Intelligent scheduling method for efficient wire embedding machine and press integrated production line

Abstract

The invention discloses an intelligent scheduling method for an efficient wire embedding machine and press integrated production line, and relates to the technical field of intelligent scheduling, operating parameters and raw material attributes of a wire embedding machine and a press are collected in real time through a high-precision sensor, a multi-dimensional dynamic data set is formed, and a prediction model is constructed by combining a historical data set and utilizing machine learning. The model calculates the optimal starting time and performs real-time calibration according to elements such as the wire embedding speed, the pressing force and the wire thickness. The system triggers dynamic regulation and control through deviation analysis so as to ensure synchronous operation of equipment and improve the production efficiency. And quality detection feedback further optimizes model parameters, improves prediction precision and enhances adaptive ability, so that continuous improvement is realized in a complex production environment, and the limitation that a traditional method is difficult to deal with dynamic changes is broken through. And through real-time data collection and regular updating, machine performance and raw material fluctuation can be adapted, and synchronous improvement of production quality and efficiency is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent scheduling technology, and particularly to an intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press. Background Art

[0002] Under the background of the rapid development of modern manufacturing industry, wire embedding technology and pressing process have become core technologies indispensable in many high-precision production fields. These technologies are widely used in multiple scenarios such as electronic product assembly, automotive parts production, and new energy fields. For example, in circuit board manufacturing, a wire embedding machine needs to precisely embed wires into predetermined positions, and then a press applies appropriate pressure to ensure the firm bonding of the wires and the substrate; in automotive wire harness assembly, wire embedding and pressing processes are used to achieve efficient wire layout and fixation to meet complex electrical connection requirements; in the production of new energy battery modules, this technology is used for the embedding and compaction of electrode connection wires to ensure the high performance and safety of battery components. With the continuous improvement of market requirements for product performance, reliability, and production efficiency, the collaborative work of wire embedding machines and presses is increasingly developing towards integration, automation, and intelligence. By building an integrated production line, enterprises can reduce manual intervention while achieving seamless connection of production processes, thereby improving overall manufacturing efficiency and meeting the needs of mass customization production.

[0003] Although the wire embedding and pressing processes can theoretically achieve efficient collaboration, in the actual production environment, the precise synchronization of operation timing is still a key technical problem restricting their performance. Specifically, during the dynamic operation of the wire embedding machine and the press, they are affected by various factors, making it difficult to maintain stable timing for their collaborative operation. For example, the operating speed of the wire embedding machine may experience slight fluctuations due to mechanical wear, motor response characteristics, or control system delays, while the action timing of the press is restricted by the response time of the hydraulic system or the transmission delay of sensor signals. At the same time, subtle differences in raw material properties (such as the diameter, hardness, or surface friction coefficient of wire materials) further exacerbate the complexity of timing matching. If the wire embedding speed instantaneously increases and the press fails to follow up in time, the wire may not be fully fixed, resulting in risks of poor contact or detachment; conversely, if the press starts too early, it may apply pressure before the wire is fully embedded, causing position deviation or material damage.

[0004] Traditional scheduling methods are usually based on fixed parameters and preset operation sequences, and it is difficult to adapt to the above-mentioned dynamic changes in real time, thus easily leading to consequences such as reduced production efficiency, decreased product consistency, and even increased scrap rate. Therefore, how to ensure the precise synchronization of wire embedding and pressing processes in the actual production scenario with multi-variable interference has become a core technical challenge that urgently needs to be solved.

[0005] For this reason, an intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press is provided. Summary of the Invention

[0006] (1) Technical Problem to be Solved

[0007] Aiming at the deficiencies of the prior art, the present invention provides an intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press. By using high-precision sensors to collect the operation parameters of the wire embedding machine and the press and the raw material properties in real time, a multi-dimensional dynamic data set is formed, and a prediction model is constructed by using machine learning in combination with the historical data set. The model calculates the optimal starting time and calibrates it in real time according to factors such as wire embedding speed, pressing force, and wire thickness. The system triggers dynamic regulation through deviation analysis to ensure the synchronous operation of the equipment and improve production efficiency. The quality inspection feedback further optimizes the model parameters, improves the prediction accuracy and enhances the adaptive ability, so as to achieve continuous improvement in a complex production environment and break through the limitations that traditional methods are difficult to cope with dynamic changes. Through real-time data collection and regular update, it can adapt to machine performance and raw material fluctuations, ensure the synchronous improvement of production quality and efficiency, and solve the technical problems recorded in the background art.

[0008] (2) Technical Solution

[0009] To achieve the above object, the present invention is realized through the following technical solutions: An intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press, including that when the real-time data collection is started, the sensor performs high-frequency measurement and integration operations on the operation parameters of the wire embedding machine and the press and the raw material properties, and forms a real-time data set D re ;

[0010] When the synchronous model construction is activated, the machine learning algorithm constructs a synchronous model M re using regression analysis based on the historical real-time data set D sc , generating the optimal starting time for pressing

[0011] When the real-time operation adjustment is triggered, receive the real-time data set D re and combine it with the synchronous model M sc to calculate the optimal pressing time Perform dynamic adjustment on the prediction and actual deviation Δt, and optimize the speeds of the wire embedding machine and the press;

[0012] When the quality feedback and model optimization are started, collect the output quality data and feedback to optimize the synchronous model M sc , identify defects by comparing with the threshold and update the model parameters, obtain the optimized model and redeploy it to transform the data form from the detection result to the optimized input;

[0013] Preferably, through a high-precision sensor network installed on the wire embedding machine and the press, the following operation parameters are monitored and recorded in real time: the wire embedding speed v of the wire embedding machine b, the pressing force f of the press p , the operation times of the wire embedding machine and the press are t b and t p ; to form a time series data set D pm = {(t, v b , f p , t b , t p ) | t ∈ T}, where T represents the time set; use a material property sensor to measure and record the key properties of the raw material in real time: wire diameter d w , material hardness h m ; to form an attribute data set D ml = {(t, d w , h m ) | | t ∈ T};

[0014] Preferably, integrate the collected operating parameters and attribute data into a real-time data set:

[0015] D re = {(t, v b , f p , t b , t p , d w , h m ) | t ∈ T}; perform outlier detection and data smoothing to generate a high-quality standardized data set D sd ;

[0016] Preferably, extract key features from the real-time data set D re , introduce the order lag v b (t - 1) of the wire embedding speed, and the feature set is defined as X = [v b , f p , t b , t p , d w , h m , v b (t - 1)], and the target variable is the optimal starting time for pressing

[0017] Preferably, use the support vector regression algorithm to construct a synchronization model M sc , and the support vector regression algorithm can effectively handle the non-linear relationship between the feature set X and the target variable , the model input is the feature set X, and the output is the optimal starting time

[0018] The optimization objective of the support vector regression algorithm is defined as follows:

[0019]

[0020] Constrained conditions:

[0021]

[0022] where: w is the weight vector; b is the bias; C is the penalty parameter; ∈ is the tolerance error; ξ i 、 are slack variables; <w,x i > is the inner product of the feature vector x i and the weight w;

[0023] Preferably, the root mean square error RMSE is used to evaluate the model performance. If the RMSE exceeds the preset threshold, the model is optimized by adjusting the feature selection or the parameters of the support vector regression algorithm;

[0024] Deploy the trained synchronization model M se to the intelligent scheduling system, and set a regular update mechanism: every fixed period, use the newly collected real-time data set D re to retrain the model and calculate the RMSE of the new model. If the RMSE of the new model is lower than that of the current model, update the deployment.

[0025] Preferably, the intelligent scheduling system receives the real-time data set D re in real time, obtains the synchronization model M sc , and uses the real-time data set D re as the input to predict the crimping start time

[0026] Calculate the predicted crimping start time and the timing deviation Δt of the actual crimping operation time t p . The formula is:

[0027]

[0028] where: t p is the actual crimping operation time;

[0029] If Δt > 0, the crimping operation is lagging, and the press speed needs to be increased or started earlier; if Δt < 0, the crimping operation is ahead, and the speed of the wire embedding machine needs to be slowed down or started later; if Δt = 0, the operation is consistent with the prediction and no adjustment is required;

[0030] Preferably, to eliminate the timing deviation Δt, the speed of the wire embedding machine or the press is dynamically adjusted according to the magnitude of the deviation. The adjustment strategy is as follows:

[0031] Press speed adjustment: If Δt > 0, increase the press speed v p , and the adjustment amplitude is calculated by the following formula:

[0032] Δv p = kp ·Δt

[0033] The adjusted press speed is:

[0034]

[0035] Wire embedding machine speed adjustment: If Δt < 0, slow down the wire embedding machine speed v b , and the adjustment range is:

[0036] Δv b = k b ·|Δt|

[0037] The adjusted wire embedding machine speed is:

[0038]

[0039] Where: v p is the press speed before adjustment; v b is the wire embedding machine speed before adjustment, Δv p is the press speed adjustment range, Δv b is the wire embedding machine speed adjustment range, k p is the press speed adjustment coefficient, determined by the press performance and process requirements; k b is the wire embedding machine speed adjustment coefficient, determined by the wire embedding machine performance; is the adjusted press speed, is the adjusted wire embedding machine speed;

[0040] Preferably, record the adjusted operation data, including the adjusted wire embedding speed press speed actual pressing start time and the timing cumulative error TCE(t), and the calculation formula is:

[0041]

[0042] Where: w is the sliding window size, indicating the evaluation time range, θ is the forgetting factor, 0 < θ ≤ 1; Δq is the quality deviation (such as wire position accuracy or unqualified rate of pressing strength); k is the quality sensitivity coefficient, with a value greater than 0, controlling the amplification effect of quality on the error;

[0043] Construct the adjustment data set D at , defined as Where: is the actual pressing start time after adjustment, measuring the deviation between the operation after adjustment and the prediction;

[0044] Preferably, collect the output quality data in real time, including the wire position accuracy p w and the pressing strength sp , preset quality standard thresholds: wire position accuracy standard p std and lamination strength standard s std ;

[0045] The condition for qualified quality is p w ≤p std and s p ≥s std . If not satisfied, unqualified data is recorded, including the real-time data set D at the corresponding moment re , the adjusted data set D at and the predicted lamination start time

[0046] Preferably, for unqualified outputs, the quality deviation index q is calculated dev . The quality parameters of the product are represented by the vector q = [p w , s p T . The standard quality vector is q std = [p std , s std T . The quality deviation vector is defined as:

[0047] Δq = q - q std

[0048] The quality deviation index q dev is defined as follows:

[0049]

[0050] In the formula: The meaning of ||A·Δq||2 is: The deviation vector Δq is transformed by the quality influence matrix A, and then its Euclidean norm ||·||2 is calculated, which is used to measure the static deviation of the quality parameters;

[0051] The matrix A represents the mutual influence between different quality parameters; The parameter λ is the weight coefficient of the dynamic term;

[0052] Preferably, a quality feedback threshold q is set th . If q dev > q th , trigger the optimization of the synchronization model; Collect the feedback data set D fk , including the real-time data set D corresponding to unqualified outputs re , the adjusted data set D at , the predicted lamination start time and the quality data p w , s p . The sample weights are distributed proportionally according to q dev ;

[0053] Based on the feedback data set D​​fk , the incremental support vector regression is used to optimize the synchronization model M sc , and the update formula is as follows:

[0054]

[0055] In the formula: w is the model weight vector, and w new is the optimized weight vector; is the adjusted crimping start time; η is the learning rate, which controls the update step size, and α i is the Lagrange multiplier of sample i, which is proportional to the quality deviation index q dev ;

[0056] is the actual crimping start time, is the predicted crimping start time, and K(D i , D) is the kernel function; m is the number of feedback samples; the optimized model is redeployed. If p wire ≤ p std and s press ≥ s std increases the proportion, the optimized synchronization model is retained Otherwise, adjust the learning rate η or the quality feedback threshold q th , and re - execute the synchronization model optimization;

[0057] (III) Beneficial effects

[0058] The present invention provides an intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press, with the following beneficial effects:

[0059] Through high - frequency real - time data acquisition, it comprehensively senses the dynamic changes of operating parameters such as wire embedding speed and crimping force, as well as the raw material properties, providing a reliable data basis for the synchronization model. The synchronization model uses the support vector regression algorithm, combines lag features to accurately predict the best crimping start time, breaking through the limitations of traditional fixed - rule scheduling. Real - time operation adjustment is through a dynamic mechanism driven by time - series deviation, flexibly adjusting the machine speed to ensure the precise synchronization of the wire embedding and crimping processes, significantly improving the production rhythm and consistency;

[0060] In the quality feedback and model optimization link, by detecting data such as wire position accuracy and crimping strength, the incremental support vector regression (ISVR) with quality deviation weighted sampling is used to optimize the model online, continuously improving the prediction accuracy and quality control effect. The closed - loop verification mechanism ensures the effectiveness of optimization, significantly reducing the scrap rate and solving the problem of unstable quality caused by multi - variable interference in the background technology;

[0061] Ensuring data reliability through high-precision sensors and data preprocessing techniques, combined with regular model updates and dynamic adjustment strategies, can flexibly cope with changes in machine performance and raw material properties. This adaptive ability enables the system to maintain efficient operation in complex production environments, breaking through the limitations of traditional methods that are difficult to handle dynamic changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic flowchart of the intelligent scheduling method for the integrated production line of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0064] Please refer to Figure 1 , the present invention provides an intelligent scheduling method for an integrated production line of an efficient wire embedding machine and a press, including:

[0065] Step 1: When real-time data collection is started, the sensor performs high-frequency measurement and integration operations on the operating parameters of the wire embedding machine and the press and the raw material properties to form a real-time data set D re ;

[0066] The said step 1 includes the following contents:

[0067] Step 101: Collection of operating parameters

[0068] Through a high-precision sensor network installed on the wire embedding machine and the press, the following operating parameters are monitored and recorded in real time:

[0069] The wire embedding speed v of the wire embedding machine b , the pressing force f of the press p , the operating times of the wire embedding machine and the press are t b and t p respectively;

[0070] The collection frequency is set to a level higher than the production beat to capture the instantaneous fluctuations of the parameters. The operating parameters are aligned by time stamps to form a time series data set D pm ={(t, v b , f p , t b , t p )∣t∈T}, where T represents the time set;

[0071] Use a material property sensor to measure and record the key properties of raw materials in real time: wire diameter d w , material hardness h m ; These properties may change due to batch or environmental factors, so continuous monitoring is required; the property data is also aligned by timestamp to form a property dataset D ml = {(t, d w , h m ) | t ∈ T};

[0072] By collecting the operating parameters of the wire embedding machine and the press in real time with high-precision sensors, small fluctuations in machine performance can be captured. This not only provides dynamic and comprehensive basic data for the subsequent synchronization model but also improves the timeliness of the data through high-frequency collection, ensuring the system's real-time perception ability of the production status, thereby improving the accuracy of prediction.

[0073] Use a material property sensor to monitor the key properties of raw materials in real time and align the operating parameter data by timestamp to ensure the ability to sense changes in raw material batches or environmental factors, enhancing the robustness and adaptability to changes in the production environment and further improving the prediction accuracy and production stability.

[0074] Step 102, Data integration and preprocessing

[0075] Integrate the collected operating parameters and property data into a real-time dataset D re = {(t, v b , f p , t b , t p , d w , h m ) | t ∈ T}; To ensure the quality of the collected data, perform outlier detection and data smoothing; standardize the real-time dataset D re using the min-max normalization method to scale each parameter to the interval [0, 1]. The formula is:

[0076]

[0077] where x is the original parameter value, x * is the standardized value, and min(x) and max(x) are the minimum and maximum values of the parameter within the sampling period respectively; the standardized dataset is denoted as

[0078] Collect the operating parameters (v b , f p , t b , t p ) of the wire embedding machine and the press and the raw material properties (d w , h m), and through integration, preprocessing, and standardization, a high-quality standardized dataset D is generated sd ; The data is scaled to the range [0, 1] through min-max standardization to eliminate the difference in parameter dimensions and improve the convergence speed and stability of model training.

[0079] Step 2: When the synchronization model construction is activated, the machine learning algorithm is based on the historical real-time dataset D re Build a synchronization model M using regression analysis sc , generating the best start time for lamination

[0080] The said Step 2 includes the following contents:

[0081] Step 201: Feature engineering and data preparation

[0082] Extract key features from the real-time dataset D re , including the wire embedding speed v b , the lamination force f p , the operation time t of the wire embedding machine b , the operation time t of the press p , the wire diameter d w , the material hardness h m ;

[0083] To capture the time series dependence, lag features are introduced, such as the first-order lag v of the wire embedding speed b (t - 1), reflecting the impact of historical states on the current synchronization. The final feature set is defined as X = [v b , f p , t b , t p , d w , h m , v b (t - 1)], and the target variable is the best start time for lamination which is calculated from the difference between the wire embedding completion time and the lamination start time in historical data;

[0084] For example, if the wire embedding speed v b suddenly increases, the lamination needs to start earlier to match the progress. The lag feature v b (t - 1) can help the model identify this speed change trend; Set the window size to 5 time units (e.g., 5 seconds). At each time point t, the system extracts the data of the previous time point from within the window and calculates and updates v b (t - 1) in real time;

[0085] Example: If the current time t = 10 seconds, then v b (t - 1) uses the wire embedding speed data at t = 9 seconds, and the window continuously scrolls to keep the data up-to-date.

[0086] Extracting key features and introducing lag features to capture temporal dependencies significantly enhances the model's perception of production dynamics. The high-dimensional design of feature set X enriches the data representation ability and provides strong support for predicting the optimal start time of lamination, improving the prediction resolution of the model;

[0087] Step 202, Model Selection and Training

[0088] Construct a synchronous model M using the support vector regression algorithm sc , and the support vector regression algorithm can effectively handle the non-linear relationship between feature set X and the target variable . The model input is feature set X, and the output is the optimal start time The optimization objective of the support vector regression algorithm is defined as follows:

[0089]

[0090] Subject to the constraint:

[0091]

[0092] In the formula: w is the weight vector, which determines the contribution of features to the prediction; b is the bias, which adjusts the benchmark of the predicted value; C is the penalty parameter, which controls the model's tolerance for errors, and its value range is (0, ∞); ∈ is the tolerance error, which defines the acceptable range of prediction errors, and its value range is [0, ∞); ξ i , are slack variables, which measure the excess error, and their value range is [0, ∞); <w, x i > is the inner product of the feature vector x i and the weight w;

[0093] Construct a synchronous model M using the support vector regression algorithm sc , and effectively fit the non-linear relationship between feature set X and the optimal start time through the optimization objective function. By minimizing the weight norm ∥w∥ and the error penalty term, seek the non-linear mapping between features and 2 , ensure the balance between prediction accuracy and model complexity. The penalty parameter C is determined through grid search and cross-validation to optimize the model's generalization ability; If the wire embedding speed v

[0094] has a non-linear relationship with the optimal start time b (such as the reduction in the lamination time advance amplitude after the speed increases to a certain extent), the support vector regression algorithm can capture this trend through a kernel function (such as the radial basis function).

[0095] ​​It should be noted that the support vector regression algorithm model adopts the radial basis function (RBF) kernel, and the form of the kernel function is:

[0096] K(x,x′)=exp(-γ||x-x′|| 2 )

[0097] Parameter setting: γ = 0.1, which is optimized by grid search combined with 5-fold cross-validation;

[0098] Step 203, Model evaluation and optimization

[0099] Use the root mean square error RMSE to evaluate the model performance, and the calculation formula is:

[0100]

[0101] In the formula: is the model's predicted crimp start time, is the actual crimp start time, and n is the number of samples;

[0102] RMSE quantifies the deviation between the predicted value and the true value, guiding the optimization of the model; if RMSE exceeds the preset threshold, optimize the model by adjusting feature selection or support vector regression algorithm parameters (such as increasing the penalty parameter C or decreasing), until the accuracy requirement is met;

[0103] Use the root mean square error (RMSE) to quantify the model performance, and optimize the model accuracy by dynamically adjusting parameters (such as the penalty parameter C and the tolerance error). The evaluation result of RMSE provides a clear direction for optimization (such as reducing the error from 0.05 seconds to 0.03 seconds), ensuring that the prediction error is controlled within the acceptable range of production, and improving the practicability and reliability of the model.

[0104] Step 204, Model deployment and update mechanism

[0105] Deploy the trained synchronization model M se to the intelligent scheduling system, and set a regular update mechanism to adapt to changes in the production environment (such as fluctuations in the hardness h m of raw materials): Every fixed period, use the newly collected real-time data set D re to retrain the model and calculate the RMSE of the new model. If the RMSE of the new model is lower than the current model, update the deployment.

[0106] Deploy the trained synchronization model M sc to the intelligent scheduling system, and maintain its adaptability to changes in the production environment through a regular update mechanism. After the update, if the RMSE improves, the model is automatically deployed to ensure the stability and prediction accuracy of the long-term operation of the system;

[0107] During use, a synchronous model M for embedding and pressing is constructed and optimized. sc Using feature engineering, a multi-dimensional feature set X is extracted, and the optimal start time of pressing is predicted through the support vector regression algorithm. The accuracy and adaptability are ensured through RMSE evaluation and regular update mechanisms.

[0108] Step 3: When a real-time operation adjustment is triggered, receive the real-time data set D. re Combined with the synchronous model M. sc Calculate the optimal pressing time. Implement dynamic adjustment for the deviation Δt between the prediction and the actual situation, and optimize the speeds of the wire embedding machine and the press.

[0109] The content of Step 3 is as follows:

[0110] Step 301: Receive real-time data and model prediction.

[0111] The intelligent scheduling system receives the real-time data set D in real time. re , including the wire embedding speed v. b , the pressing force f. p , the operation time t of the wire embedding machine. b and the operation time t of the press. p , the wire diameter d. w , the material hardness h. m ; Obtain the synchronous model M. sc , and use the real-time data set D. re as the input to predict the pressing start time. The calculation formula is:

[0112]

[0113] In the formula: is the predicted pressing start time, indicating the ideal time point when the press should start operating.

[0114] During use, the intelligent scheduling system receives the real-time data set D. re , using the synchronous model M. sc Predict the optimal pressing start time. The prediction process has a fast response speed and seamless connection with the production rhythm, ensuring that the system can respond to dynamic changes in a timely manner and providing an accurate time benchmark for subsequent adjustments.

[0115] Step 302: Calculate the timing deviation.

[0116] Calculate the predicted pressing start time. The actual pressing operation time t. p The timing deviation Δt is calculated, and the formula is:

[0117]

[0118] Where: Δt is the timing deviation, representing the difference between the actual operation and the predicted time, and t p is the actual pressing operation time;

[0119] If Δt > 0, the pressing operation is lagging, and the press speed needs to be increased or started earlier; if Δt < 0, the pressing operation is ahead, and the speed of the wire embedding machine needs to be slowed down or the start delayed; if Δt = 0, the operation is consistent with the prediction and no adjustment is required; for example, if seconds, and t p = 5.2 seconds, then Δt = 0.2 seconds, indicating that the pressing is lagging by 0.2 seconds, and acceleration is required to make up for the time difference.

[0120] During use, calculate the timing deviation Δt between the predicted time and the actual operation time t p . This can provide a clear adjustment direction and amplitude. The quantitative analysis of the deviation enhances the scientific nature and pertinence of the adjustment, ensuring the efficiency and controllability of operation optimization.

[0121] Step 303. Dynamically adjust the machine speed

[0122] To eliminate the timing deviation Δt, dynamically adjust the speed of the wire embedding machine or the press according to the magnitude of the deviation. The adjustment strategy is as follows:

[0123] Press speed adjustment: If Δt > 0, increase the press speed v p , and the adjustment amplitude is calculated by the following formula:

[0124] Δv p = k p ·Δt

[0125] The adjusted press speed is:

[0126]

[0127] Wire embedding machine speed adjustment: If Δt < 0, slow down the speed of the wire embedding machine v b , and the adjustment amplitude is:

[0128] Δv b = k b ·|Δt|

[0129] The adjusted speed of the wire embedding machine is:

[0130]

[0131] Where: v p is the press speed before adjustment; v b is the speed of the wire embedding machine before adjustment, Δv p is the adjustment amplitude of the press speed, and Δv bThe speed adjustment range of the wire embedding machine, k p The speed adjustment coefficient of the press, determined by the press performance and process requirements; k b The speed adjustment coefficient of the wire embedding machine, determined by the wire embedding machine performance; The adjusted press speed, The adjusted wire embedding machine speed;

[0132] Where: k p And k b The calculation methods are as follows:

[0133]

[0134] In the formula: Δv max,p And Δv max,b : The maximum speed change ranges of the press and the wire embedding machine respectively (unit: m / s); Δt max,p And Δt max,b : The maximum response times of the press and the wire embedding machine respectively (unit: s).

[0135] Example: If the maximum speed change of the press Δv max,p = 0.5 m / s and the response time Δt max,p = 1 s, then k p = 0.5 m / s.

[0136] Based on the timing deviation Δt, dynamically adjust the speed of the wire embedding machine or the press. For example, increase the press speed v p to v' p to make up for the lag. Combining with the equipment performance parameters, achieve precise synchronization of the wire embedding and pressing processes. The adjustment strategy is flexible, improving the system's adaptive ability and response speed;

[0137] Step 304, Record and transfer the adjustment effect

[0138] Record the operation data after adjustment, including the adjusted wire embedding speed press speed actual pressing start time and the timing cumulative error TCE(t), and the calculation formula is:

[0139]

[0140] In the formula: w is the sliding window size, representing the evaluation time range, θ is the forgetting factor, 0 < θ ≤ 1; Δq is the quality deviation (such as the wire position accuracy or the unqualified rate of the pressing strength); k is the quality sensitivity coefficient, with a value greater than 0, controlling the amplification effect of quality on the error;

[0141] Integrate these data into the adjustment dataset D at , defined as Wherein: is the adjusted actual lamination start time, measuring the deviation between the adjusted operation and the prediction;

[0142] During use, record the adjusted operation data (such as ) and the timing cumulative error TCE(t), providing complete data support for quality feedback and model optimization. The comprehensiveness and systematicness of the records ensure the effectiveness of the feedback mechanism and promote the continuous improvement of the production process.

[0143] Step Four: When quality feedback and model optimization are initiated, collect the output quality data and feedback to optimize the synchronous model M sc , identify defects by comparing with the threshold and update the model parameters to obtain the optimized model and redeploy it, transforming the data form from the detection result to the optimized input;

[0144] Step 401: Quality data collection and evaluation

[0145] Real-time collect the output quality data through the quality inspection equipment, including the wire position accuracy p w and the lamination strength s p , preset the quality standard thresholds: the wire position accuracy standard p std and the lamination strength standard s std ; the quality qualification condition is p w ≤p std and s p ≥s std . If not satisfied, record the unqualified data, including the real-time data set D re , the adjusted data set D at and the predicted lamination start time

[0146] Real-time collect the output quality data through the quality inspection equipment, such as the wire position accuracy p w and the lamination strength s p . Compare with the standard threshold to accurately identify unqualified products. Real-time monitoring improves the timeliness of quality control and provides a reliable basis for deviation analysis and model optimization.

[0147] Step 402: Quality feedback and deviation analysis

[0148] For unqualified outputs, calculate the quality deviation index q dev , used to quantify the degree of quality defects, and introduce matrix operations and dynamic change analysis. Assume that the quality parameters of the product can be represented by a vector, for example, q = [p w , s p T (representing the position accuracy and lamination strength respectively), and the standard quality vector is q​std = [p std , s std T , the mass deviation vector is defined as:

[0149] Δq = q - q std

[0150] Based on this, the mass deviation index q dev is defined as follows:

[0151]

[0152] In the formula: The meaning of ||A·Δq||2 is: The deviation vector Δq is transformed by the mass influence matrix A, and then its Euclidean norm ||·||2 is calculated, which is used to measure the static deviation of the quality parameters;

[0153] The matrix A is a 2×2 matrix, which characterizes the mutual influence between different quality parameters. For example, the position accuracy p w may have an indirect impact on the lamination strength s p and the matrix elements can be determined through process analysis or data fitting; The meaning of is: By integrating the rate of change of the mass deviation vector over time (i.e., the Euclidean norm of), the dynamic trend of quality defects is captured; The parameter λ is the weight coefficient of the dynamic term, and its value is greater than 0, which is used to balance the importance of static deviation and dynamic change and can be adjusted according to actual needs; The time interval from t0 to t represents the evaluation time range, such as a certain stage in the production process.

[0154] Through the mass deviation index q dev the severity of quality defects is quantified, and the nonlinear design improves the accuracy and pertinence of feedback, highlighting significant quality problems and providing scientific guidance for model optimization.

[0155] Step 403, Model Optimization Trigger and Data Preparation

[0156] Set the quality feedback threshold q th , if q dev > q th , trigger synchronous model optimization; Collect the feedback data set D fk , which includes the real-time data set D corresponding to unqualified outputs re , the adjusted data set D at , the predicted lamination start time and the quality data p w , s p , to enhance the adaptability of the model to quality problems, the sample weights are distributed in proportion to q dev ;

[0157] ​Based on the feedback dataset D fk , the incremental support vector regression is used to optimize the synchronization model M sc , and the update formula is as follows:

[0158]

[0159] In the formula: w is the model weight vector, and w new is the optimized weight vector; is the adjusted crimping start time; η is the learning rate, controlling the update step size, and α i is the Lagrange multiplier of sample i, which is proportional to the quality deviation index q dev ;

[0160] is the actual crimping start time, is the predicted crimping start time, K(D i , D) is the kernel function, measuring the similarity between sample D i and the current input D; m is the number of feedback samples;

[0161] The optimized model is redeployed to verify the quality improvement effect in subsequent production; if p wire ≤p std and s press ≥s std increases, the optimized synchronization model is retained, otherwise the learning rate η or the quality feedback threshold q th is adjusted, and the synchronization model optimization in step 404 is re-executed;

[0162] Set the quality feedback threshold q th (such as 0.1), and optimization is triggered only when q dev >q th to avoid wasting resources. The weighted sampling mechanism based on the quality deviation index q dev makes the model pay more attention to serious quality problems, improving the optimization efficiency and pertinence; the incremental support vector regression is used to optimize the synchronization model M sc online, and the weight w new is dynamically updated to adapt to production changes. The deviation-oriented optimization mechanism ensures the continuous improvement of the model prediction accuracy and enhances the system stability. The optimized model is deployed to step 3, and the quality improvement effect is verified in production (such as p wire deviation is reduced to 0.05 mm). The closed-loop verification mechanism ensures the effectiveness of optimization and improves the sustainability of production quality and the system reliability.

[0163] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0164] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0165] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only for some logical function divisions. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be electrical, mechanical, or other forms.

[0166] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0167] As described above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, and all should be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. An intelligent scheduling method for an integrated production line of an efficient buried wire machine and a press, characterized in that: including When the real-time data collection is started, the sensor performs high-frequency measurement and integration operations on the operating parameters of the wire embedding machine and the press and the raw material properties to form a real-time data set D re ; When the synchronous model construction is activated, the machine learning algorithm is based on the historical real-time data set D re Adopt regression analysis to construct the synchronous model M sc , and generate the best start time for lamination When a real-time operation adjustment is triggered, receive the real-time data set D re Combine with the synchronization model M sc Calculate the optimal pressing time Implement dynamic adjustment for the prediction and actual deviation Δt, and optimize the speeds of the wire embedding machine and the press When quality feedback and model optimization are initiated, collect output quality data and synchronously feedback and optimize model M sc , identify defects by comparing with the threshold and update model parameters to obtain the optimized model and redeploy it to transform the data form from the detection result to the optimized input.

2. The intelligent scheduling method according to claim 1, characterized in that: Through a high-precision sensor network installed on the wire embedding machine and the press, the operating parameters are monitored and recorded in real time: The wire embedding speed v of the wire embedding machine b , the pressing force f of the press p , and the operation times of the wire embedding machine and the press are t b and t p respectively; a time series data set D pm ={(t, v b , f p , t b , t p )|t ∈ T}, where T represents the time set; Use a material property sensor to measure and record in real time the key properties of raw materials: wire diameter d w , material hardness h m ; form an attribute data set D ml = {(t, d w , h m ) | t ∈ T}.

3. The intelligent scheduling method according to claim 2, characterized in that: Integrate the collected operating parameters and attribute data into a real-time data set D re , extract key features from the real-time data set D re , introduce the order lag v b (t - 1) of the wire embedding speed, and then define the feature set X. The target variable is the best start time of pressing Construct a synchronization model M using the support vector regression algorithm sc , and the support vector regression algorithm can effectively handle the non-linear relationship between the feature set X and the target variable . The input of the model is the feature set X, and the output is the optimal start time 4. The intelligent scheduling method according to claim 3, characterized in that: The root mean square error RMSE is used to evaluate the model performance. If the RMSE exceeds the preset threshold, the model is optimized by adjusting the feature selection or the parameters of the support vector regression algorithm; Deploy the trained synchronization model M se to the intelligent scheduling system and set a regular update mechanism: use the newly collected real-time dataset D re to retrain the model and calculate the RMSE of the new model. If the RMSE of the new model is lower than that of the current model, update the deployment.

5. The intelligent scheduling method according to claim 4, characterized in that: The intelligent scheduling system receives the real-time data set D in real time re , obtains the synchronization model M sc , and uses the real-time data set D re as the input to predict the lamination start time Calculate the predicted crimping start time Actual crimping operation time t p The timing deviation Δt. If Δt > 0, the crimping operation lags, and the press speed needs to be increased or the start time advanced; if Δt < 0, the crimping operation is ahead, and the speed of the wire embedding machine needs to be slowed down or the start time delayed; if Δt = 0, the operation is consistent with the prediction and no adjustment is required.

6. The intelligent scheduling method according to claim 5, characterized in that: The speed of the wire embedding machine or the press is dynamically adjusted according to the magnitude of the deviation, and the adjustment strategy is as follows: Press speed adjustment: If Δt > 0, increase the press speed v p , and the adjustment range is: Δv p = k p ·Δt; The adjusted press speed is: Wire embedding machine speed adjustment: If Δt < 0, slow down the wire embedding machine speed v b , and the adjustment range is: Δv b = k b ·|Δt|; The adjusted wire embedding machine speed is: where: v p is the press speed before adjustment; v b is the wire embedding machine speed before adjustment, Δv p is the press speed adjustment range, Δv b is the wire embedding machine speed adjustment range, k p is the press speed adjustment coefficient; k b is the wire embedding machine speed adjustment coefficient, determined by the performance of the wire embedding machine; is the press speed after adjustment, is the wire embedding machine speed after adjustment.

7. The intelligent scheduling method according to claim 6, characterized in that: Record the adjusted operating data, including the adjusted wire embedding speed Press speed Actual pressing start time And the timing cumulative error TCE(t), and summarize and construct the adjusted data set D at , defined as The calculation formula of the time series cumulative error TCE(t) is: where: w is the sliding window size, is the forgetting factor, Δq is the mass deviation, and k is the mass sensitivity coefficient.

8. The intelligent scheduling method according to claim 7, characterized in that: Real-time collect the output quality data, including the wire position accuracy p w and the lamination strength s p , preset the quality standard thresholds: the wire position accuracy standard p std and the lamination strength standard s std ; The quality qualification condition is p w ≤p std and s p ≥s std , if not satisfied, record the unqualified data, including the real-time data set D at the corresponding moment re , adjust the data set D at and the predicted crimping start time 9. The intelligent scheduling method according to claim 8, characterized in that: Calculate the quality deviation index q for nonconforming output dev , the quality parameters of the product are represented by the vector q = [p w , s p T , the standard quality vector is q std = [p std , s std T , the quality deviation vector is defined as: Δq = q - q std ;​​ Quality deviation index q dev is defined as follows: In the formula: The meaning of ||A·Δq||2 is: The deviation vector Δq is transformed through the quality influence matrix A, and then its Euclidean norm ||·||2 is calculated, which is used to measure the static deviation of the quality parameters; The matrix A represents the mutual influence between different quality parameters; The parameter λ is the weight coefficient of the dynamic term.

10. The intelligent scheduling method according to claim 9, characterized in that: Set the quality feedback threshold q th , if q dev >q th , trigger synchronous model optimization; collect the feedback data set D fk , including the real-time data set D corresponding to unqualified outputs re , adjust the data set D at , predict the pressing start time and the quality data p w , s p , the sample weights are proportionally allocated according to q dev ; Based on the feedback dataset D fk , the incremental support vector regression is used to optimize the synchronization model M sc ; Redploy the optimized model If p wire ≤ p std and s press ≥ s std The proportion increases, then retain the optimized synchronization model Otherwise, adjust the learning rate η or the quality feedback threshold q th and re - execute the optimization of the synchronization model

Images