifies the edge of the board through image processing) is set on the discharge side of the shearing mechanism 4 and is used to measure the actual length of the cut board after shearing. Sensing unit 8 is used to acquire at least one physical parameter related to the shearing process in real time; The central controller 7 is connected to the front-end visual positioning unit 51, the back-end re-inspection visual unit 52, the sensing unit 8, and the drive mechanism 6 via signals. The central controller 7 is configured to dynamically adjust the feeding length of the drive mechanism 6 based on the feedback signals from the front-end vision positioning unit 51 and the back-end re-inspection vision unit 52, as well as the real-time data from the sensing unit 8, in order to compensate for the fixed length error caused by shearing springback.

[0021] The sheet material enters the device, and the front-end vision positioning unit 51 (such as an industrial camera in conjunction with a line laser emitter) images the initial end of the sheet material. The central controller 7 identifies the end position coordinates P1 through image processing. The central controller 7 calculates the feeding amount based on the target length L_t and the initial compensation value (which can be 0 initially) and instructs the drive mechanism 6 to perform precise feeding. During the shearing process, pressure sensor 81 monitors the pre-pressure of clamping mechanism 3 on the plate; temperature sensor 82 monitors the temperature of the blade area of ​​shearing mechanism 4. After the material is fed into place, the central controller 7 instructs the clamping mechanism 3 to work, clamping the plate, and then instructs the hydraulic cylinder of the shearing mechanism 4 to drive the blade downward to complete the shearing. After shearing, the plates are separated. The back-end re-inspection vision unit 52 images the plates and measures their actual length L_a. This data, along with the actual positioning length L_p (calculated from feedback by the servo encoder), is fed back to the central controller 7.

[0022] Through the collaboration of the front-end visual positioning unit 51 and the back-end re-inspection visual unit 52, the system not only achieves positioning before feeding but also obtains real result feedback after each shearing. Combined with the process physical parameters collected by the sensing unit 8, the central controller 7 obtains complete data on "positioning, pressure, temperature" and "actual length," laying a data foundation for analyzing and learning the rebound pattern and realizing intelligent dynamic compensation. This breaks through the limitations of traditional open-loop or single feedback control in terms of system architecture.

[0023] Specifically, sensing unit 8 includes: Pressure sensor 81 is installed on clamping mechanism 3 to detect the clamping force on the plate during shearing in real time; And / or temperature sensor 82, located near the blade of the shearing mechanism 4, for real-time detection of the blade's operating temperature.

[0024] The clamping mechanism 3 includes a hydraulic cylinder 31, which is fixedly connected to the frame 1 via a bracket. A lifting plate 32 is fixedly connected to the output end of the hydraulic cylinder 31. The lower end of the lifting plate 32 is fixedly connected to the pressure plate 33 via a spring rod 34. A pressure sensor 81 is set at the connection position between the spring rod 34 and the lifting plate 32 to measure the clamping force on the plate during shearing.

[0025] The clamping mechanism 3 specifically includes a hydraulic cylinder 31, a lifting plate 32, a spring rod 34, and a pressure plate 33. The hydraulic cylinder 31 is fixed to the frame 1 by a bracket, and the lower end of its piston rod is fixedly connected to the lifting plate 32. The lower part of the lifting plate 32 is connected to the pressure plate 33 by the spring rod 34. The pressure sensor 81 is a weighing sensor or a strain gauge force sensor, which is set at the connection between the spring rod 34 and the lifting plate 32 (e.g., installed in a sensor mounting base on the lower surface of the lifting plate 32, with the upper end of the spring rod 34 pressing against the force measuring point of the pressure sensor 81). The temperature sensor 82 is preferably an infrared temperature sensor or a thermocouple, installed on the side of the upper or lower blade holder of the shearing mechanism 4, with its probe pointing towards the working area of ​​the blade edge.

[0026] When the central controller 7 issues a clamping command, the hydraulic cylinder 31 pushes the lifting plate 32 downward, pressing the pressure plate 33 against the sheet metal via the spring rod 34. The spring rod 34 provides flexible cushioning to prevent rigid impact damage to the sheet metal surface. The pressure sensor 81 detects the clamping force transmitted through the spring rod 34 in real time; this signal reflects the actual clamping state of the sheet metal. The temperature sensor 82 monitors the blade's operating temperature in real time, as increased blade temperature reduces edge hardness, alters shear friction, and consequently affects springback. This design ensures that the clamping force and blade temperature signals acquired by the system are direct and effective, providing reliable input variables for accurate modeling.

[0027] Specifically, the central controller 7 is configured to have: The springback calculation module is used to calculate the actual springback based on the actual length of the plate measured by the back-end re-inspection vision unit 52 and the corresponding actual positioning length. The prediction model update module is used to update the rebound amount prediction model based on the actual rebound amount and the corresponding sensor unit 8 data. The feeding compensation module is used to predict the springback amount of the next shear based on the updated model, and adjust the feeding target value of the drive mechanism 6 accordingly.

[0028] Springback Calculation Module: After each shearing re-inspection, this module receives the actual length L_a[n] of the plate from the back-end re-inspection vision unit 52 and the actual positioning length L_p[n] from the servo system. It immediately calculates the actual springback of this shearing: ΔL_a[n] = L_a[n] – L_p[n]. This value is the direct basis for compensation.

[0029] Prediction Model Update Module: This module uses the calculated ΔL_a[n], along with parameters such as the clamping force F[n] and blade temperature T[n] collected by sensor unit 8 during the shearing process, and the sheet metal specifications (such as thickness H and width W, which can be obtained from the identification module 9 or manually preset) as a training sample. The system maintains an initial model (such as the linear model Y=K1). F+K2 T+K3 H+B) This module uses new samples and algorithms such as recursive least squares to update the model coefficients (K1, K2, K3, B) online, making the model increasingly consistent with the current production reality.

[0030] Feeding compensation module: Before the next (n+1th) shearing, this module calls the updated model, inputs the currently monitored clamping force F[n+1] (estimated), blade temperature T[n+1], and sheet material specifications into the model, and predicts the possible springback amount ΔL_c[n+1] for this shearing. Subsequently, it adjusts the feeding target instruction to: L_t–ΔL_c[n+1] and sends it to the drive mechanism 6 for execution.

[0031] These three modules form a complete intelligent closed loop of "measurement-learning-prediction-feedforward compensation". The system no longer simply performs hysteresis correction based on the previous error, but can proactively predict the springback amount of the next shear under specific working conditions and perform advance compensation. This model-based feedforward control greatly improves the system's response speed and compensation accuracy, achieving adaptive learning and optimization.

[0032] The module is a software program functional unit executed in the central controller 7.

[0033] Specifically, the central controller 7 is also connected to the clamping mechanism 3 and is configured to dynamically adjust the output force of the clamping mechanism 3 based on feedback from the pressure sensor 81, ensuring that the clamping force remains stable within a set range during shearing. When controlling the clamping mechanism 3, the central controller 7 not only issues clamping commands but also reads the pressure value from the pressure sensor 81 in real time. For example, the target clamping force is set to F_s. During the clamping process, the controller uses a closed-loop control algorithm to dynamically adjust the hydraulic cylinder 31, stabilizing the measured pressure within a small range near F_s. This ensures that the clamping state of the sheet metal is consistent each time it is sheared, eliminating random errors introduced by fluctuations in the clamping force.

[0034] Specifically, the working timing of the front-end visual positioning unit 51 and the back-end re-inspection visual unit 52 is uniformly scheduled by the central controller 7, and the image acquisition trigger signals of both are synchronized with the encoder position signal of the drive mechanism 6 or the action phase signal of the shearing mechanism 4. To avoid motion blur and ensure consistent acquisition positions, the central controller 7 precisely synchronizes the acquisition triggers of the front-end visual positioning unit 51 and the back-end re-inspection visual unit 52. For example, the signal that triggers the acquisition of the front-end visual positioning unit 51 comes from the "Z-phase zero point" signal of the servo motor encoder of the drive mechanism 6 superimposed with a position offset, ensuring that the image is taken each time the plate is transported to the same relative position. The signal that triggers the acquisition of the back-end re-inspection visual unit 52 comes from the lifting signal of the pressure plate 33 of the hydraulic cylinder displacement sensor of the shearing mechanism 4. This ensures high repeatability and comparability of the detection data.

[0035] Specifically, it also includes an identification module 9 connected to the central controller 7, used to identify batch or material codes on the sheet material; the central controller 7 stores and manages the springback prediction model parameters and historical data under different identification codes in separate partitions. An identification module 9 (such as a QR code scanner or RFID reader) is added at the feeding end to read the batch codes on the sheet material. The central controller 7 establishes an internal database to independently store and manage the parameters and historical data of a set of springback prediction models for sheet materials with different codes. When switching between different batches of materials, the system automatically calls the corresponding model parameters, achieving one-click model switching without the need for relearning and accumulation, significantly improving production flexibility.

[0036] Specifically, the springback prediction model is a multiple linear regression model or a neural network model; the input variables of the model include at least: the clamping force F and / or blade temperature T collected by the sensing unit 8, and the specification parameters of the sheet material; the output variable of the model is the predicted springback amount ΔL_ct.

[0037] Multiple linear regression model: suitable for scenarios with relatively clear parameter relationships and initial data volume. Its form can be: ΔL_c = aF + bT + c H+d. The input variables are clamping force F, blade temperature T, and plate thickness H, and the output is the predicted springback amount ΔL_c. Its advantages are simple model, fast calculation, and easy interpretation.

[0038] Neural network models are suitable for complex scenarios with strong nonlinear relationships and abundant data. A fully connected feedforward network can be used, consisting of an input layer (nodes corresponding to F, T, H, etc.), several hidden layers, and an output layer (nodes representing ΔL_c). Neural networks can better fit the complex nonlinear mapping relationship between clamping force, blade temperature, material properties, and springback, exhibiting superior predictive potential.

[0039] Specifically, when the rebound prediction model is a neural network model, the central controller 7 is configured to use transfer learning in the initial stage. Based on a pre-trained base network model, it is rapidly fine-tuned using initial sampling data from the current production line to accelerate model convergence. When a neural network model is selected and initially applied to a new production line or with new materials, the system faces the problem of limited initial samples and slow model training. In this case, the system employs transfer learning technology. Specifically, the central controller 7 pre-stores a base network model trained on a large amount of historical data (this model has learned the general rules between rebound and various parameters). In the new scenario, the system retains the basic feature extraction layer of this network, only replacing and retraining the final fully connected output layer. The network is then fine-tuned using a small amount of initial sampling data from the new production line. This allows the new model to quickly adapt to the specificities of the new scenario, significantly reducing the amount of data and time required for model convergence, and enabling rapid deployment and effectiveness of the intelligent system.

[0040] A shearing process for a shearing device used in the production of automotive leaf springs includes the following steps: S1: Positioning step, the initial end position of the board is identified by the front-end vision positioning unit 51; S2: Feeding step, the central controller 7 controls the drive mechanism 6 to drive the feeding mechanism 2 to send the sheet to the shearing station according to the target length L_t and the predicted springback compensation amount. S3: Shearing step, control the clamping mechanism 3 to clamp the plate, and then control the shearing mechanism 4 to perform shearing; S4: Measurement step, the actual length L_a of the blanked plate is measured by the back-end re-inspection vision unit 52; S5: Update step, the central controller 7 calculates the actual rebound amount and updates the rebound amount prediction model based on the actual length L_a, the actual positioning length L_p and the current process parameters collected by the sensing unit 8. S6: Prediction step, based on the updated model and real-time process parameters, predicts the springback amount of the next shearing, which is used to compensate for the next feeding step S2.

[0041] Work process: After the sheet material enters the device, the drive mechanism 6 drives the feeding mechanism 2 to transport it to the shearing station. The front-end vision positioning unit 51 first performs imaging positioning on the initial end of the sheet material. The central controller 7 calculates the feeding command based on the set target length and initial compensation value and sends it to the drive mechanism 6 to execute the feeding.

[0042] During the feeding process and in preparation for shearing, the sensing unit 8 starts to work. Specifically, the pressure sensor 81 installed on the clamping mechanism 3 monitors the clamping force on the sheet material in real time; at the same time, the temperature sensor 82 installed near the blade of the shearing mechanism 4 detects the working temperature of the blade in real time.

[0043] After the material is fed into place, the central controller 7 controls the pressing mechanism 3 to operate. Its hydraulic cylinder 31 pushes the lifting plate 32, which in turn presses down the pressure plate 33 via the spring rod 34 to securely fix the material. Subsequently, the shearing mechanism 4 starts to complete the shearing.

[0044] After the shearing action is completed, the back-end re-inspection vision unit 52 immediately images the dropped sheet, measures its actual length, and feeds this data back to the central controller 7. The springback calculation module inside the central controller 7 then calculates the actual springback amount generated by this shearing based on the actual length and the actual length at the time of feeding and positioning. The prediction model update module then uses this springback amount data, along with process parameters such as clamping force and blade temperature collected by the pressure sensor 81 and temperature sensor 82 during this shearing process, as well as the sheet specification information obtained through the identification module 9, as a training sample to update its internal self-learning springback prediction model.

[0045] Based on the latest updated model, the feeding compensation module can predict the amount of springback that will occur during the next shearing by combining real-time process parameters. The central controller 7 then dynamically adjusts the target length of the next feed for the drive mechanism 6 based on this prediction. Throughout the process, the image acquisition of the front-end vision positioning unit 51 and the back-end re-inspection vision unit 52 is uniformly scheduled by the central controller 7 and triggered synchronously with the encoder signal of the drive mechanism 6 or the action phase of the shearing mechanism 4 to ensure the consistency and accuracy of the detection.

[0046] Through the above working process, the system forms an intelligent closed loop of "positioning-shearing-re-inspection-learning-prediction-compensation", which can actively adapt to changes in material properties and working conditions, dynamically compensate for shear springback, and thus achieve stable fixed-length shearing of the leaf spring.

[0047] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A shearing device for producing automotive leaf springs, comprising a frame (1), a feeding mechanism (2), a clamping mechanism (3), a shearing mechanism (4), and a driving mechanism (6) for feeding the feeding mechanism (2), characterized in that, Also includes: A front-end visual positioning unit (51) is disposed at the feeding end of the shearing mechanism (4) and is used to identify the initial end position of the plate. The back-end re-inspection vision unit (52) is set on the discharge side of the shearing mechanism (4) and is used to measure the actual length of the dropped plate after shearing. The sensing unit (8) is used to acquire at least one physical parameter related to the shearing process in real time; The central controller (7) is connected to the front-end visual positioning unit (51), the back-end re-inspection visual unit (52), the sensing unit (8), and the drive mechanism (6) respectively. The central controller (7) is configured to dynamically adjust the feeding length of the drive mechanism (6) based on the feedback signals from the front-end visual positioning unit (51) and the back-end re-inspection visual unit (52), as well as the real-time data from the sensing unit (8), in order to compensate for the fixed length error caused by shearing springback.

2. The shearing device for producing automotive leaf springs according to claim 1, characterized in that, The sensing unit (8) includes: A pressure sensor (81) is installed on the clamping mechanism (3) to detect the clamping force on the plate during shearing in real time; And / or a temperature sensor (82) is disposed near the blade of the shearing mechanism (4) for real-time detection of the blade's operating temperature.

3. The shearing device for producing automotive leaf springs according to claim 1, characterized in that, The clamping mechanism (3) includes a hydraulic cylinder (31), which is fixedly connected to the frame (1) by a bracket. The output end of the hydraulic cylinder (31) is fixedly connected to a lifting plate (32). The lower end of the lifting plate (32) is fixedly connected to a pressure plate (33) by a spring rod (34). The pressure sensor (81) is set at the connection position between the spring rod (34) and the lifting plate (32) to measure the clamping force on the plate during shearing.

4. The shearing device for producing automotive leaf springs according to claim 1 or 2, characterized in that, The central controller (7) is configured to have: The springback calculation module is used to calculate the actual springback based on the actual length of the plate measured by the back-end re-inspection vision unit (52) and the corresponding actual positioning length. The prediction model update module is used to update the rebound amount prediction model based on the actual rebound amount and the corresponding sensor unit (8) data. The feeding compensation module is used to predict the springback amount of the next shear based on the updated model and adjust the feeding target value of the drive mechanism (6) accordingly.

5. The shearing device for producing automotive leaf springs according to claim 2 or 3, characterized in that, The central controller (7) is also connected to the clamping mechanism (3) and is configured to dynamically adjust the output force of the clamping mechanism (3) based on the feedback from the pressure sensor (81) so that the clamping force during shearing is stabilized within a set range.

6. The shearing device for producing automotive leaf springs according to claim 1, characterized in that, The working sequence of the front-end visual positioning unit (51) and the back-end re-inspection visual unit (52) is uniformly scheduled by the central controller (7), and the image acquisition trigger signals of both are synchronized with the encoder position signal of the drive mechanism (6) or the action phase signal of the shearing mechanism (4).

7. The shearing device for producing automotive leaf springs according to claim 4, characterized in that, The springback prediction model is a multiple linear regression model or a neural network model; the input variables of the model include at least: the clamping force F and / or blade temperature T collected by the sensing unit (8), and the specification parameters of the sheet material; the output variable of the model is the predicted springback amount ΔL_ct.

8. The shearing device for producing automotive leaf springs according to claim 1, characterized in that, It also includes an identification module (9) connected to the central controller (7) for identifying batch or material codes on the sheet material; the central controller (7) stores and manages the rebound prediction model parameters and historical data under different identifications in partitions.

9. The shearing device for producing automotive leaf springs according to claim 7, characterized in that, When the rebound prediction model is a neural network model, the central controller (7) is configured to use transfer learning in the initial stage, based on a pre-trained base network model, and to quickly fine-tune it in combination with the initial sampling data of the current production line to accelerate model convergence.

10. A shearing process for a shearing device used in the production of automotive leaf springs as described in any one of claims 1-9, characterized in that, Includes the following steps: S1: Positioning step, the initial end position of the board is identified by the front-end visual positioning unit (51); S2: Feeding step, the central controller (7) controls the drive mechanism (6) to drive the feeding mechanism (2) to send the plate to the shearing station according to the target length L_t and the predicted springback compensation amount; S3: Shearing step, control the clamping mechanism (3) to clamp the plate, and then control the shearing mechanism (4) to perform shearing; S4: Measurement step, measure the actual length L_a of the blanked plate through the back-end re-inspection vision unit (52); S5: Update step, the central controller (7) calculates the actual rebound amount and updates the rebound amount prediction model based on the actual length L_a, the actual positioning length L_p and the current process parameters collected by the sensing unit (8). S6: Prediction step, based on the updated model and real-time process parameters, predicts the springback amount of the next shearing, which is used to compensate for the next feeding step S2.

Images