Robot-based cold extrusion process automation control method and system

Abstract

The application discloses a robot-based cold extrusion forming process automatic control method and system, belongs to the technical field of non-electric variable adjustment and control, and comprises the following steps: S1, initializing and configuring a cold extrusion device; S2, obtaining a quality prediction model; S3, outputting a real-time defect risk value, acquiring a cylinder pressure value in real time, and analyzing to obtain a filling state label, and obtaining construction execution instructions based on the real-time defect risk value and in combination with the filling state label; S4, obtaining an execution cylinder adjustment determination label, if the execution cylinder adjustment determination label is adjustment up to standard, then the adjustment is completed, if the execution cylinder adjustment determination label is speed reduction substandard, then speed reduction secondary correction processing is performed, and if the execution cylinder adjustment determination label is speed increase substandard, then speed increase secondary correction processing is performed; and S5, performing initial configuration updating of the quality prediction model, so that the technical problem of insufficient cold extrusion process production precision in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of non-electric variable regulation and control technology, and in particular to a robot-based automated control method and system for cold extrusion molding processes. Background Technology

[0002] Existing automated control systems for cold extrusion molding processes achieve the desired shape by placing the billet into the mold cavity and forcing it to undergo plastic flow at room temperature or lower under external force.

[0003] For example, the Chinese invention patent with publication number CN120540442A discloses a control method and system for PE insulation pipe production equipment, which includes: adjusting heating temperature control parameters based on the particle size distribution characteristics of raw materials, identifying viscosity state by combining screw torque and speed, adjusting the feed rate of each segment according to the viscosity difference, calculating molding uniformity and adjusting foaming agent injection parameters by combining cell structure, and adjusting cooling zone flow rate parameters by combining heat load assessment.

[0004] The above-mentioned technology has at least the following technical problems:

[0005] Existing cold extrusion processes often overemphasize die design and external loading conditions, neglecting the dynamic impact of equipment operation during process execution. Especially under high-frequency production or high-pressure extrusion conditions, the die-works contact area experiences significant localized temperature rise due to intense friction and plastic deformation. This localized temperature rise alters the material's flowability and frictional properties, consequently affecting pressure transmission and speed control. Because current technology fails to effectively respond to these changes, deviations in the actuator's operating parameters occur. Specifically, the actual cylinder angular velocity cannot consistently reach the target angular velocity, thus reducing the precision and consistency of part production. Therefore, the technical problem of insufficient production precision in cold extrusion processes stems from the localized temperature rise effect in the die-works contact area caused by high-frequency processing, leading to deviations in equipment operating parameters. Summary of the Invention

[0006] To address the technical problem in existing technologies where high-frequency processing causes localized heating in the mold-workpiece contact area, leading to deviations in equipment operating parameters and consequently insufficient production precision in cold extrusion processes, this invention provides a robot-based automated control method and system for cold extrusion molding processes. The technical solution is as follows:

[0007] On the one hand, a robot-based automated control method for cold extrusion molding is provided. This method includes: S1, after receiving a part processing instruction, the robot transports the blank to a designated position and simultaneously initializes the cold extrusion equipment; S2, the cold extrusion equipment is started, real-time construction execution parameters are acquired, and analyzed with historical construction execution parameter sets to obtain a historical construction training set, which is then used to train a quality prediction model; S3, real-time construction execution parameters are acquired again and input into the quality prediction model, outputting real-time defect risk values, real-time cylinder pressure values ​​are acquired, and the filling status is analyzed to obtain the filling status indicator. S4. Based on the real-time defect risk value and combined with the filling status label analysis, the construction execution command is obtained; S5. Based on the construction execution command, the operating status of the execution cylinder is adjusted, and the operating status of the execution cylinder is obtained after adjustment. The execution cylinder adjustment judgment label is obtained. If the execution cylinder adjustment judgment label is that the adjustment meets the standard, the adjustment is completed. If the execution cylinder adjustment judgment label is that the speed reduction does not meet the standard, the speed reduction secondary correction process is performed. If the execution cylinder adjustment judgment label is that the speed increase does not meet the standard, the speed increase secondary correction process is performed; S6. The quality parameters of the finished part after the part processing is completed are obtained, and the initial configuration update of the quality prediction model is performed accordingly.

[0008] On the other hand, a robot-based automated control system for cold extrusion molding is provided. This system includes: an initial configuration module, a model building module, a construction instruction acquisition module, a secondary calibration module, and a model configuration update module. The initial configuration module, upon receiving a part processing instruction from the robot, moves the blank to a designated location and simultaneously initializes the cold extrusion equipment. The model building module starts the cold extrusion equipment, acquires real-time construction execution parameters, analyzes them with historical construction execution parameter sets to obtain a historical construction training set, and uses this set to train a quality prediction model. The construction instruction acquisition module again acquires real-time construction execution parameters and inputs them into the quality prediction model, outputting real-time defect risk. The system employs several mechanisms: a risk value module, a secondary correction module, and a model configuration update module. The first module acquires the real-time pressure value of the actuator cylinder and analyzes it to obtain a filling status label. The second module generates a secondary correction module to adjust the actuator cylinder's operating status based on the construction execution command. After adjustment, it acquires the actuator cylinder's operating status and obtains an actuator cylinder adjustment judgment label. If the label indicates the adjustment meets the standard, the adjustment is complete. If the label indicates the speed reduction does not meet the standard, a second speed reduction correction is performed. If the label indicates the speed increase does not meet the standard, a second speed increase correction is performed. The third module acquires the quality parameters of the finished part after processing and updates the initial configuration of the quality prediction model accordingly.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0010] 1. The robot-based automated control method for cold extrusion molding provided by this invention acquires real-time construction execution parameters and trains a quality prediction model by combining historical construction training sets. This enables real-time output of real-time defect risk values ​​and analysis of filling status labels, thereby achieving intelligent decision-making on construction execution instructions. This effectively solves the technical problem in the prior art where the local heating effect in the contact area between the mold and the billet caused by high-frequency processing leads to deviations in equipment operating parameters, resulting in insufficient production precision in the cold extrusion process.

[0011] 2. This invention adjusts the target pressure and target linear velocity of the actuator cylinder based on the construction execution instructions of the defect risk level and filling status label, thereby enabling the correction of the equipment execution parameters according to the real-time working conditions, and thus improving the machining accuracy and optimizing the machining efficiency of the parts.

[0012] 3. Through secondary correction processing of speed reduction and speed increase, the linear speed and pressure deviation of the actuator cylinder under different working conditions can be dynamically adjusted, thereby achieving stable operation of the equipment under different load changes and avoiding the decrease in processing accuracy or equipment abnormality caused by insufficient flow, flow overflow or temperature changes.

[0013] 4. This invention uses statistical analysis of construction component quality evaluation parameters and initial configuration updates of the quality prediction model to incrementally train the model, thereby improving subsequent training efficiency. This enables rapid adaptation to new working conditions and continuous optimization of quality prediction capabilities, effectively solving the problem of low training efficiency in existing technologies where each training session requires starting from scratch.

[0014] 5. This invention monitors the hydraulic pump output flow, hydraulic oil viscosity, hydraulic valve opening, lubricant dynamic viscosity, and billet contact surface temperature in real time, and analyzes the secondary adjustment value of the actuator cylinder. This allows for more precise optimization and adjustment of the actuator cylinder's linear speed and pressure, thereby ensuring the stability of the cold extrusion process while improving work efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0016] Figure 1 A macroscopic flowchart of the robot-based automated control method for cold extrusion molding provided in the embodiments of this application;

[0017] Figure 2 A flowchart illustrating the steps of a robot-based automated control method for cold extrusion molding provided in this application embodiment;

[0018] Figure 3 A schematic diagram of the structure of the robot-based automated control method for cold extrusion molding provided in the embodiments of this application;

[0019] Figure 4 This is a schematic diagram of the structure of an automated control system for a robot-based cold extrusion molding process provided in an embodiment of this application. Detailed Implementation

[0020] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0021] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0022] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0023] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0024] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0025] like Figure 1The diagram shown is a flowchart of a robot-based automated control method for cold extrusion molding provided in this application embodiment. The method includes the following steps: S1. After receiving a part processing instruction, the robot transports the blank to a designated position and simultaneously initializes the cold extrusion equipment; S2. The cold extrusion equipment is started, real-time construction execution parameters are acquired, and analyzed against historical construction execution parameter sets to obtain a historical construction training set. This set is then used for training to obtain a quality prediction model; S3. Real-time construction execution parameters are acquired again and input into the quality prediction model, outputting real-time defect risk values. The real-time cylinder pressure value is also acquired and analyzed. S4. Based on the real-time defect risk value and the analysis of the filling status label, the construction execution command is obtained; S5. The operating status of the execution cylinder is adjusted according to the construction execution command, and the operating status of the execution cylinder is obtained after the adjustment. The execution cylinder adjustment judgment label is obtained. If the execution cylinder adjustment judgment label is that the adjustment meets the standard, the adjustment is completed. If the execution cylinder adjustment judgment label is that the speed reduction does not meet the standard, the speed reduction secondary correction process is performed. If the execution cylinder adjustment judgment label is that the speed increase does not meet the standard, the speed increase secondary correction process is performed; S6. The quality parameters of the finished part after the part processing is completed are obtained, and the initial configuration of the quality prediction model is updated accordingly.

[0026] In this embodiment, it should be noted that the database mentioned in this solution is an intelligent database that stores a large amount of configuration information and can be updated in real time. The database stores a number of data information, such as the adjustment amount of various parameters including the standardized distance threshold and the actuator displacement threshold (these thresholds are obtained by statistical analysis based on historical experimental data, specifically, the maximum standardized distance and the maximum actuator displacement corresponding to the products produced in the historical database that are qualified products are used as the standardized distance threshold and the actuator displacement threshold, respectively), the actuator pressure adjustment value (obtained by calculating the average value of the historical actuator pressure adjustment values), and the datasets corresponding to each historical sample vector, etc.

[0027] By introducing robotic handling and equipment initialization configuration into the cold extrusion molding process, the billet loading and process preparation stages are automated, avoiding human error. A quality prediction model is established by acquiring real-time construction execution parameters and comparing them with historical parameter sets, enabling the process to predict defect risks in advance during operation, thereby improving the controllability of part processing. During construction, construction execution instructions are dynamically generated by combining real-time defect risk values ​​and filling status labels, achieving adaptive adjustment of the actuator cylinder's operating status, thus ensuring the matching relationship between pressure and speed. Furthermore, in cases of actuator cylinder deviation, secondary correction processing can be used to correct the issue, ensuring the stability and accuracy of the execution actions. Finally, the quality prediction model is updated based on the quality parameters of the processed parts, allowing the model to continuously optimize as the working conditions evolve, thus achieving closed-loop control and accuracy assurance throughout the entire cold extrusion molding process.

[0028] Furthermore, the historical construction training set is obtained through the following methods: Real-time construction execution parameters are acquired and processed to obtain real-time construction feature vectors. These parameters include real-time cylinder pressure, real-time cylinder displacement, real-time cylinder linear velocity, and real-time slider velocity. A historical construction execution parameter set is acquired and processed to obtain historical sample vectors. This set includes historical cylinder pressure, historical cylinder displacement, historical cylinder linear velocity, and historical slider velocity for each historical sample. Each historical sample vector is then subjected to standardized distance analysis with the real-time construction feature vector to obtain the standardized distance between each historical sample vector and the real-time construction feature vector, and these standardized distances are labeled as standardized distances. A preset standardized distance threshold is obtained from the database and compared with each standardized distance. The dataset corresponding to each historical sample vector whose standardized distance is below the standardized distance threshold is selected as the historical construction training set.

[0029] In this embodiment, real-time construction execution parameters are obtained and processed to obtain a real-time construction feature vector. Specifically, the cylinder pressure, cylinder displacement, cylinder speed, and slider speed are collected in real time by sensors, and these data are normalized to form a unified feature vector to obtain the real-time construction feature vector.

[0030] Obtain the set of historical construction execution parameters and process them to obtain each historical sample vector. Specifically, extract historical construction execution parameter data from the database, convert each historical construction record into a corresponding feature vector, and ensure that the data dimension is consistent with the real-time feature vector.

[0031] Standardized distance analysis specifically involves calculating the standardized distance between each historical sample vector and the current real-time construction feature vector. This standardization measure the similarity between each historical sample and the current construction status. Standardization can eliminate the influence of differences in dimensions or values.

[0032] The historical construction training set is obtained by filtering. Specifically, a preset standardized distance threshold is obtained from the database, and the standardized distance of each historical sample is compared with the threshold. Only historical sample vectors with distances below the threshold are selected to form the historical construction training set, ensuring that the training data is highly correlated with the current working conditions and excluding irrelevant or significantly deviated historical data.

[0033] By performing standardized distance analysis on real-time construction characteristics and historical construction data, and selecting historical samples that are highly correlated with the current construction status as the training set, the defects of blindly using all historical data for training, which may introduce noise or irrelevant information, are avoided. This improves the accuracy of the quality prediction model and enhances the model's responsiveness and prediction accuracy under different construction conditions, ensuring that defect risk prediction during cold extrusion molding is more scientific and effective.

[0034] Furthermore, a quality prediction model is obtained. The specific method is as follows: the construction execution parameters of each historical sample in the historical construction training set are processed to obtain the input feature vector corresponding to each historical sample; the input feature vector corresponding to each historical sample is trained using the support vector machine algorithm to obtain the weight vector and bias term; the input feature vector is then linearly combined to obtain the linear combination result, which is used to characterize the relative position of the input features on the model decision plane; the linear combination result is input into the Sigmoid function for normalization processing, and the real-time defect risk value is output, thus obtaining the quality prediction model.

[0035] In this embodiment, the real-time defect risk value is obtained using the following method:

[0036] ;

[0037] ;

[0038] ;

[0039] In the formula, R represents the real-time defect risk value, x represents the input feature vector, w represents the weight vector, b represents the bias, T represents the transpose of the input feature vector, and P represents the weight vector. c Indicates the cylinder pressure, D c Indicates the displacement of the actuator cylinder, V c Indicates the linear velocity of the actuator, V s ω1 represents the slider speed, ω2 represents the actuator cylinder pressure weight, ω3 represents the actuator cylinder displacement weight, and ω4 represents the actuator cylinder linear velocity weight.

[0040] It should be noted that the real-time defect risk value ranges from (0,1), and the higher the value, the greater the risk of quality defects.

[0041] The optimal weight vector and optimal bias are found using the following formula:

[0042] ;

[0043] The constraints are: ;

[0044] , ;

[0045] ;

[0046] In the formula, min represents the minimum value, w represents the weight vector, and w * Let b represent the optimal weight vector, and b represent the bias. * Let y denote the optimal bias, C denote the misclassification penalty coefficient, and y i Let ξ represent the i-th historical defect tag, where i represents the tag number (i = 1, 2, ..., N), and N represents the total number of historical defect tags. i Let σ represent the allowed misclassification value for the i-th time. 2 This represents the variance of the training samples.

[0047] By introducing Lagrange multipliers to solve the dual problem, the optimal weights w and biases b are calculated:

[0048] Solving the dual problem using formulas: ;

[0049] ;

[0050] ;

[0051] constraint: ;

[0052] In the formula, ɑ i Let represent the contribution of the i-th historical defect label to the decision hyperplane, from which the weights and biases are calculated: ; ;

[0053] In the formula, ɑ i * Let W(α) represent the optimal Lagrange multiplier value obtained by maximizing the dual objective function W(α), where α is the optimal value of the Lagrange multiplier. i Variables in the optimization process, α i * It is the optimal solution after optimization, ɑ * Let x represent the set of optimal Lagrange multiplier values. k Let y be any support vector. k This represents the historical labels corresponding to the support vectors.

[0054] By analyzing the construction execution parameters, including cylinder pressure, cylinder displacement, cylinder linear velocity, and slider speed, a quality prediction model is constructed. This model takes into account the interrelationships between these parameters. For example, cylinder pressure directly acts on the billet, causing it to undergo plastic deformation, thus affecting the filling state and displacement. Cylinder displacement reflects the degree of billet flow in the mold; insufficient displacement may lead to incomplete local filling, and changes in displacement will also affect the stability of cylinder pressure. Cylinder linear velocity determines the pressure application time and deformation rate; excessive speed may cause excessive pressure peaks or billet rebound, while excessively low speed may lead to reduced production efficiency. Slider speed, in sync with cylinder linear velocity, affects the uniformity of billet flow and the completeness of filling; deviations from the set slider speed will cause pressure fluctuations and abnormal displacement.

[0055] It should be noted that the error variance of the training samples can be obtained by calculating the variance between the actual measured system output value and the model prediction value.

[0056] It should also be noted that real-time construction execution parameters, including real-time actuator pressure, real-time actuator displacement, real-time actuator linear velocity, and real-time slider speed, can be acquired collaboratively with the back-end control system via the equipment's built-in monitoring module. Specifically, actuator pressure is monitored in real-time by a high-precision pressure sensor, and the analog signal is converted into a digital signal by an analog-to-digital converter before being input to the control system. Actuator displacement is acquired by a linear displacement sensor (such as a magnetostrictive displacement sensor), capable of micron-level position resolution. Actuator linear velocity is calculated in real-time from displacement data using a differential calculation module. Slider speed is measured by a photoelectric encoder or laser velocimeter, calculated through synchronous sampling of the slider's movement path and time. All real-time data is uniformly connected to the equipment's main control system via a data acquisition card and communicates at high speed with the back-end management system through a bus protocol, enabling real-time display, dynamic recording, and automatic comparison of construction execution parameters.

[0057] By extracting features from the construction execution parameters of each historical sample in the historical construction training set, and training a weight vector and bias term based on the support vector machine algorithm, a quality prediction model reflecting the relative position of input features on the model's decision plane is constructed. By inputting the linear combination result into the sigmoid function to output a real-time defect risk value, potential defect risks during cold extrusion can be identified in advance. This allows for accurate analysis of real-time construction status, timely prediction of potential defects, and avoidance of part quality problems. Furthermore, by combining real-time defect risk information, key parameters such as cylinder pressure and speed can be dynamically adjusted to achieve adaptive correction of the equipment. This effectively improves the production efficiency and automation level of the cold extrusion process while ensuring part production safety and quality.

[0058] Furthermore, the filling status label is obtained as follows: Based on the real-time execution cylinder pressure value and the change degree analysis of the execution cylinder pressure value at the previous moment, the pressure change degree value is obtained; the execution cylinder displacement is obtained and compared with the preset execution cylinder displacement threshold in the database to obtain the displacement reaching judgment label. If the execution cylinder displacement is above the execution cylinder displacement threshold, the displacement reaching judgment label is considered as displacement meeting the standard; otherwise, the displacement reaching judgment label is considered as displacement not meeting the standard; the preset pressure change degree threshold in the database is obtained and compared with the pressure change degree value. If the pressure change degree value is above the pressure change degree threshold and the displacement reaching judgment label is considered as displacement meeting the standard, the filling status label is "fully filled" and the cold extrusion process ends; otherwise, the filling status label is "not fully filled" and the cold extrusion process continues.

[0059] In this embodiment, by analyzing the real-time cylinder pressure value and its degree of change, and combining the cylinder displacement with a preset threshold, a filling status label is generated to determine the filling status of the part during the cold extrusion molding process. This allows for real-time identification of whether the part is fully filled, avoiding cracks, folds, or material shortages caused by continued extrusion filling after actual filling is completed, thereby improving the quality and reliability of the molded parts. At the same time, by timely determining the filling status, the duration and equipment parameters of the cold extrusion process can be reasonably controlled, avoiding excessive pressure that could damage the mold or blank, thus improving production efficiency while ensuring product quality.

[0060] Furthermore, the construction execution instructions are obtained as follows: A preset defect risk threshold is retrieved from the database and compared with the real-time defect risk value. If the real-time defect risk value is less than the defect risk threshold, the defect risk level is low; otherwise, the defect risk level is high. Based on the defect risk level and the filling status label, the construction execution instructions are obtained. These instructions include adjusting the target pressure and target linear velocity of the execution cylinder. If the defect risk level is high and the filling status label is "not fully filled," the construction execution instructions are to increase the target pressure and decrease the target linear velocity of the execution cylinder. If the defect risk level is high and the filling status label is "fully filled," the construction execution instructions are to decrease the target pressure and decrease the target linear velocity of the execution cylinder. If the defect risk level is low and the filling status label is "not fully filled," the construction execution instructions are to increase the target pressure and increase the target linear velocity of the execution cylinder. If the defect risk level is low and the filling status label is "fully filled," the construction execution instructions are to maintain the current target pressure and increase the target linear velocity of the execution cylinder.

[0061] In this embodiment, by analyzing real-time defect risk values ​​and comparing them with preset defect risk thresholds, a defect risk level is generated, thereby determining the rationality of the equipment execution parameters during the current cold extrusion molding process. When the defect risk level is high, it indicates that the equipment execution parameters are set unreasonably, such as excessive cylinder pressure or excessive linear speed, which may lead to part damage or mold wear. When the defect risk level is low, it indicates that the current equipment state is relatively safe. At this time, the execution parameters may be too low, and production efficiency can be improved without increasing the defect risk by appropriately increasing the pressure or linear speed.

[0062] By combining the analysis of filling status labels, further construction execution instructions are generated, including adjustments to the target pressure and target linear speed of the execution cylinder. These directly affect the filling degree and forming quality of the billet. Adjusting the pressure controls material flow and filling effect, while adjusting the linear speed controls the material deformation rate during cold extrusion, thus balancing part integrity and production efficiency. In high-risk situations where filling is substandard, increasing pressure and decreasing linear speed improves the filling effect while reducing the risk of damage. In low-risk situations where filling is substandard, increasing pressure and linear speed accelerates the forming process and improves the filling effect. In other cases, pressure and speed are maintained or adjusted according to the filling status and risk level to ensure stable part quality while optimizing production efficiency.

[0063] Furthermore, the target pressure and target linear velocity of the actuator are adjusted. Specifically, the following methods are used: The current operating status of the actuator is obtained, including its linear velocity and pressure. A difference analysis is performed between the actuator and a defect risk threshold and the real-time defect risk value to obtain a defect risk difference value. This difference value is then matched with a database to obtain the actuator linear velocity adjustment value. Finally, the actuator linear velocity is adjusted based on the defect risk threshold and compared with the real-time defect risk value. If the real-time defect risk value is less than the threshold, the actuator linear velocity is increased based on the adjustment value; otherwise, the adjustment is based on the actuator pressure adjustment value. The linear velocity of the actuator is reduced to obtain the target linear velocity of the actuator, and the target linear velocity adjustment of the actuator is completed. A difference analysis is performed between the actuator pressure value and the actuator pressure threshold to obtain the actuator pressure difference value. This value is then matched with a database to obtain the actuator pressure adjustment value. The actuator pressure value is compared with the actuator pressure threshold. If the defective pressure value is less than the defective pressure threshold, the actuator pressure is increased based on the actuator pressure adjustment value; otherwise, the actuator pressure is decreased based on the actuator pressure adjustment value. This process yields the actuator target pressure and completes the actuator target pressure adjustment.

[0064] In this embodiment, the linear velocity of the actuator cylinder can be detected by an encoder. The pressure value of the actuator cylinder can be detected by a pressure sensor.

[0065] Based on the defect risk threshold and the difference between the real-time defect risk value, the defect risk difference value is obtained. Specifically, the absolute difference is obtained by subtracting the defect risk threshold from the real-time defect risk value and then dividing the absolute difference by the defect risk threshold.

[0066] The execution cylinder line speed adjustment value is obtained by matching the defect risk difference value with the database. Specifically, after obtaining the current defect risk difference value, it is compared with the historical defect risk difference values ​​stored in the database, and the absolute value of the difference between each historical data point and the current defect risk difference value is calculated. Then, a weight is calculated based on the absolute value of the difference for each historical data point. Specifically, the weight is equal to the reciprocal of the absolute value of the difference, and then normalized so that the sum of all weights is 1. Finally, the historical execution cylinder line speed adjustment value corresponding to each historical defect risk difference value is multiplied by its normalized weight, and the weighted results are summed to obtain the final execution cylinder line speed adjustment value.

[0067] The difference between the execution pressure value and the execution cylinder pressure threshold is analyzed to obtain the execution cylinder pressure difference value. Specifically, the absolute value of the absolute difference is obtained by subtracting the execution cylinder pressure threshold from the execution pressure value and then dividing the absolute difference by the execution cylinder pressure threshold.

[0068] The actuator cylinder pressure adjustment value is obtained by matching the actuator cylinder pressure difference value with the database. Specifically, after obtaining the actuator cylinder pressure difference value, it is compared with the historical actuator cylinder pressure difference value data stored in the database, and the absolute value of the difference between each historical data point and the actuator cylinder pressure difference value is calculated. Then, a weight is calculated based on the absolute value of the difference for each historical data point. Specifically, the weight is equal to the reciprocal of the absolute value of the difference, and then normalized so that the sum of all weights is 1. Finally, the historical actuator cylinder pressure adjustment value corresponding to each historical actuator cylinder pressure difference value is multiplied by its normalized weight, and the weighted results are summed to obtain the final actuator cylinder pressure adjustment value.

[0069] By monitoring the operating status of the actuator cylinder in real time and dynamically analyzing the difference between the real-time defect risk value and the preset threshold, the target pressure and target linear velocity of the actuator cylinder can be finely adjusted. By matching the preset adjustment parameters in the database, corresponding pressure and speed adjustments can be generated for different levels of defect risk differences, thereby achieving closed-loop control of the actuator cylinder's movement. This improves the stability and consistency of the molding process and maintains the accuracy of equipment operating parameters during the processing of different parts or batches, avoiding local overpressure or insufficient speed problems caused by equipment parameter deviations, thus further ensuring production safety and part processing quality.

[0070] Further, a secondary speed reduction correction process is performed. Specifically, the following steps are taken: The flow rate of the hydraulic pump output to the actuator cylinder is obtained and marked as the actual flow rate; the preset minimum stable flow rate in the database is obtained and compared with the actual flow rate to obtain a speed reduction effect tracking label. If the actual flow rate is greater than the minimum stable flow rate, the speed reduction effect tracking label indicates flow overflow, and flow compensation is performed; if the actual flow rate is less than the minimum stable flow rate, the speed reduction effect tracking label indicates insufficient flow, and flow reduction is performed; if the actual flow rate equals the minimum stable flow rate, the speed reduction effect tracking label indicates equipment malfunction, and an equipment malfunction alert is issued; the flow compensation and flow reduction processes are jointly marked as working oil flow rate correction processing; based on the adjusted actuator cylinder linear velocity and the actuator cylinder target linear velocity... Speed ​​difference analysis is performed to obtain the speed difference value of the actuator cylinder, and this value is matched with the database to obtain the working oil flow constraint factor. The difference between the actual flow rate and the minimum stable flow rate is processed to obtain the stable flow rate difference, which is used as the working oil flow correction value. Based on the working oil flow correction processing analysis, if the working oil flow correction processing is a flow compensation processing, the working oil flow correction value is increased based on the working oil flow constraint factor to obtain the comprehensive working oil flow adjustment value; if the working oil flow correction processing is a flow reduction processing, the working oil flow correction value is decreased based on the working oil flow constraint factor to obtain the comprehensive working oil flow adjustment value. The working oil flow is then adjusted based on the comprehensive working oil flow adjustment value.

[0071] In this embodiment, the difference between the adjusted linear velocity of the actuator cylinder and the target linear velocity of the actuator cylinder is analyzed to obtain the difference value of the linear velocity of the actuator cylinder. Specifically, the absolute value of the difference between the adjusted linear velocity of the actuator cylinder and the target linear velocity of the actuator cylinder is taken to obtain the absolute difference between the linear velocity of the actuator cylinder and the target linear velocity. The absolute difference between the linear velocity of the actuator cylinder and the target linear velocity of the actuator cylinder is divided by the target linear velocity of the actuator cylinder to obtain the difference value of the linear velocity of the actuator cylinder.

[0072] The working oil flow constraint factor is obtained by matching the actuator linear velocity difference value with the database. Specifically, after obtaining the actuator linear velocity difference value, it is compared with the historical actuator linear velocity difference values ​​stored in the database. The absolute value of the difference between each historical data point and the actuator linear velocity difference value is calculated. Then, a weight is calculated based on the absolute value of the difference for each historical data point. Specifically, the weight is equal to the reciprocal of the absolute value of the difference, and then normalized so that the sum of all weights is 1. Finally, the historical working oil flow constraint factor corresponding to each historical actuator linear velocity difference value is multiplied by its normalized weight, and the weighted results are summed to obtain the final working oil flow constraint factor.

[0073] The working oil flow rate correction value is increased based on the working oil flow rate constraint factor to obtain the comprehensive working oil flow rate regulation value. Specifically, this is done by multiplying the working oil flow rate correction value by the working oil flow rate constraint factor and then adding the working oil flow rate correction value back to the original value. Alternatively, the working oil flow rate correction value is decreased based on the working oil flow rate constraint factor to obtain the comprehensive working oil flow rate regulation value. Specifically, this is done by multiplying the working oil flow rate correction value by the working oil flow rate constraint factor and then subtracting the working oil flow rate correction value back to the original value.

[0074] By monitoring the flow rate of the hydraulic pump output to the actuator in real time and comparing it with the preset minimum stable flow rate in the database, the speed reduction effect can be analyzed and traced to achieve precise secondary speed reduction correction. By analyzing the difference between the actual flow rate and the minimum stable flow rate, different speed reduction effect states can be identified: when the actual flow rate is less than the minimum stable flow rate, the label displays "Insufficient Flow," indicating that flow compensation is needed to ensure the actuator's linear speed is reduced to the target value; when the actual flow rate is greater than the minimum stable flow rate, the label displays "Flow Overflow," indicating that flow reduction is needed to prevent the actuator's speed from decreasing too quickly or becoming unstable; when the actual flow rate equals the minimum stable flow rate, the label displays "Equipment Abnormality," indicating that the speed reduction effect is not due to a problem with the hydraulic flow rate, but may be due to a fault in the hydraulic system or the actuator itself. In this case, an equipment abnormality warning will be issued to prevent safety risks.

[0075] Furthermore, a secondary speed-up correction process is performed. The specific method is as follows: Equipment operation influencing parameters are obtained, including hydraulic pump output flow, hydraulic oil viscosity, hydraulic valve opening, lubricant dynamic viscosity, and billet contact surface temperature. Based on the analysis of these parameters, a secondary adjustment value for the actuator cylinder linear speed is obtained. The current actuator cylinder linear speed is updated and compared with the target linear speed to obtain a second difference value. This second difference value is then matched with the database to obtain a secondary adjustment constraint factor for the actuator cylinder linear speed. The secondary adjustment value and constraint factor are coupled to obtain a second comprehensive adjustment value for the actuator cylinder linear speed. This is then used to perform a secondary speed-up process on the actuator cylinder linear speed (specifically, the current actuator cylinder linear speed is added to the second comprehensive adjustment value to obtain the actuator cylinder linear speed after the secondary speed-up process).

[0076] In this embodiment, the secondary adjustment value of the linear velocity of the actuator is obtained by the following method:

[0077] ;

[0078] ;

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] In the formula, ER represents the secondary adjustment of the cylinder linear velocity, and ω F P represents the weight of the actual outflow rate of the hydraulic pump. F This indicates the effect of adjusting the hydraulic pump outflow rate, where F represents the actual outflow rate of the hydraulic pump. ref ω represents the flow rate reference value. o P represents the dynamic viscosity weight of hydraulic oil. O This indicates the effect of hydraulic oil dynamic viscosity adjustment, where V represents the hydraulic oil dynamic viscosity. ref V represents the reference oil viscosity. crit ω represents the critical value of oil viscosity. v P represents the dynamic viscosity weight of hydraulic oil. V This indicates the factor affecting the hydraulic valve opening adjustment, where Θ represents the actual hydraulic valve opening. safe P represents the hydraulic valve opening threshold. L This represents the effect of lubricant dynamic viscosity adjustment, ω. L H represents the dynamic viscosity weight of the hydraulic oil, where H represents the dynamic viscosity of the lubricant.ref H represents the reference viscosity of the lubricant. crit P represents the critical viscosity of the lubricant. T T represents the effect of adjusting the temperature of the contact surface of the billet. c T represents the temperature of the contact surface of the blank. nom Indicates the calibrated contact temperature, T stop This indicates the temperature threshold of the contact surface of the blank.

[0084] The weights of hydraulic oil dynamic viscosity are determined by collecting historical sample data from the system under different operating conditions, followed by data cleaning and normalization to ensure that all parameters are within the same dimension. Subsequently, using multiple input parameters (such as cylinder pressure, displacement, and speed) as independent variables and system response results (such as execution accuracy) as dependent variables, a multiple linear relationship model is established using the least squares regression method. The regression coefficients of each input parameter are obtained by minimizing the sum of squared residuals. To avoid weight bias caused by strong correlations among input parameters, correlation analysis methods (such as Pearson correlation analysis) are used to calculate the linear correlation between any two input parameters. Parameters with excessively high correlations are corrected to reduce collinearity. Finally, the corrected regression coefficients of each parameter are normalized to a sum of 1, and this sum is used as the weight of each parameter.

[0085] The parameters affecting equipment operation, including hydraulic pump output flow, hydraulic oil viscosity, hydraulic valve opening, lubricant dynamic viscosity, and billet contact surface temperature, can be obtained through the equipment's back-end management system. This system integrates various internal sensing and monitoring modules: a hydraulic pump flow sensor for real-time monitoring of the hydraulic system's instantaneous output flow; a hydraulic oil temperature and pressure composite sensor for acquiring oil temperature and pressure parameters to calculate viscosity values ​​using an oil property database; a hydraulic valve displacement sensor for monitoring valve opening changes; a lubricant viscosity sensor for online monitoring via a miniature vibratory viscometer; and an embedded infrared temperature sensor for acquiring the billet contact surface temperature. All sensor data is aggregated by the data acquisition unit and transmitted via an industrial Ethernet interface to the back-end management system for unified storage and real-time analysis. The system uses timestamp synchronization to automatically extract and record each operational parameter, ensuring the continuity and accuracy of parameter acquisition.

[0086] Flow reference values, reference oil viscosity, critical oil viscosity values, hydraulic valve opening thresholds, lubricant reference viscosity, calibrated contact temperature, and blank contact surface temperature thresholds can be directly obtained from the database.

[0087] The output flow of the hydraulic pump, the viscosity of the hydraulic oil, the opening of the hydraulic valve, the dynamic viscosity of the lubricant, and the temperature of the billet contact surface can all be obtained by real-time acquisition through their respective corresponding sensor devices. Among them, the output flow of the hydraulic pump is measured by a flow sensor; the viscosity of the hydraulic oil is obtained by an online viscosity monitor; the opening of the hydraulic valve is measured by a spool displacement sensor; the dynamic viscosity of the lubricant is obtained by an online lubricating oil monitor sensor; the temperature of the billet contact surface is measured in real time by an infrared temperature sensor or a thermocouple. All the above sensors are connected to the background data management system, and the collected data is uploaded to the background management system in real time through the industrial Ethernet. The background management system reads, records, analyzes, and manages these parameters uniformly to achieve multi-parameter linkage monitoring.

[0088] By analyzing the equipment operation influencing parameters including the output flow of the hydraulic pump, the viscosity of the hydraulic oil, the opening of the hydraulic valve, the dynamic viscosity of the lubricant, and the temperature of the billet contact surface, the secondary adjustment value of the actuator cylinder linear velocity is obtained. This is considering the mutual influence relationship between these parameters. For example: The output flow of the hydraulic pump directly affects the speed and pressure of the actuator cylinder. When the flow is insufficient, the valve opening may need to be increased to compensate for the flow, but the changes in the viscosity of the hydraulic oil and the dynamic viscosity of the lubricant will affect the valve response and the flow transmission efficiency; at the same time, the increase in the temperature of the billet contact surface will cause the local oil temperature to rise, thereby reducing the oil viscosity, decreasing the fluid resistance, and causing fluctuations in the flow and pressure; conversely, if the temperature is too low, the oil viscosity increases, and the valve may need a larger opening to maintain the same flow, resulting in a decrease in the actuator cylinder linear velocity. A closed-loop influence is formed among the parameters: the flow affects the pressure and the linear velocity, the viscosities of the oil and the lubricant affect the flow response, the valve opening regulates the flow, and the temperature change in turn regulates the oil viscosity, thus jointly acting on the speed and pressure performance of the actuator cylinder.

[0089] Update and obtain the current actuator cylinder linear velocity (and mark it as the updated value of the actuator cylinder linear velocity), and analyze the degree of difference from the target linear velocity of the actuator cylinder to obtain the second degree of difference value of the actuator cylinder linear velocity. The specific method is: subtract the target linear velocity of the actuator cylinder from the updated value of the actuator cylinder linear velocity to obtain the updated difference of the target linear velocity of the actuator cylinder, and divide the updated difference of the target linear velocity of the actuator cylinder by the target linear velocity of the actuator cylinder to obtain the second degree of difference value of the actuator cylinder linear velocity.

[0090] The secondary adjustment constraint factor for the execution cylinder linear speed is obtained by matching the second difference value of the execution cylinder linear speed with the database. Specifically, after obtaining the current second difference value of the execution cylinder linear speed, it is compared with the historical second difference values ​​of the execution cylinder linear speed stored in the database. The absolute value of the difference between each historical data point and the current second difference value is calculated. Then, a weight is calculated based on the absolute value of the difference for each historical data point. Specifically, the weight is equal to the reciprocal of the absolute value of the difference, and then normalized so that the sum of all weights is 1. Next, the secondary adjustment constraint factor for the historical execution cylinder linear speed corresponding to each historical second difference value is multiplied by its normalized weight, and the weighted results are summed to obtain the final secondary adjustment constraint factor for the execution cylinder linear speed.

[0091] The second comprehensive adjustment value of the execution cylinder linear velocity is obtained by coupling the secondary adjustment value of the execution cylinder linear velocity with the secondary adjustment constraint factor. Specifically, the second comprehensive adjustment value of the execution cylinder linear velocity is obtained by adding the product of the secondary adjustment value of the execution cylinder linear velocity and the secondary adjustment constraint factor.

[0092] By real-time monitoring and analysis of parameters affecting equipment operation, a secondary correction process for the actuator linear speed is achieved, thus resolving the issue of the original speed increase failing to reach the reasonable target. During continuous high-intensity cold extrusion, the hydraulic pump output flow, hydraulic oil viscosity, hydraulic valve opening, lubricant dynamic viscosity, and billet contact surface temperature all change over time. These factors can cause the actuator linear speed to fall below the target value, impacting production efficiency and reducing processing capacity per unit time. By analyzing the parameters affecting equipment operation, a secondary adjustment value for the actuator linear speed can be calculated. This value is then coupled with a secondary adjustment constraint factor obtained from a database matching the difference with the target linear speed to generate a second comprehensive adjustment value for the actuator linear speed. This effectively offsets the effects of temperature changes and equipment operating status variations on the linear speed, ensuring the actuator linear speed is close to the optimal setting, improving the processing efficiency of the cold extrusion forming process, and reducing problems such as incomplete part filling or uneven processing caused by insufficient speed. Furthermore, through secondary correction, the equipment status can be continuously monitored, ensuring that the equipment operates within a safe range during long-term continuous operation, reducing mechanical wear and the risk of failure.

[0093] Furthermore, the initial configuration update of the quality prediction model is performed as follows: After the cold extrusion molding process reaches the preset duration, the construction execution parameters and construction part quality evaluation parameters during this period are statistically analyzed. The construction part quality evaluation parameters include the number of cracks and the area of ​​fold defects. The construction part quality evaluation parameters are correlated and matched with the corresponding real-time construction execution parameters and real-time defect risk values ​​to construct quality evaluation training samples, thereby triggering the initial configuration update of the quality prediction model. The initial configuration update is specifically performed by calling the training configuration parameters from the previous round and performing incremental training in combination with the quality evaluation training samples to obtain the incrementally trained quality prediction model and its corresponding training configuration parameters, which are then used as the updated training configuration parameters, thus completing the initial configuration update of the quality prediction model. The training configuration parameters include the learning rate, regularization coefficient, and optimizer momentum parameter.

[0094] In this embodiment, the statistical analysis of construction execution parameters and construction part quality evaluation parameters is as follows: After the cold extrusion molding process reaches the preset time, the background management system acquires construction execution parameters such as cylinder pressure, linear velocity, and hydraulic pump flow rate from various sensors during that time period. Simultaneously, it acquires quality evaluation parameters for the completed parts, such as the number of cracks and the area of ​​folding defects. The number of cracks can be determined by using an industrial vision inspection system (high-definition camera) to capture images of the molded part surface, and then using image processing algorithms (such as edge detection) to identify the location and number of cracks. The acquired image data is then uploaded to the background data management system in real time. The area of ​​folding defects can be determined by using an industrial vision inspection system to acquire images of the part surface, combining image segmentation algorithms to extract the folding defect area, calculating the defect pixel area, converting it into the actual physical area, and storing it synchronously in the background data management system.

[0095] The quality evaluation training samples are constructed by associating and matching the construction execution parameters of each part with the corresponding quality evaluation parameters and real-time defect risk values ​​to form a complete training sample record. By integrating the data from multiple parts, a training dataset containing multiple sample records is obtained, which is used for subsequent training to obtain the quality evaluation training samples.

[0096] Incremental training updates the model by calling the configuration parameters from the previous training round (such as learning rate, regularization coefficient, and optimizer momentum parameters) and performing incremental training on the existing model using new quality assessment training samples. During incremental training, only the model weights are updated for the new data. (During incremental training, the system inputs the newly constructed quality assessment training samples into the existing quality prediction model, uses the model's current weights to predict the new data, and calculates the error between the predicted value and the actual real-time defect risk value. Subsequently, the gradient of the loss function with respect to each weight is calculated through backpropagation, and then combined with the optimizer, such as stochastic gradient descent with momentum, the model weights are updated. Only the relevant weights are adjusted for the new data, while retaining the existing weight structure, ultimately resulting in the updated quality prediction model.) This avoids retraining from scratch, generating the updated model and its corresponding training configuration parameters. Updating model weights only for new data allows for rapid absorption of new data features, improving the model's ability to predict defect risks under current construction conditions, while avoiding retraining from scratch each time, improving training efficiency, and maintaining the stability of predictions based on historical data. Update the training configuration parameters, specifically by marking the incrementally trained training parameters as the updated training configuration parameters, completing the initial configuration update of the quality prediction model, and preparing for the next round of prediction and real-time correction.

[0097] By updating the initial configuration of the quality prediction model, the model can be trained from scratch without having to start from scratch, significantly improving training speed and efficiency. At the same time, by continuously incorporating the latest construction execution parameters and quality evaluation parameters into the model, the prediction accuracy and adaptability can be enhanced, ensuring that the subsequent cold extrusion molding process can more quickly and accurately reflect defect risks in real-time monitoring and cylinder parameter adjustment. This improves production efficiency while ensuring part processing quality and reduces the risk of misjudgment or adjustment delays caused by model lag.

[0098] like Figure 4The diagram shows the structure of a robot-based automated control system for cold extrusion molding provided in this embodiment of the application. The robot-based automated control system for cold extrusion molding includes: an initial configuration module, a model building module, a construction instruction acquisition module, a secondary correction module, and a model configuration update module. The initial configuration module is used to move the blank to a designated position and simultaneously initialize the cold extrusion equipment after the robot receives a part processing instruction. The model building module is used to start the cold extrusion equipment, acquire real-time construction execution parameters, and analyze them with historical construction execution parameter sets to obtain a historical construction training set, thereby training a quality prediction model. The construction instruction acquisition module is used to acquire real-time construction execution parameters again and... The data is input into the quality prediction model, which outputs real-time defect risk values, acquires real-time cylinder pressure values, and analyzes them to obtain filling status labels. Based on the real-time defect risk values ​​and the filling status labels, the construction execution instructions are obtained. The secondary correction module is used to adjust the operating status of the cylinder based on the construction execution instructions, and acquires the operating status of the cylinder after adjustment, obtaining the cylinder adjustment judgment label. If the cylinder adjustment judgment label indicates that the adjustment meets the standard, the adjustment is completed. If the cylinder adjustment judgment label indicates that the speed reduction does not meet the standard, a speed reduction secondary correction process is performed. If the cylinder adjustment judgment label indicates that the speed increase does not meet the standard, a speed increase secondary correction process is performed. The model configuration update module is used to acquire the quality parameters of the finished part after the part processing is completed, and thereby update the initial configuration of the quality prediction model.

[0099] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0104] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A robot-based automated control method for cold extrusion molding process, characterized in that, Includes the following steps: S1. After receiving the part processing instruction, the robot transports the blank to the designated position and simultaneously initializes the cold extrusion equipment. S2. Start the cold extrusion equipment, obtain real-time construction execution parameters, and analyze them with the historical construction execution parameter set to obtain the historical construction training set. Use this set to train the quality prediction model. S3. Obtain the real-time construction execution parameters again and input them into the quality prediction model. Output the real-time defect risk value, obtain the execution cylinder pressure value and execution cylinder displacement in real time, and analyze to obtain the filling status label. Based on the real-time defect risk value and the filling status label, analyze to obtain the construction execution command. S4. Adjust the operating status of the actuator cylinder based on the construction execution command, and obtain the operating status of the actuator cylinder after adjustment. Obtain the actuator cylinder adjustment judgment label. If the actuator cylinder adjustment judgment label indicates that the adjustment meets the standard, the adjustment is completed. If the actuator cylinder adjustment judgment label indicates that the speed reduction does not meet the standard, perform a second speed reduction correction process. If the actuator cylinder adjustment judgment label indicates that the speed increase does not meet the standard, perform a second speed increase correction process. S5. Obtain the quality parameters of the finished part after the part processing is completed, and then update the initial configuration of the quality prediction model.

2. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for obtaining the historical construction training set is as follows: Real-time construction execution parameters are obtained and processed to obtain a real-time construction feature vector. The real-time construction execution parameters include real-time execution cylinder pressure, real-time execution cylinder displacement, real-time execution cylinder linear velocity, and real-time slider velocity. Obtain the historical construction execution parameter set and process it to obtain each historical sample vector. The historical construction execution parameter set includes the historical execution cylinder pressure, historical execution cylinder displacement, historical execution cylinder linear velocity, and historical slider velocity of each historical sample. Each historical sample vector is subjected to standardized distance analysis with the real-time construction feature vector to obtain the standardized distance between each historical sample vector and the real-time construction feature vector, and these distances are marked as standardized distances. Obtain the preset standardized distance threshold in the database, compare it with each standardized distance, and select the dataset corresponding to each historical sample vector whose standardized distance is below the standardized distance threshold as the historical construction training set.

3. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for obtaining the quality prediction model is as follows: The construction execution parameters of each historical sample in the historical construction training set are processed to obtain the input feature vector corresponding to each historical sample. The input feature vectors corresponding to each historical sample are trained by the support vector machine algorithm to obtain weight vectors and bias terms. The input feature vectors are then linearly combined to obtain a linear combination result, which is used to characterize and reflect the relative position of the input features on the model decision plane. The linear combination result is input into the Sigmoid function for normalization, and the real-time defect risk value is output, thus obtaining the quality prediction model.

4. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for obtaining the filled status label is as follows: The pressure change value is obtained by analyzing the change between the real-time cylinder pressure value and the cylinder pressure value at the previous moment. The displacement of the actuator cylinder is obtained and compared with the preset displacement threshold of the actuator cylinder in the database to obtain the displacement reaching judgment label. If the displacement of the actuator cylinder is above the displacement threshold, the displacement reaching judgment label means that the displacement meets the standard; otherwise, the displacement reaching judgment label means that the displacement does not meet the standard. Obtain the preset pressure change threshold from the database and compare it with the pressure change value. If the pressure change value is above the pressure change threshold and the displacement reaches the judgment label as displacement meets the standard, the filling status label is "fully filled" and the cold extrusion process ends. Otherwise, the filling status label is "not fully filled" and the cold extrusion process continues.

5. The automated control method for robot-based cold extrusion molding process as described in claim 4, characterized in that: The specific method for obtaining the construction execution instruction is as follows: Obtain the preset defect risk threshold in the database and compare it with the real-time defect risk value. If the real-time defect risk value is less than the defect risk threshold, the defect risk level is low; otherwise, the defect risk level is high. Based on the analysis of defect risk level and filling status label, the construction execution instructions are obtained, which include adjusting the target pressure of the actuator cylinder and adjusting the target linear velocity of the actuator cylinder. If the defect risk level is high and the filling status label is incompletely filled, the construction execution instruction is to increase the target pressure of the actuator cylinder and decrease the target linear velocity of the actuator cylinder. If the defect risk level is high and the filling status label is fully filled, the construction execution instruction is to reduce the target pressure of the actuator cylinder and reduce the target linear velocity of the actuator cylinder. If the defect risk level is low and the filling status label is not fully filled, the construction execution instruction is to increase the target pressure of the actuator cylinder and increase the target linear velocity of the actuator cylinder. If the defect risk level is low and the filling status label is fully filled, the construction execution instruction is to maintain the current target pressure of the actuator cylinder and increase the target linear velocity of the actuator cylinder.

6. The automated control method for robot-based cold extrusion molding process as described in claim 5, characterized in that: The specific methods for adjusting the target pressure and target linear velocity of the actuator cylinder are as follows: Obtain the current operating status of the actuator, which includes the actuator linear velocity and actuator pressure value; Based on the defect risk threshold and the difference analysis between it and the real-time defect risk value, the defect risk difference value is obtained. Based on the defect risk difference value and the database, the cylinder line speed adjustment value is obtained. Based on the defect risk threshold, and compared with the real-time defect risk value, if the real-time defect risk value is less than the defect risk threshold, the linear velocity of the actuator cylinder is increased based on the actuator cylinder linear velocity adjustment value; otherwise, the linear velocity of the actuator cylinder is decreased based on the actuator cylinder pressure adjustment value. Thus, the target linear velocity of the actuator cylinder is obtained, and the target linear velocity adjustment of the actuator cylinder is completed. The degree of difference between the execution pressure value and the execution cylinder pressure threshold is analyzed to obtain the execution cylinder pressure difference value. The execution cylinder pressure adjustment value is obtained by matching the execution cylinder pressure difference value with the database. Based on the comparison between the execution pressure value and the execution cylinder pressure threshold, if the defect pressure value is less than the defect pressure threshold, the pressure of the execution cylinder is increased based on the execution cylinder pressure adjustment value; otherwise, the pressure of the execution cylinder is decreased based on the execution cylinder pressure adjustment value. This process yields the target pressure of the execution cylinder and completes the target pressure adjustment of the execution cylinder.

7. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for the deceleration secondary correction process is as follows: Obtain the flow rate of the working oil output from the hydraulic pump to the actuator cylinder, and mark it as the actual flow rate; The system retrieves the preset minimum stable flow rate from the database and compares it with the actual flow rate to obtain a speed reduction effect tracking label. If the actual flow rate is greater than the minimum stable flow rate, the speed reduction effect tracking label indicates flow overflow, and flow compensation is performed. If the actual flow rate is less than the minimum stable flow rate, the speed reduction effect tracking label indicates insufficient flow, and flow reduction is performed. If the actual flow rate is equal to the minimum stable flow rate, the speed reduction effect tracking label indicates device malfunction, and a device malfunction alert is issued. The flow compensation process and the flow reduction process are combined and labeled as the working fluid flow correction process; The difference between the adjusted linear velocity of the actuator cylinder and the target linear velocity of the actuator cylinder is analyzed to obtain the difference value of the linear velocity of the actuator cylinder, and then matched with the database to obtain the working oil flow constraint factor. The difference between the actual flow rate and the minimum stable flow rate is processed to obtain the stable flow rate difference, which is used as the working oil flow rate correction value. Based on the working oil flow rate correction processing analysis, if the working oil flow rate correction processing is a flow compensation processing, the working oil flow rate correction value is increased based on the working oil flow rate constraint factor to obtain the comprehensive adjustment value of the working oil flow rate. If the working oil flow correction process is a flow reduction process, then the working oil flow correction value is reduced based on the working oil flow constraint factor to obtain the working oil flow comprehensive adjustment value, and then the working oil flow is adjusted based on the working oil flow comprehensive adjustment value.

8. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for performing the acceleration secondary correction process is as follows: Obtain the parameters affecting equipment operation, including hydraulic pump output flow rate, hydraulic oil viscosity, hydraulic valve opening degree, lubricant dynamic viscosity, and billet contact surface temperature; Based on the analysis of parameters affecting equipment operation, the secondary adjustment value of the actuator linear velocity is obtained; Update the current linear velocity of the execution cylinder, analyze the difference between it and the target linear velocity of the execution cylinder, obtain the second difference value of the linear velocity of the execution cylinder, and match it with the database to obtain the secondary adjustment constraint factor of the linear velocity of the execution cylinder; The second comprehensive adjustment value of the actuator linear speed is obtained by coupling the second adjustment constraint factor of the actuator linear speed, and then the second speed-up processing of the actuator linear speed is performed.

9. The automated control method for robot-based cold extrusion molding process as described in claim 1, characterized in that: The specific method for updating the initial configuration of the quality prediction model is as follows: After the cold extrusion molding process reaches the preset time, the construction execution parameters and construction part quality evaluation parameters within that time are statistically analyzed. The construction part quality evaluation parameters include the number of cracks and the area of ​​folding defects. The quality evaluation parameters of construction parts are associated and matched with the corresponding real-time construction execution parameters and real-time defect risk values ​​to construct quality evaluation training samples, thereby triggering the initial configuration update of the quality prediction model. The initial configuration update specifically refers to: By calling the training configuration parameters from the previous round and combining them with the quality evaluation training samples for incremental training, the incrementally trained quality prediction model and its corresponding training configuration parameters are obtained, and these are used as the updated training configuration parameters, thereby completing the initial configuration update of the quality prediction model. The training configuration parameters include the learning rate, regularization coefficient, and optimizer momentum parameter.

10. A system applying the robot-based automated control method for cold extrusion molding as described in any one of claims 1-9, characterized in that, include: The module includes an initial configuration module, a model building module, a construction instruction acquisition module, a secondary correction module, and a model configuration update module. The initial configuration module is used to move the blank to a designated position and simultaneously initialize the cold extrusion equipment after the robot receives the part processing instruction. The model building module is used to start the cold extrusion equipment, obtain real-time construction execution parameters, and analyze them with the historical construction execution parameter set to obtain the historical construction training set, thereby training the quality prediction model. The construction instruction acquisition module is used to acquire real-time construction execution parameters again and input them into the quality prediction model, output real-time defect risk values, acquire real-time cylinder pressure values, analyze and obtain filling status labels, and obtain construction execution instructions based on real-time defect risk values ​​and filling status labels. The secondary correction module is used to adjust the operating status of the actuator cylinder based on the construction execution command, and obtain the operating status of the actuator cylinder after adjustment to obtain the actuator cylinder adjustment judgment label. If the actuator cylinder adjustment judgment label is that the adjustment meets the standard, the adjustment is completed. If the actuator cylinder adjustment judgment label is that the speed reduction does not meet the standard, the speed reduction secondary correction process is performed. If the actuator cylinder adjustment judgment label is that the speed increase does not meet the standard, the speed increase secondary correction process is performed. The model configuration update module is used to obtain the quality parameters of the finished part after the part processing is completed, and thereby perform the initial configuration update of the quality prediction model.

Images