Correction method and system of double-station solder ball welding machine

Abstract

The invention discloses a correction method and system of a double-station solder ball welding machine, and belongs to the technical field of welding equipment correction, and the correction method comprises the following steps: arranging various sensors on the double-station solder ball welding machine, and collecting welding data; preprocessing the collected welding data to obtain welding processing data; performing feature extraction on the welding processing data through a feature layer fusion method to obtain a welding feature vector; building a correction prediction model based on a support vector machine, and predicting the welding feature vector obtained in real time to obtain a prediction result; deviation detection judgment is conducted on abnormal data in the prediction result, and welding data needing to be corrected are obtained; and based on the welding data needing to be corrected, corresponding correction measures are taken for the welding data with deviation in each station.

Description

Technical Field

[0001] The invention relates to the technical field of welding equipment correction, and in particular to a correction method and system for a double-station solder ball welding machine. Background Art

[0002] As a type of solder ball welding equipment, the dual-station solder ball welding machine has two working positions and can perform soldering operations on two different soldering points at the same time, greatly improving production efficiency. However, in the actual production process, due to the influence of various factors, such as equipment aging, ambient temperature changes, etc., the welding accuracy and stability of the dual-station solder ball welding machine often decrease, seriously affecting the quality and reliability of the product.

[0003] In addition, some methods are based on analysis based on only a single sensor data, so the information acquisition is not comprehensive and it is difficult to accurately reflect the real state of the welding process. There is a lack of advance prediction of whether there is a deviation in the dual-station solder ball welding machine, which is not conducive to improving welding efficiency.

[0004] The present invention is proposed in view of the above-mentioned problems, and aims to provide a correction method and system for a double-station solder ball welding machine to solve the deficiencies in the prior art. Summary of the invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a correction method and system for a double-station solder ball welding machine to solve the problems raised in the above background technology.

[0006] The purpose of the present invention can be achieved by the following technical solution: A correction method for a double-station solder ball welding machine comprises the following steps:

[0007] Step 1: Collect welding data by deploying various sensors on the dual-station solder ball welding machine; the welding data includes solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0008] Step 2: pre-process the collected welding data to obtain welding processing data; extract features from the welding processing data by a feature layer fusion method to obtain a welding feature vector;

[0009] Step 3: construct a correction prediction model based on support vector machine, predict the welding feature vector obtained in real time, and obtain the prediction result;

[0010] Step 4: Perform deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected;

[0011] Step 5: Based on the welding data that needs to be corrected, take corresponding corrective measures for the welding data with deviations in each workstation.

[0012] Preferably, in step 1, the various sensors include:

[0013] Using the laser tracker and industrial camera, the laser beam emitted by them is used as a reference to establish a three-dimensional reference coordinate system in the working space of the welding machine to obtain the position coordinates of the welding point;

[0014] Displacement sensor, used to measure the displacement data of welding head movement;

[0015] Force sensor, installed at the contact point between the welding head and the workpiece, used to measure the pressure data during welding;

[0016] Temperature sensor, used to measure the temperature of the welding area and obtain temperature data;

[0017] Motion sensor, used to measure the rotation angle of the motor;

[0018] The welding data of the double-station solder ball welding machine is determined based on the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle.

[0019] Preferably, in step 2, the welding data is preprocessed, including:

[0020] The 3σ criterion is adopted to regard the data that deviates from the mean by more than 3 times the standard deviation as outliers and eliminate or correct them to remove outliers in the welding data; a low-pass filter is used to filter the noise in the welding data; the normalization method is used to standardize the welding data with different dimensions and numerical ranges; the synchronization of welding data is achieved by aligning the timestamps when the data is collected and filling the missing values ​​with the interpolation method.

[0021] Preferably, feature extraction is performed on the welding process data, including:

[0022] Get displacement data through the formula Calculate the average speed of the welding head during welding and determine the displacement characteristics; where v is the average speed of the welding head, Δw is the displacement data of the welding head; Δt is the welding time;

[0023] Obtain pressure data, calculate the maximum pressure, minimum pressure and average pressure during welding, and determine the pressure characteristics;

[0024] Obtain temperature data, calculate the temperature change difference between adjacent time points and divide it by the time interval to obtain the temperature change rate and determine the temperature characteristics;

[0025] Obtain the rotation angle of the motor and determine the rotation angle characteristics;

[0026] Based on the displacement features, pressure features, temperature features, and rotation angle features, they are fused through weighted summation according to preset weights to obtain a welding feature vector.

[0027] Preferably, constructing a correction prediction model includes:

[0028] Based on historical welding data, obtain the historical welding feature vector X m and the corresponding label Y m ; Wherein, m = 1, 2, ..., M, M is the number of historical welding feature vectors;

[0029] Label Y corresponding to the historical welding feature vector m Including normal and deviation, it can be expressed as:

[0030] Y m =1, indicating that the historical welding feature vector is normal welding data;

[0031] Y m =-1, indicating that the historical welding feature vector is welding data with deviation;

[0032] Taking the historical welding feature vector as the input sample, the decision function in the correction prediction model is constructed based on the support vector machine;

[0033] The decision function is expressed as:

[0034]

[0035] In the formula, f(x) is the decision function value; x is the sample to be tested; α m is the Lagrange multiplier, each input sample X m Corresponds to a Lagrange multiplier; K(X i ,x) is the kernel function, which is used to calculate the input sample X m The similarity with the sample x to be tested; b is the bias term, which is used to adjust the position of the decision boundary;

[0036] Use gradient descent to solve for α m and b; by calculating the loss function about α m and the gradient of b, and then update α in the opposite direction of the gradient m and b, gradually reduce the value of the loss function in an iterative manner, and finally find the minimum value of the loss function; wherein the loss function can select the hinge loss function;

[0037] According to the calculated decision function value, the corresponding label of the sample x to be tested is determined; if the decision function value f(x) ≥ 0, the prediction result is positive, that is, the label of the sample x to be tested is Y m =1; if the decision function value f(x) < 0, the prediction result is negative, that is, the label of the sample x to be tested is Y m = -1, and issue a warning prompt, indicating that there is a deviation;

[0038] The accuracy of the model is obtained by calculating the ratio of the number of samples correctly predicted by the model to the total number of predicted samples; the accuracy is compared with the preset standard value. If the obtained accuracy is greater than or equal to the preset standard value, it means that the model prediction is accurate and the performance of the model meets the standard; otherwise, the model is optimized until the calculated accuracy reaches the preset standard value.

[0039] Preferably, the method for obtaining the prediction result is as follows:

[0040] According to the constructed correction prediction model, the welding feature vector obtained in real time is predicted, the welding feature vector is input into the correction prediction model, and the label corresponding to the welding feature vector is determined and obtained according to the calculated decision function value, which is recorded as the prediction result;

[0041] The prediction results include normal data and abnormal data; the label is Y m = 1 is marked as normal data; the welding feature vector labeled as Y m = -1 is marked as abnormal data.

[0042] Preferably, in step 4, deviation detection and judgment are performed on abnormal data in the prediction results, including:

[0043] Obtain welding data from abnormal data, including welding point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0044] For the displacement data, the difference between the weld point position coordinates and the displacement data of the weld head movement is calculated to obtain the displacement deviation;

[0045] By comparing the displacement deviation WY with the preset standard displacement deviation value WYZ, if WY≤WYZ, it is determined that the position deviation meets the standard; otherwise, it is determined that the displacement data has deviation and needs to be corrected;

[0046] For pressure and temperature deviations, the pressure data is compared with the preset standard pressure range. If the pressure data is within the standard pressure range, it is determined that there is no deviation in the pressure data; otherwise, it is determined that there is a deviation in the pressure data and correction is required;

[0047] By comparing the temperature data with the preset standard temperature range, if the temperature data is within the standard temperature range, it is determined that the temperature data has no deviation; otherwise, it is determined that the temperature data has a deviation and needs to be corrected;

[0048] For the rotation angle, by comparing the rotation angle with a preset standard rotation angle value, if the rotation angle ≤ the standard rotation angle value, it means that the rotation angle meets the standard; otherwise, it is determined that there is a deviation in the rotation angle and correction is required.

[0049] Preferably, in step five, the corrective measures include:

[0050] For displacement data correction, the size and direction of the position deviation of the corresponding workstation are calculated based on the prediction results, and instructions are sent to the control system to control the movement of the X-axis, Y-axis, and Z-axis motors to move the welding head in the opposite direction by a corresponding distance, so that the welding point position returns to the correct position;

[0051] For pressure correction, the pressure output by the pressure applying device is adjusted in real time to ensure that the pressure during welding is always within the standard pressure range;

[0052] For temperature correction, the heating power or heating time is adjusted according to the prediction results. At the same time, the heating parameters are dynamically adjusted according to the real-time temperature changes in combination with the feedback control algorithm to ensure the stability of the welding temperature. Among them, the feedback control algorithm is an existing technology, and the specific process will not be described in detail.

[0053] For rotation angle correction, the angle compensation is performed by controlling the rotating motor and sending instructions to the motor controller to make the motor rotate the corresponding angle until the rotation angle is less than or equal to the standard rotation angle value.

[0054] In order to solve the above problems, the present invention also provides a correction system for a double-station solder ball welding machine, comprising:

[0055] The data acquisition module collects welding data by deploying various sensors on the dual-station solder ball welding machine; the welding data includes the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0056] The preprocessing and feature extraction module preprocesses the collected welding data to obtain welding processing data; extracts features from the welding processing data through the feature layer fusion method to obtain the welding feature vector;

[0057] The correction prediction module builds a correction prediction model based on the support vector machine, predicts the welding feature vector obtained in real time, and obtains the prediction result;

[0058] Deviation judgment module, which performs deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected;

[0059] The correction module takes corresponding corrective measures for the welding data with deviations in each workstation based on the welding data that needs to be corrected.

[0060] Compared with the existing solutions, the present invention achieves the following beneficial effects:

[0061] By deploying various sensors on a dual-station solder ball welding machine, the present invention can comprehensively and accurately collect welding data, effectively remove noise and interference, and ensure the accuracy of subsequent analysis. Compared with the traditional method of judging welding conditions based on experience, this greatly improves the scientificity and reliability of data acquisition and can significantly improve the stability of welding quality.

[0062] The present invention uses a feature layer fusion method to extract features from welding processing data, which can mine deep and representative features in the data and obtain accurate welding feature vectors;

[0063] The present invention uses a correction prediction model constructed based on a support vector machine to predict the welding feature vector acquired in real time and identify abnormal data, thereby improving production efficiency and reducing production costs.

[0064] The present invention can quickly and accurately obtain welding data that needs to be corrected by performing deviation detection and judgment on abnormal data in the prediction results; corresponding corrective measures are taken in time for welding data with deviations in each workstation to achieve dynamic adjustment of the welding process; this timely feedback and correction mechanism effectively reduces welding defects, improves the yield rate of products, enhances the competitiveness of products in the market, and provides strong support for enterprises to improve overall production levels and management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The present invention will be further described below in conjunction with the accompanying drawings.

[0066] Figure 1 The present invention is a flowchart of a correction method for a dual-station solder ball welding machine.

[0067] Figure 2 This is a module structure diagram of the correction system of the double-station solder ball welding machine proposed by the present invention. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0069] Embodiment 1, as Figure 1 As shown, the present invention is a correction method for a double-station solder ball welding machine, comprising the following steps:

[0070] Step 1: Collect welding data by deploying various sensors on the double-station solder ball welding machine; the welding data includes solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0071] In step 1, the various sensors include:

[0072] Using the laser tracker and industrial camera, the laser beam emitted by them is used as a reference to establish a three-dimensional reference coordinate system in the working space of the welding machine to obtain the position coordinates of the welding point;

[0073] Displacement sensor, used to measure the displacement data of welding head movement;

[0074] Force sensor, installed at the contact point between the welding head and the workpiece, used to measure the pressure data during welding;

[0075] Temperature sensor, used to measure the temperature of the welding area and obtain temperature data;

[0076] Motion sensor, used to measure the rotation angle of the motor;

[0077] Determine the welding data of the double-station solder ball welding machine based on the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0078] Step 2: pre-process the collected welding data to obtain welding processing data; extract features from the welding processing data by a feature layer fusion method to obtain a welding feature vector;

[0079] In step 2, the welding data is preprocessed, including:

[0080] The 3σ criterion is adopted to treat data that deviates from the mean by more than 3 times the standard deviation as outliers and eliminate or correct them to remove outliers in welding data; a low-pass filter is used to filter noise in welding data; a normalization method is used to standardize welding data with different dimensions and numerical ranges; the synchronization of welding data is achieved by aligning the timestamps when the data is collected and filling the missing values ​​with interpolation;

[0081] Furthermore, feature extraction is performed on the welding process data, including:

[0082] Get displacement data through the formula Calculate the average speed of the welding head during welding and determine the displacement characteristics; where v is the average speed of the welding head, Δw is the displacement data of the welding head; Δt is the welding time;

[0083] Obtain pressure data, calculate the maximum pressure, minimum pressure and average pressure during welding, and determine the pressure characteristics;

[0084] Obtain temperature data, calculate the temperature change difference between adjacent time points and divide it by the time interval to obtain the temperature change rate and determine the temperature characteristics;

[0085] Obtain the rotation angle of the motor and determine the rotation angle characteristics;

[0086] Based on the displacement features, pressure features, temperature features, and rotation angle features, the welding feature vector is obtained by weighted summation according to the preset weights.

[0087] Step 3: construct a correction prediction model based on support vector machine, predict the welding feature vector obtained in real time, and obtain the prediction result;

[0088] In step three, a correction prediction model is constructed, including:

[0089] Based on historical welding data, obtain the historical welding feature vector X m and the corresponding label Y m ; Wherein, m = 1, 2, ..., M, M is the number of historical welding feature vectors;

[0090] Label Y corresponding to the historical welding feature vector m Including normal and deviation, it can be expressed as:

[0091] Y m =1, indicating that the historical welding feature vector is normal welding data;

[0092] Y m =-1, indicating that the historical welding feature vector is welding data with deviation;

[0093] Taking the historical welding feature vector as the input sample, the decision function in the correction prediction model is constructed based on the support vector machine;

[0094] The decision function is expressed as:

[0095]

[0096] In the formula, f(x) is the decision function value; x is the sample to be tested; α m is the Lagrange multiplier, each input sample X m Corresponding to a Lagrange multiplier, representing the input sample X m Importance in the decision boundary; K(X i ,x) is the kernel function, which is used to calculate the input sample X m The similarity with the sample x to be tested; b is the bias term, which is used to adjust the position of the decision boundary;

[0097] It should be noted that through K(X i ,x) The kernel function maps the data to a high-dimensional space, so that the originally linearly inseparable data becomes separable in the new space, which is used to capture the data features;

[0098] Use gradient descent to solve for α m and b; by calculating the loss function about α mand the gradient of b, and then update α in the opposite direction of the gradient m and b, gradually reduce the value of the loss function in an iterative manner, and finally find the minimum value of the loss function; wherein the loss function can select the hinge loss function;

[0099] According to the calculated decision function value, the corresponding label of the sample x to be tested is determined; if the decision function value f(x) ≥ 0, the prediction result is positive, that is, the label of the sample x to be tested is Y m =1; if the decision function value f(x) < 0, the prediction result is negative, that is, the label of the sample x to be tested is Y m = -1, and issue a warning prompt, indicating that there is a deviation;

[0100] The accuracy of the model is obtained by calculating the ratio of the number of samples correctly predicted by the model to the total number of predicted samples;

[0101] The accuracy is compared with the preset standard value. If the obtained accuracy is greater than or equal to the preset standard value, it means that the model prediction is accurate and the performance of the model meets the standard; otherwise, the model is optimized until the calculated accuracy reaches the preset standard value; wherein the preset standard value is set according to the prediction accuracy of whether there is a deviation in the dual-station solder ball welding machine in actual application;

[0102] It should be noted that the optimization of the model includes regularly collecting new welding data, including successful and failed welding cases, retraining and optimizing the model, and improving the prediction accuracy of the model as the data continues to accumulate, further improving the stability of the welding quality;

[0103] Furthermore, the method for obtaining the prediction results is as follows:

[0104] According to the constructed correction prediction model, the welding feature vector obtained in real time is predicted, the welding feature vector is input into the correction prediction model, and the label corresponding to the welding feature vector is determined and obtained according to the calculated decision function value, which is recorded as the prediction result;

[0105] The prediction results include normal data and abnormal data; the label is Y m = 1 is marked as normal data; the welding feature vector labeled as Y m = -1 welding feature vector is marked as abnormal data;

[0106] Step 4: Perform deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected;

[0107] In step 4, deviation detection and judgment are performed on abnormal data in the prediction results, including:

[0108] Obtain welding data from abnormal data, including welding point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0109] For the displacement data, the difference between the weld point position coordinates and the displacement data of the weld head movement is calculated to obtain the displacement deviation;

[0110] By comparing the displacement deviation WY with the preset standard displacement deviation value WYZ, if WY≤WYZ, it is determined that the position deviation meets the standard; otherwise, it is determined that the displacement data has deviation and needs to be corrected;

[0111] For pressure and temperature deviations, the pressure data is compared with the preset standard pressure range. If the pressure data is within the standard pressure range, it is determined that there is no deviation in the pressure data; otherwise, it is determined that there is a deviation in the pressure data and correction is required;

[0112] By comparing the temperature data with the preset standard temperature range, if the temperature data is within the standard temperature range, it is determined that the temperature data has no deviation; otherwise, it is determined that the temperature data has a deviation and needs to be corrected;

[0113] For the rotation angle, by comparing the rotation angle with the preset standard rotation angle value, if the rotation angle is ≤ the standard rotation angle value, it means that the rotation angle meets the standard; otherwise, it is determined that the rotation angle has a deviation and needs to be corrected;

[0114] Among them, the standard displacement deviation value and the standard rotation angle value are set according to the specific requirements of experts in this field for deviation errors in actual applications;

[0115] The standard pressure range includes an upper pressure limit and a lower pressure limit, which are set according to the specific requirements for the pressure of the double-station solder ball welding machine in actual applications;

[0116] The standard temperature range includes the upper and lower temperature limits, which are set by the specific requirements for the temperature of the double-station solder ball welding machine in actual applications;

[0117] Step 5: Based on the welding data that needs to be corrected, take corresponding corrective measures for the welding data with deviations in each workstation;

[0118] For displacement data correction, the size and direction of the position deviation of the corresponding workstation are calculated based on the prediction results, and instructions are sent to the control system to control the movement of the X-axis, Y-axis, and Z-axis motors to move the welding head in the opposite direction by a corresponding distance, so that the welding point position returns to the correct position;

[0119] For pressure correction, the pressure output by the pressure applying device is adjusted in real time to ensure that the pressure during welding is always within the standard pressure range;

[0120] For temperature correction, the heating power or heating time is adjusted according to the prediction results. At the same time, the heating parameters are dynamically adjusted according to the real-time temperature changes in combination with the feedback control algorithm to ensure the stability of the welding temperature. Among them, the feedback control algorithm is an existing technology, and the specific process will not be described in detail.

[0121] For rotation angle correction, the angle compensation is performed by controlling the rotating motor and sending instructions to the motor controller to make the motor rotate the corresponding angle until the rotation angle is less than or equal to the standard rotation angle value.

[0122] Embodiment 2, as Figure 2 As shown, the present invention is a correction system for a double-station solder ball welding machine, comprising:

[0123] The data acquisition module collects welding data by deploying various sensors on the dual-station solder ball welding machine; the welding data includes the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle;

[0124] The preprocessing and feature extraction module preprocesses the collected welding data to obtain welding processing data; extracts features from the welding processing data through the feature layer fusion method to obtain the welding feature vector;

[0125] The correction prediction module builds a correction prediction model based on the support vector machine, predicts the welding feature vector obtained in real time, and obtains the prediction result;

[0126] Deviation judgment module, which performs deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected;

[0127] The correction module takes corresponding corrective measures for the welding data with deviations in each workstation based on the welding data that needs to be corrected.

[0128] In the several embodiments provided by the present invention, it should be understood that the disclosed system can be implemented in other ways. For example, the above-described embodiments of the invention are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0129] The modules described as separate accessories may or may not be physically separated, and the accessories shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0130] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0131] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. The correction method of the double-station solder ball welding machine is characterized in that: The following steps are involved: Step 1: Collect welding data by deploying various sensors on the double-station solder ball welding machine; the welding data includes solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle; Step 2: pre-process the collected welding data to obtain welding processing data; extract features from the welding processing data by a feature layer fusion method to obtain a welding feature vector; Step 3: construct a correction prediction model based on support vector machine, predict the welding feature vector obtained in real time, and obtain the prediction result; Step 4: Perform deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected; Step 5: Based on the welding data that needs to be corrected, take corresponding corrective measures for the welding data with deviations in each workstation.

2. The correction method of the double-station solder ball welding machine according to claim 1 is characterized in that: In step 1, the various sensors include: Using the laser tracker and industrial camera, the laser beam emitted by them is used as a reference to establish a three-dimensional reference coordinate system in the working space of the welding machine to obtain the position coordinates of the welding point; Displacement sensor, used to measure the displacement data of welding head movement; Force sensor, installed at the contact point between the welding head and the workpiece, used to measure the pressure data during welding; Temperature sensor, used to measure the temperature of the welding area and obtain temperature data; Motion sensor, used to measure the rotation angle of the motor; The welding data of the double-station solder ball welding machine is determined based on the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle.

3. The correction method of the double-station solder ball welding machine according to claim 2 is characterized in that: In step 2, the welding data is preprocessed, including: The 3σ criterion is adopted to regard the data that deviates from the mean by more than 3 times the standard deviation as outliers and eliminate or correct them to remove outliers in the welding data; a low-pass filter is used to filter the noise in the welding data; the normalization method is used to standardize the welding data with different dimensions and numerical ranges; the synchronization of welding data is achieved by aligning the timestamps when the data is collected and filling the missing values ​​with the interpolation method.

4. The correction method of the double-station solder ball welding machine according to claim 3 is characterized in that: Feature extraction of welding process data, including: Get displacement data through the formula Calculate the average speed of the welding head during welding and determine the displacement characteristics; where v is the average speed of the welding head, Δw is the displacement data of the welding head; Δt is the welding time; Obtain pressure data, calculate the maximum pressure, minimum pressure and average pressure during welding, and determine the pressure characteristics; Obtain temperature data, calculate the temperature change difference between adjacent time points and divide it by the time interval to obtain the temperature change rate and determine the temperature characteristics; Obtain the rotation angle of the motor and determine the rotation angle characteristics; Based on the displacement features, pressure features, temperature features, and rotation angle features, they are fused through weighted summation according to preset weights to obtain a welding feature vector.

5. The correction method of the double-station solder ball welding machine according to claim 1 is characterized in that: Construct a corrective prediction model, including: Based on historical welding data, obtain the historical welding feature vector X m and the corresponding label Y m ; Wherein, m = 1, 2, ..., M, M is the number of historical welding feature vectors; Label Y corresponding to the historical welding feature vector m Including normal and deviation, it can be expressed as: Y m =1, indicating that the historical welding feature vector is normal welding data; Y m =-1, indicating that the historical welding feature vector is welding data with deviation; Taking the historical welding feature vector as the input sample, the decision function in the correction prediction model is constructed based on the support vector machine; The decision function is expressed as: In the formula, f(x) is the decision function value; x is the sample to be tested; α m is the Lagrange multiplier, each input sample X m Corresponds to a Lagrange multiplier; K(X i ,x) is the kernel function, which is used to calculate the input sample X m The similarity with the sample x to be tested; b is the bias term, which is used to adjust the position of the decision boundary; Use gradient descent to solve for α m and b; by calculating the loss function about α m and the gradient of b, and then update α in the opposite direction of the gradient m and b, gradually reduce the value of the loss function in an iterative manner, and finally find the minimum value of the loss function; wherein the loss function can select the hinge loss function; According to the calculated decision function value, the corresponding label of the sample x to be tested is determined; if the decision function value f(x) ≥ 0, the prediction result is positive, that is, the label of the sample x to be tested is Y m =1; if the decision function value f(x) < 0, the prediction result is negative, that is, the label of the sample x to be tested is Y m = -1, and issue a warning prompt, indicating that there is a deviation; The accuracy of the model is obtained by calculating the ratio of the number of samples correctly predicted by the model to the total number of predicted samples; the accuracy is compared with the preset standard value. If the obtained accuracy is greater than or equal to the preset standard value, it means that the model prediction is accurate and the performance of the model meets the standard; otherwise, the model is optimized until the calculated accuracy reaches the preset standard value.

6. The correction method of the double-station solder ball welding machine according to claim 5 is characterized in that: The method to obtain the prediction results is as follows: According to the constructed correction prediction model, the welding feature vector obtained in real time is predicted, the welding feature vector is input into the correction prediction model, and the label corresponding to the welding feature vector is determined and obtained according to the calculated decision function value, which is recorded as the prediction result; The prediction results include normal data and abnormal data; the label is Y m = 1 is marked as normal data; the welding feature vector labeled as Y m = -1 is marked as abnormal data.

7. The correction method of the double-station solder ball welding machine according to claim 6 is characterized in that: In step 4, deviation detection and judgment are performed on abnormal data in the prediction results, including: Obtain welding data from abnormal data, including welding point position coordinates, displacement data, pressure data, temperature data, and rotation angle; For the displacement data, the difference between the weld point position coordinates and the displacement data of the weld head movement is calculated to obtain the displacement deviation; By comparing the displacement deviation WY with the preset standard displacement deviation value WYZ, if WY≤WYZ, it is determined that the position deviation meets the standard; otherwise, it is determined that the displacement data has deviation and needs to be corrected; For pressure and temperature deviations, the pressure data is compared with the preset standard pressure range. If the pressure data is within the standard pressure range, it is determined that there is no deviation in the pressure data; otherwise, it is determined that there is a deviation in the pressure data and correction is required; By comparing the temperature data with the preset standard temperature range, if the temperature data is within the standard temperature range, it is determined that the temperature data has no deviation; otherwise, it is determined that the temperature data has a deviation and needs to be corrected; For the rotation angle, by comparing the rotation angle with a preset standard rotation angle value, if the rotation angle ≤ the standard rotation angle value, it means that the rotation angle meets the standard; otherwise, it is determined that there is a deviation in the rotation angle and correction is required.

8. The correction method of the double-station solder ball welding machine according to claim 7, characterized in that: The corrective measures described include: For displacement data correction, the size and direction of the position deviation of the corresponding workstation are calculated based on the prediction results, and instructions are sent to the control system to control the movement of the X-axis, Y-axis, and Z-axis motors to move the welding head in the opposite direction by a corresponding distance, so that the welding point position returns to the correct position; For pressure correction, the pressure output by the pressure applying device is adjusted in real time to ensure that the pressure during welding is always within the standard pressure range; For temperature correction, the heating power or heating time is adjusted according to the prediction results. At the same time, the heating parameters are dynamically adjusted according to the real-time temperature changes in combination with the feedback control algorithm to ensure the stability of the welding temperature. Among them, the feedback control algorithm is an existing technology, and the specific process will not be described in detail. For rotation angle correction, the angle compensation is performed by controlling the rotating motor and sending instructions to the motor controller to make the motor rotate the corresponding angle until the rotation angle is less than or equal to the standard rotation angle value.

9. A correction system for a dual-station solder ball welding machine, implemented based on the correction method for a dual-station solder ball welding machine according to any one of claims 1 to 8, characterized in that: include: The data acquisition module collects welding data by deploying various sensors on the dual-station solder ball welding machine; the welding data includes the solder point position coordinates, displacement data, pressure data, temperature data, and rotation angle; The preprocessing and feature extraction module preprocesses the collected welding data to obtain welding processing data; extracts features from the welding processing data through the feature layer fusion method to obtain the welding feature vector; The correction prediction module builds a correction prediction model based on the support vector machine, predicts the welding feature vector obtained in real time, and obtains the prediction result; Deviation judgment module, which performs deviation detection and judgment on abnormal data in the prediction results to obtain welding data that needs to be corrected; The correction module takes corresponding corrective measures for the welding data with deviations in each workstation based on the welding data that needs to be corrected.

Images