Multi-robot cooperative automobile side wall production line parameter control optimization system

Abstract

The application discloses a multi-robot cooperative automobile side wall production line parameter control optimization system and relates to the technical field of side wall production parameter control. The system comprises the following steps: collecting welding current and welding times of a robot each time spot welding is performed, obtaining spot welding current data, collecting corresponding nugget diameters after spot welding is completed, and obtaining spot welding diameter data; performing merging, dividing and data elimination processing to obtain spot welding effective sample data; performing current optimization analysis to obtain the corresponding relationship between welding current and welding times under ideal nugget diameters and obtain current times reference information; and controlling and optimizing the welding current of the robot. The application is used to solve the problem that the existing side wall production parameter control technology cannot establish the relationship between welding current and welding times under ideal nugget diameters according to historical data of welding current and nugget diameters and cannot control and optimize welding current in a real time manner when controlling and optimizing the welding current of robot resistance spot welding.

Description

Technical Field

[0001] This invention relates to the field of side panel production parameter control technology, specifically a multi-robot collaborative automotive side panel production line parameter control optimization system. Background Technology

[0002] Side panel production parameter control technology is a core process control technology in the field of intelligent manufacturing of automotive bodies for the entire mass production process of side panel components. With the dimensional accuracy, structural strength, surface quality of side panel products and the stability and efficiency of the production process as the core objectives, it integrates automotive manufacturing process mechanisms, measurement and control technology, data analysis and intelligent optimization algorithms to accurately monitor, identify anomalies, dynamically control and iteratively optimize key production parameters of side panel components from stamping and forming, spot welding assembly to gluing and joining.

[0003] Existing side panel production parameter control technologies often employ a phased current compensation method when optimizing the welding current for robotic resistance spot welding. This is because the electrode tip gradually wears down with each welding cycle, leading to a gradual decrease in the weld nugget diameter and eventually defects such as incomplete welds and weld failures. Therefore, before electrode re-grinding or replacement, the welding current must be gradually increased to compensate for the losses caused by contact electrode wear. However, phased current compensation is performed at fixed welding intervals, such as once every 100 welds. The current adjustment is not synchronized with the actual wear state of the electrode; within a fixed interval, the electrode continues to wear down, but the current remains constant. This leads to a gradual decrease in the weld nugget diameter within the weld range, from slightly larger to near the acceptable lower limit, and even some later weld nugget diameters being slightly lower than the standard value, resulting in poor welding consistency. As a load-bearing component of the vehicle body, the side panel has high requirements for the consistency of the weld nugget diameter at its key weld points. The gradual decrease in the weld nugget diameter within the weld range will lead to large dispersion in weld strength, which will reduce the rigidity and collision safety of the overall side panel structure. Therefore, existing side panel production parameter control technologies cannot establish the relationship between welding current and welding number under the ideal weld nugget diameter based on historical data of welding current and weld nugget diameter when controlling and optimizing the welding current of robotic resistance spot welding, and cannot control and optimize the welding current in real time. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art. It obtains spot welding current data by collecting the welding current and welding count during each spot welding operation by the robot, and collects the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data. This data is then merged, divided, and processed to remove outliers, resulting in valid spot welding sample data. Current optimization analysis is performed to obtain the correspondence between welding current and welding count under the ideal weld nugget diameter, providing current-count reference information. The robot's welding current is then controlled and optimized. This addresses the problem that existing side panel production parameter control technologies, when controlling and optimizing the welding current for robot resistance spot welding, cannot establish the relationship between welding current and welding count under the ideal weld nugget diameter based on historical data of welding current and weld nugget diameter, and cannot perform real-time control and optimization of the welding current.

[0005] To achieve the above objectives, this application provides a multi-robot collaborative parameter control and optimization system for an automotive side panel production line, including a parameter acquisition module, a data cleaning module, an optimization analysis module, and a control optimization module;

[0006] The parameter acquisition module includes a first acquisition unit and a second acquisition unit. The first acquisition unit is used to acquire the welding current and welding number during each spot welding by the robot to obtain spot welding current data. The second acquisition unit is used to acquire the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data.

[0007] The data cleaning module is used to merge and divide the spot welding current data and spot welding diameter data, and perform data de-identification processing to obtain valid spot welding sample data.

[0008] The optimization analysis module performs current optimization analysis based on effective sample data of spot welding, obtains the correspondence between welding current and welding number under the ideal weld nugget diameter, and obtains current number reference information;

[0009] The control optimization module optimizes the welding current of the robot based on the current frequency reference information and the number of welding operations performed by the robot.

[0010] Furthermore, the first acquisition unit is configured with a first acquisition strategy, which includes:

[0011] In the production of automobile side panels, any resistance spot welding point is designated as the first spot welding point; the robots of the same model that weld the first spot welding point are designated as spot welding robot 1 to spot welding robot n, where n is the total number of spot welding robots.

[0012] The magnitude of the current applied to the welding area through the electrode when the spot welding robot welds the first weld point is recorded as the welding current, and the magnitude of the welding current set by the spot welding robot is recorded as the preset current; the number of times the spot welding robot welds the first weld point from the time the electrode is ground or replaced is recorded as the welding count.

[0013] For spot welding robot 1, during each spot welding, the corresponding welding current and welding number are collected and recorded as the spot welding parameter information of spot welding robot 1. The spot welding parameter information of all spot welding robots is collected repeatedly and merged to obtain spot welding current data. The preset current corresponding to each welding number is obtained, and the preset current when the welding number is 0 is recorded as WI.

[0014] Furthermore, the second acquisition unit is configured with a second acquisition strategy, which includes:

[0015] For spot welding robot 1, after each spot welding is completed, the diameter of the weld nugget is collected and recorded as the spot welding diameter information of spot welding robot 1; the spot welding parameter information of all spot welding robots is collected repeatedly to obtain the spot welding diameter data.

[0016] Obtain the ideal weld nugget diameter of the first weld joint, denoted as the ideal weld nugget diameter AL.

[0017] Furthermore, the data cleaning module is configured with data cleaning strategies, which include:

[0018] The range of welding times is obtained based on the spot welding current data and is sequentially recorded as welding times 1 to welding times m, where m is the maximum value of welding times in the welding current data; for welding times 1, based on the spot welding current data and spot welding diameter data, the welding current and weld nugget diameter corresponding to all welding times 1 are recorded as the original current diameter set;

[0019] The welding current and weld nugget diameter collected in the same time in the original current-diameter set are combined into a data pair, denoted as welding data pair (AI, AD). After completion, the first welding dataset is obtained, where AI represents welding current and AD represents weld nugget diameter.

[0020] Calculate the median ME of the welding current in the first welding dataset and calculate the absolute deviation of the median MA. Welding currents that are not located in [ME-k1×1.4826×MA, ME+k1×1.4826×MA] are recorded as outliers and marked as abnormal currents, where k1 is a set scaling factor. Repeatedly obtain the outliers corresponding to the weld nugget diameter in the first welding dataset and mark them as abnormal diameters.

[0021] For any welding data pair in the first welding dataset, if it contains abnormal current or abnormal diameter, it is marked as an abnormal data pair and removed. After completion, the second welding dataset is obtained.

[0022] Furthermore, data cleaning strategies also include:

[0023] Calculate the mean values ​​of welding current and weld nugget diameter in the second welding dataset, denoted as IU and DU respectively, and form a column vector, denoted as mean vector U0=[IU, DU]. T ;Any welding data pair in the second welding dataset is denoted as the first data pair (BI, BD);

[0024] Based on the welding current and weld nugget diameter in the second welding dataset, calculate the sample variance of the welding current and the sample variance of the weld nugget diameter, denoted as CII and CDD respectively, and calculate the sample covariance of the welding current and the weld nugget diameter, denoted as CID.

[0025] The covariance matrix of welding current and weld nugget diameter is constructed and denoted as the first covariance matrix, which is as follows: , where M represents the first covariance matrix.

[0026] Furthermore, data cleaning strategies also include:

[0027] Convert the first data pair (BI, BD) into a column vector, denoted as [BI, BD]. T Then calculate [BI, BD] based on M. T Calculate the Mahalanobis distance BG to the mean vector U0; repeat the calculation of the Mahalanobis distances for all welding data pairs in the second welding dataset to obtain the Mahalanobis distance set.

[0028] Outliers in the Mahalanobis distance set are filtered according to the 3σ principle and recorded as outlier distances. Welding data pairs corresponding to outlier distances are marked as outlier data pairs and removed. After completion, the effective welding dataset with welding number 1 is obtained.

[0029] Repeatedly obtain the valid welding dataset corresponding to all welding times to obtain valid spot welding sample data.

[0030] Furthermore, the optimization analysis module is configured with optimization analysis strategies, which include:

[0031] Based on the valid welding dataset of welding number 1, obtain the sum of the corresponding Mahalanobis distances for all welding data pairs, denoted as the weighted basis AQ; for any welding current in the valid welding dataset, denoted as EI, and the corresponding Mahalanobis distance denoted as EG, calculate EI×EG / AQ, denoted as the weighted current of EI;

[0032] Repeatedly calculate the weighted current of all welding currents in the effective welding dataset, and sum them to obtain the representative value of the welding current, which is denoted as the representative current of welding number 1. Repeatedly obtain the representative value of the welding current to obtain the representative diameter of welding number 1.

[0033] Repeatedly obtain the representative current and representative diameter for all welding cycles, and form a data pair with each welding cycle and the corresponding representative current and representative diameter, denoted as sample point YN=(RN, NI, ND), where RN, NI and ND represent the welding cycle, representative current and representative diameter, respectively.

[0034] Furthermore, the optimization analysis strategy also includes:

[0035] Based on the preset current corresponding to the number of welding, the sample points with the same preset current are grouped into one category and called the sample set. They are then sequentially named from sample set 1 to sample set v in descending order of the number of welding, where v is the total number of sample sets.

[0036] For sample set 1, the range of welding times in sample set 1 is denoted as the number interval [AN1, BN1]; and the welding times and representative diameter in the sample set are linearly fitted to obtain the ND relationship model of sample set 1, denoted as ND1=ek1×RN1+b1, where ek1 and b1 represent the slope and intercept, respectively.

[0037] Arrange the sample points in sample set 1 in ascending order of welding number, and denote any sample point as YN1j=(RN1j, NI1j, ND1j), where j represents the position number; obtain the representative current M1I and representative diameter M1D of the sample point at the end position, and denote the end current diameter 1. Repeat the process of obtaining the representative current Q2I and representative diameter Q2D of the sample point at the beginning position in sample set 2, and denote it as the beginning current diameter 2.

[0038] Furthermore, the optimization analysis strategy also includes:

[0039] Calculate (Q2I-M1I) / (Q2D-M1D), denoted as the slope IK1 of sample set 1; substitute RN1j into the ND relationship model of sample set 1, and denot the result as PD1j; calculate (AL-PD1j)×IK1, denoted as the target current corresponding to the welding number RN1j;

[0040] Repeatedly obtain the target current corresponding to all welding times in sample set 1, and perform nonlinear fitting on the welding times and target current to obtain the relationship curve between welding times and target current corresponding to the interval [AN1, BN1], which is denoted as the target current curve;

[0041] Repeat the number of times intervals of all sample sets, and obtain the target current curves for all number intervals, which are recorded as the current number reference information.

[0042] Furthermore, the control optimization module is configured with control optimization strategies, which include:

[0043] For any spot welding robot, after the previous spot welding is completed, the corresponding welding count is obtained and recorded as the current welding count;

[0044] Based on the current welding cycle range, the corresponding target current is obtained using the corresponding target current curve, recorded as the next target current, and the welding current for the next spot weld is adjusted to the next target current; the welding current of all spot welding robots is controlled repeatedly.

[0045] The beneficial effects of this invention are as follows: This invention obtains spot welding current data by collecting the welding current and welding count during each spot welding operation of the robot, and obtains spot welding diameter data by collecting the corresponding weld nugget diameter after spot welding is completed; the spot welding current data and spot welding diameter data are merged and divided, and data anomaly removal is performed to obtain effective spot welding sample data; based on the effective spot welding sample data, current optimization analysis is performed to obtain the correspondence between welding current and welding count under the ideal weld nugget diameter, and obtain current count reference information; based on the current count reference information and the number of welding operations of the robot, the welding current of the robot is controlled and optimized; when controlling and optimizing the welding current of robot resistance spot welding, the relationship between welding current and welding count under the ideal weld nugget diameter can be established based on the historical data of welding current and weld nugget diameter, and the welding current can be controlled and optimized in real time to improve the consistency of welding;

[0046] This invention calculates Mahalanobis distance using the covariance matrix, enabling the filtering of data that are abnormal under the correlation of two variables, rather than relying solely on their respective thresholds. This allows for a more comprehensive elimination of outliers for welding current and weld nugget diameter, which exhibit statistical dependence. Grouping by welding number and then independently removing outliers allows for the use of different statistical characteristics to judge different wear stages, avoiding filtering errors caused by mixing data from different wear stages. Weighting the current of each sample using its Mahalanobis distance reflects the representativeness of the sample distribution characteristics, thus better reflecting typical welding conditions. Using representative currents and diameters from different welding numbers as sample points quantifies the relationship between electrode usage and welding current under ideal weld nugget diameters, allowing for dynamic compensation of welding current with wear and improving the consistency of spot welding. Fitting separately within each welding number interval allows for a more precise fit of the welding current relationship at different wear stages, improving the reliability of control optimization. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the system of the present invention;

[0048] Figure 2 This is a flowchart illustrating the steps of the method of the present invention;

[0049] Figure 3 This is a flowchart illustrating the process of obtaining the effective welding dataset according to the present invention.

[0050] Figure 4 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1, please refer to Figure 1 As shown, this application provides a multi-robot collaborative parameter control and optimization system for an automotive side panel production line, including a parameter acquisition module, a data cleaning module, an optimization analysis module, and a control optimization module;

[0053] The parameter acquisition module includes a first acquisition unit and a second acquisition unit. The first acquisition unit is used to acquire the welding current and welding number during each spot welding by the robot to obtain spot welding current data. The second acquisition unit is used to acquire the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data.

[0054] The first acquisition unit is configured with a first acquisition strategy, which includes: recording any resistance spot welding point in the production of the automobile side panel as the first welding point; and sequentially recording the robots of the same model that weld the first welding point as spot welding robot 1 to spot welding robot n, where n is the total number of spot welding robots. In the production of the automobile side panel, there are many resistance spot welding positions, and the specifications of the welding points at different positions are often inconsistent, and their relationship with the welding current is also different. Therefore, the data cannot be used interchangeably. If the specifications of the welding points at different positions are consistent, then the data can be used interchangeably.

[0055] The current applied to the welding area by the electrode when the spot welding robot welds the first weld point is recorded as the welding current. The welding current set by the spot welding robot is recorded as the preset current, which is the sum of the set reference welding current and the compensation current. The number of times the spot welding robot welds the first weld point from the time the electrode is ground or replaced is recorded as the welding count. The preset current is the command value issued by the controller, and the actual welding current is the real response during the spot welding process.

[0056] For spot welding robot 1, during each spot welding operation, the corresponding welding current and welding count are collected. When the welding count is 0, no data is collected. This data is recorded as the spot welding parameter information of spot welding robot 1. The spot welding parameter information of all spot welding robots is collected repeatedly and merged to obtain the spot welding current data. The preset current corresponding to each welding count is obtained. The preset current when the welding count is 0 is denoted as WI. The number of welding counts collected refers to the number of electric weldings that have been completed, excluding the current spot welding. For example, when welding for the 5th time, the welding count is 4. WI represents the preset current for current compensation. It also represents that when the motor is not worn, the weld nugget diameter is the ideal weld nugget diameter when the welding current is WI.

[0057] The collected welding current is the actual welding current. The formation of the weld nugget diameter depends on the Joule heat generated by the actual current passing through the workpiece in the welding circuit. The preset current is only the theoretical output command value of the welding machine. The actual welding current will have a small but not negligible deviation from the preset current due to objective factors. If the preset current is used directly, the subsequent mapping relationship will be disconnected from the actual welding process, which will lead to unreliable control optimization.

[0058] The number of welding cycles represents the electrode wear state. The number of welding cycles is a representative variable of the changes in electrode geometry and contact state over time, which directly affects contact resistance, heating distribution, and thus the formation of the weld nugget, i.e., the weld nugget diameter.

[0059] The second acquisition unit is configured with a second acquisition strategy, which includes: for spot welding robot 1, after each spot welding is completed, the diameter of the weld nugget is acquired and recorded as the spot welding diameter information of spot welding robot 1; the spot welding parameter information of all spot welding robots is repeatedly acquired to obtain the spot welding diameter data; the weld nugget diameter is the most intuitive indicator of weld quality and directly reflects whether the welding is appropriate.

[0060] Obtain the ideal weld nugget diameter of the first weld joint, denoted as the ideal weld nugget diameter AL;

[0061] In practice, the electrode tip wears down gradually with each welding cycle due to high-temperature thermal stress, mechanical pressure, and metal adhesion. This causes the weld nugget diameter to gradually decrease. Since the electrode cannot be replaced immediately after welding, the welding current must be gradually increased to compensate for the Joule heat loss caused by the increased contact resistance before the electrode is repaired or replaced, ensuring that the weld nugget quality remains within the acceptable range. Because of the use of staged current compensation, the collected welding current and weld nugget diameter will also exhibit the characteristics of staged jumps.

[0062] The data cleaning module is used to merge and divide spot welding current data and spot welding diameter data, and to perform data de-identification processing to obtain valid spot welding sample data.

[0063] The data cleaning module is configured with a data cleaning strategy, which includes: obtaining the range of welding times based on spot welding current data, and recording them sequentially as welding times 1 to welding times m, where m is the maximum value of welding times in the welding current data, i.e., the maximum number of welding times for the electrode; for welding times 1, based on spot welding current data and spot welding diameter data, recording the welding current and weld nugget diameter corresponding to all welding times 1 as the original current-diameter set; i.e., the data corresponding to welding times 1;

[0064] Please see Figure 3 As shown, the welding current and weld nugget diameter collected in the same time in the original current-diameter set are combined into a data pair, denoted as welding data pair (AI, AD). After completion, the first welding dataset is obtained, where AI represents welding current and AD represents weld nugget diameter.

[0065] Calculate the median ME of the welding current in the first welding dataset and calculate the absolute deviation of the median MA. Welding currents that are not located in [ME-k1×1.4826×MA, ME+k1×1.4826×MA] are recorded as abnormal values ​​corresponding to the welding current and marked as abnormal currents, where k1 is a set proportional coefficient. Repeatedly obtain the abnormal values ​​corresponding to the weld nugget diameter in the first welding dataset and mark them as abnormal diameters. In this embodiment, k1=3, which can be set flexibly and is generally [2, 3].

[0066] The mean is easily affected by extreme values. The data center is determined by the median, and the robustness of the data fluctuation level is measured by the absolute deviation of the median. Finally, the outlier threshold of a single indicator is defined to identify any extreme abnormal sample of AI or AD that jumps out of its normal fluctuation range, thereby reducing the interference of extreme values ​​for subsequent two-dimensional analysis.

[0067] For example, the first welding dataset is {(14.0, 6.0), (14.1, 6.1), (13.9, 5.9), (14.2, 6.2), (13.8, 5.8), (14.0, 6.0), (16.0, 5.7), (13.9, 5.9), (14.1, 6.1), (14.0, 6.0), (13.8, 5.8), (14.2, 6.2), (14.0, 4.5), (13.9, 6.0), (14.1, 6.1), (14.0, 5.9), (13.8, 5.8), (14.2, 6.2), (14.0, 6.0), (13.9, 5.9), (14.1, 6.1), (14.2, 5.0)} The values ​​(13.8, 6.0), (14.0, 6.0), (13.9, 5.9), (14.1, 6.1), (14.0, 6.0), (13.9, 5.8), (14.2, 6.3), (14.0, 6.0)} are in kA and mm. The median welding current ME = 14.0 is calculated, and the absolute deviation of the median MA = 0.1 kA. Then [ME - k1 × 1.4826 × MA, ME + k1 × 1.4826 × MA] is [13.5552, 14.4448]. The welding current of (16.0, 5.7) is 16.0, which is not in this range, so it is an abnormal current. The abnormal diameter is screened again, and the weld nugget diameter of (14.0, 4.5) is an abnormal diameter.

[0068] For any welding data pair in the first welding dataset, if it contains abnormal current or abnormal diameter, it is marked as an abnormal data pair and removed. After completion, the second welding dataset is obtained.

[0069] Calculate the mean values ​​of welding current and weld nugget diameter in the second welding dataset, denoted as IU and DU respectively, and form a column vector, denoted as mean vector U0=[IU, DU]. T T represents vector transpose, and any welding data pair in the second welding dataset is denoted as the first data pair (BI, BD).

[0070] Based on the welding current and weld nugget diameter in the second welding dataset, the sample variance of the welding current and the sample variance of the weld nugget diameter are calculated and denoted as CII and CDD, respectively. The sample covariance of the welding current and the weld nugget diameter is also calculated and denoted as CID. CII reflects the fluctuation range of the welding current, CDD reflects the fluctuation range of the weld nugget diameter, and CID reflects the degree of linear correlation between the two.

[0071] The covariance matrix of welding current and weld nugget diameter is constructed and denoted as the first covariance matrix, which is as follows: , where M represents the first covariance matrix.

[0072] Convert the first data pair (BI, BD) into a column vector, denoted as [BI, BD]. T Then calculate [BI, BD] based on M. T Calculate the Mahalanobis distance BG to the mean vector U0; repeat the calculation of the Mahalanobis distances for all welding data pairs in the second welding dataset to obtain the Mahalanobis distance set.

[0073] Outliers in the Mahalanobis distance set are filtered according to the 3σ principle and recorded as outlier distances. Welding data pairs corresponding to outlier distances are marked as outlier data pairs and removed. After completion, the effective welding dataset with welding number 1 is obtained.

[0074] Repeatedly obtain the valid welding dataset corresponding to all welding times to obtain valid spot welding sample data;

[0075] In the specific implementation process, Mahalanobis distance is used to measure the degree of deviation of each sample point in the two-dimensional distribution. This can identify welding data pairs where the individual indicators are normal, but the pairing relationship deviates from the overall distribution. This can make up for the defect of missing abnormal pairings in the single indicator screening.

[0076] The optimization analysis module performs current optimization analysis based on effective sample data of spot welding, obtains the correspondence between welding current and welding number under the ideal weld nugget diameter, and obtains current number reference information;

[0077] Based on the valid welding dataset with 1 welding cycle, obtain the sum of the Mahalanobis distances corresponding to all welding data pairs, denoted as the basic weight AQ; for any welding current in the valid welding dataset, denoted as EI, and the corresponding Mahalanobis distance denoted as EG, calculate EI×EG / AQ, denoted as the weighted current of EI; the Mahalanobis distance is the distribution distance from the sample to the two-dimensional center, i.e., the mean vector. The smaller the Mahalanobis distance, the closer the sample is to the two-dimensional center, the higher the fit with the overall working condition, the stronger the representativeness, and the greater the weight assigned.

[0078] Repeatedly calculate the weighted current of all welding currents in the effective welding dataset, and sum them to obtain the representative value of the welding current, which is denoted as the representative current of welding number 1. Repeatedly obtain the representative value of the welding current to obtain the representative diameter of welding number 1.

[0079] Repeatedly obtain the representative current and representative diameter for all welding cycles, and form a data pair with each welding cycle and the corresponding representative current and representative diameter, denoted as sample point YN=(RN, NI, ND), where RN, NI and ND represent the welding cycle, representative current and representative diameter, respectively.

[0080] Based on the preset current corresponding to the number of welding operations, sample points with the same preset current are grouped into one category and denoted as sample set. The original data is collected on the basis of staged compensation of welding arc, and the preset current in the unified compensation range is the same. They are then denoted as sample set 1 to sample set v in descending order of the number of welding operations, where v is the total number of sample sets.

[0081] For sample set 1, the range of welding times in sample set 1 is denoted as the number interval [AN1, BN1]. A linear fit is performed on the welding times and representative diameter in the sample set to obtain the ND relationship model of sample set 1, denoted as ND1 = ek1 × RN1 + b1, where ND1 and RN1 represent the welding times and representative diameter of sample set 1, respectively. Here, ek1 and b1 represent the slope and intercept, respectively. Electrode wear causes the weld nugget diameter to decrease approximately linearly. For example, ek1 = -0.002 mm / time, that is, the weld nugget diameter decreases by 0.002 mm for each welding.

[0082] Arrange the sample points in sample set 1 in ascending order of welding number, and denote any sample point as YN1j=(RN1j, NI1j, ND1j), where j represents the position number; obtain the representative current M1I and representative diameter M1D of the sample point at the end position, and denote the end current diameter 1. Repeat the process of obtaining the representative current Q2I and representative diameter Q2D of the sample point at the beginning position in sample set 2, and denote it as the beginning current diameter 2.

[0083] Calculate (Q2I-M1I) / (Q2D-M1D), denoted as the slope IK1 of sample set 1; based on the changes in welding current and representative diameter in adjacent compensation intervals, calculate the welding current required to increase the weld nugget by 1 mm. For intermediate sample sets, such as sample set 2, which is adjacent to sample set 1 and sample set 3, two slopes can be calculated. The average value is then taken as the final slope.

[0084] Substitute RN1j into the ND relationship model of sample set 1 and denote the result as PD1j; calculate (AL-PD1j)×IK1, and denote it as the target current corresponding to the welding number RN1j;

[0085] For example, in the ND relationship model ND1=-0.002×RN1+6.02, AL=6.0mm, IK1=1kA / mm, WI=14kA, then when RN1j=50, PD1j=5.9mm, then (AL-PD1j)×IK1=0.1k4, and the corresponding target current is 14.1kA;

[0086] Repeatedly obtain the target current corresponding to all welding times in sample set 1, and perform nonlinear fitting on the welding times and target current to obtain the relationship curve between welding times and target current corresponding to the interval [AN1, BN1], which is denoted as the target current curve;

[0087] Repeat the number of times intervals of all sample sets, and obtain the target current curves for all number intervals, which are recorded as current number reference information;

[0088] In practice, all welding cycles and target currents can be fitted into a relationship curve, but the slope of the curves in adjacent intervals must be continuous at the connection point to avoid sudden changes in current.

[0089] The control optimization module optimizes the welding current of the robot based on the current frequency reference information and the number of welding operations performed by the robot.

[0090] For any spot welding robot, after the previous spot welding is completed, the corresponding welding count is obtained and recorded as the current welding count;

[0091] Based on the current welding cycle range, the corresponding target current is obtained using the corresponding target current curve, recorded as the next target current, and the welding current for the next spot weld is adjusted to the next target current; the welding current of all spot welding robots is controlled repeatedly.

[0092] In the specific implementation process, the target current curve provides real-time dynamic compensation for the motor wear of the spot welding robot, which can maintain a stable weld nugget diameter during the generation of the car side panel and improve the consistency of the weld.

[0093] Example 2, please refer to Figure 2 As shown, this application provides a method for parameter control optimization of a multi-robot collaborative automotive side panel production line, including the following steps:

[0094] Step S1 involves collecting the welding current and the number of welds during each spot welding operation by the robot to obtain spot welding current data, and collecting the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data. Step S1 includes the following sub-steps:

[0095] Step S101: Record any resistance spot welding point in the production of the automobile side panel as the first spot welding point; record the robots of the same model that weld the first spot welding point as spot welding robot 1 to spot welding robot n, where n is the total number of spot welding robots.

[0096] Step S102: Record the magnitude of the current applied to the welding area by the electrode when the spot welding robot welds the first weld point as the welding current, and record the magnitude of the welding current set by the spot welding robot as the preset current; record the number of times the spot welding robot has welded the first weld point since grinding or replacing the electrode as the welding count.

[0097] Step S103: For spot welding robot 1, during each spot welding, the corresponding welding current and welding number are collected and recorded as the spot welding parameter information of spot welding robot 1. The spot welding parameter information of all spot welding robots is collected repeatedly and merged to obtain spot welding current data. The preset current corresponding to each welding number is obtained and the preset current when the welding number is 0 is recorded as WI.

[0098] Step S104: For spot welding robot 1, after each spot welding is completed, the diameter of the weld nugget is collected and recorded as the spot welding diameter information of spot welding robot 1; the spot welding parameter information of all spot welding robots is collected repeatedly to obtain the spot welding diameter data.

[0099] Step S105: Obtain the weld nugget diameter under ideal conditions for the first weld point, denoted as the ideal weld nugget diameter AL.

[0100] Step S2 involves merging and dividing the spot welding current data and spot welding diameter data, and performing data deduplication to obtain valid spot welding sample data. Step S2 includes the following sub-steps:

[0101] Step S201: Obtain the range of welding times based on spot welding current data, and record them sequentially as welding times 1 to welding times m, where m is the maximum value of welding times in the welding current data; for welding times 1, based on spot welding current data and spot welding diameter data, record the welding current and weld nugget diameter corresponding to all welding times 1 as the original current diameter set;

[0102] Step S202: Combine the welding current and weld nugget diameter collected in the same time in the original current diameter set into a data pair, denoted as welding data pair (AI, AD). After completion, the first welding dataset is obtained, where AI represents welding current and AD represents weld nugget diameter.

[0103] Step S203: Calculate the median ME of the welding current in the first welding dataset and calculate the absolute deviation of the median MA. Welding currents that are not located in [ME-k1×1.4826×MA, ME+k1×1.4826×MA] are recorded as outliers and marked as abnormal currents, where k1 is a set scaling factor. Repeatedly obtain the outliers corresponding to the weld nugget diameter in the first welding dataset and mark them as abnormal diameters.

[0104] Step S204: For any welding data pair in the first welding dataset, if it contains abnormal current or abnormal diameter, it is marked as an abnormal data pair and removed. After completion, the second welding dataset is obtained.

[0105] Step S205: Calculate the mean values ​​of welding current and weld nugget diameter in the second welding dataset, denoted as IU and DU respectively, and form a column vector, denoted as mean vector U0=[IU, DU]. T ;Any welding data pair in the second welding dataset is denoted as the first data pair (BI, BD);

[0106] Step S206: Based on the centralized welding current and weld nugget diameter in the second welding dataset, calculate the sample variance of the welding current and the sample variance of the weld nugget diameter, denoted as CII and CDD respectively, and calculate the sample covariance of the welding current and the weld nugget diameter, denoted as CID.

[0107] Step S207, and construct the covariance matrix of welding current and weld nugget diameter, denoted as the first covariance matrix, which is as follows: , where M represents the first covariance matrix.

[0108] Step S208: Convert the first data pair (BI, BD) into a column vector, denoted as [BI, BD]. T Then calculate [BI, BD] based on M. T Calculate the Mahalanobis distance BG to the mean vector U0; repeat the calculation of the Mahalanobis distances for all welding data pairs in the second welding dataset to obtain the Mahalanobis distance set.

[0109] Step S209: Filter outliers in the Mahalanobis distance set according to the 3σ principle, record them as outlier distances, mark the welding data pairs corresponding to the outlier distances as outlier data pairs, and remove them. After completion, the effective welding dataset with welding number 1 is obtained.

[0110] Step S210: Repeatedly acquire the valid welding dataset corresponding to all welding times to obtain valid spot welding sample data.

[0111] Step S3 involves performing current optimization analysis based on valid spot welding sample data to obtain the correspondence between welding current and welding count under the ideal weld nugget diameter, thus obtaining current count reference information. Step S3 includes the following sub-steps:

[0112] Step S301: Based on the valid welding dataset of welding number 1, obtain the sum of the corresponding Mahalanobis distances of all welding data pairs, denoted as the weighted basis AQ; for any welding current in the valid welding dataset, denoted as EI, and the corresponding Mahalanobis distance denoted as EG, calculate EI×EG / AQ, denoted as the weighted current of EI;

[0113] Step S302: Repeatedly calculate the weighted current of all welding currents in the effective welding dataset, and sum them to obtain the representative value of the welding current, which is recorded as the representative current of welding number 1. Repeatedly obtain the representative value of the welding current to obtain the representative diameter of welding number 1.

[0114] Step S303: Repeatedly obtain the representative current and representative diameter for all welding cycles, and form a data pair with each welding cycle and the corresponding representative current and representative diameter, denoted as sample point YN=(RN, NI, ND), where RN, NI and ND represent the welding cycle, representative current and representative diameter, respectively.

[0115] Step S304: Based on the preset current corresponding to the number of welding, the sample points with the same preset current are grouped into one category and called the sample set. They are then sequentially called sample set 1 to sample set v in descending order of the number of welding, where v is the total number of sample sets.

[0116] Step S305: For sample set 1, the range of welding times in sample set 1 is denoted as the number interval [AN1, BN1]; and the welding times and representative diameter in the sample set are linearly fitted to obtain the ND relationship model of sample set 1, denoted as ND1=ek1×RN1+b1, where ek1 and b1 represent the slope and intercept, respectively.

[0117] Step S306: Arrange the sample points in sample set 1 in ascending order according to the number of welding, and denot any sample point as YN1j=(RN1j, NI1j, ND1j), where j represents the position number; and obtain the representative current M1I and representative diameter M1D of the sample point at the end position, which is called the end current diameter 1. Repeat the process of obtaining the representative current Q2I and representative diameter Q2D of the sample point at the beginning position in sample set 2, which is called the beginning current diameter 2.

[0118] Step S307: Calculate (Q2I-M1I) / (Q2D-M1D), denoted as the slope IK1 of the sample set 1; substitute RN1j into the ND relationship model of the sample set 1, and denot the result as PD1j; calculate (AL-PD1j)×IK1, denoted as the target current corresponding to the welding number RN1j.

[0119] Step S308: Repeatedly obtain the target current corresponding to all welding times in sample set 1, and perform nonlinear fitting on the welding times and target current to obtain the relationship curve between welding times and target current corresponding to the number interval [AN1, BN1], which is denoted as the target current curve;

[0120] Step S309: Repeat the number of times the entire sample set is obtained, and obtain the target current curve for all number of times the interval is recorded as the current number reference information.

[0121] Step S4: Based on the current frequency reference information and the number of welding operations performed by the robot, optimize the control of the robot's welding current; Step S4 includes the following sub-steps:

[0122] Step S401: For any spot welding robot, after the previous spot welding is completed, obtain the corresponding welding count and record it as the current welding count;

[0123] Step S402: Based on the current welding cycle range, obtain the corresponding target current using the corresponding target current curve, record it as the next target current, and adjust the welding current for the next spot welding to the next target current; repeat the control of welding current for all spot welding robots.

[0124] Example 3, please refer to Figure 4 As shown, Figure 4 A schematic diagram of an electronic device is provided, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call the instructions in the memory. When the computer-readable instructions are executed by the processor, steps such as those in the parameter control optimization method for a multi-robot collaborative automotive side panel production line are performed to achieve the following functions: collecting the welding current and welding count for each spot welding by the robot to obtain spot welding current data; collecting the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data; merging and dividing the spot welding current data and spot welding diameter data, and performing data deduplication to obtain valid spot welding sample data; performing current optimization analysis based on the valid spot welding sample data to obtain the correspondence between welding current and welding count under the ideal weld nugget diameter, obtaining current count reference information; and controlling and optimizing the robot's welding current based on the current count reference information and the robot's welding count.

[0125] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] Example 4: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs steps such as those in the parameter control optimization method for a multi-robot collaborative automotive side panel production line to achieve the following functions: collecting the welding current and welding count during each spot welding by the robot to obtain spot welding current data; collecting the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data; merging and dividing the spot welding current data and spot welding diameter data, and performing data de-identification processing to obtain valid spot welding sample data; performing current optimization analysis based on the valid spot welding sample data to obtain the correspondence between welding current and welding count under the ideal weld nugget diameter, and obtaining current count reference information; and controlling and optimizing the robot's welding current based on the current count reference information and the robot's welding count.

[0127] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0128] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A multi-robot collaborative parameter control and optimization system for automotive side panel production lines, characterized in that, It includes a parameter acquisition module, a data cleaning module, an optimization analysis module, and a control optimization module; The parameter acquisition module includes a first acquisition unit and a second acquisition unit. The first acquisition unit is used to acquire the welding current and welding number during each spot welding by the robot to obtain spot welding current data. The second acquisition unit is used to acquire the corresponding weld nugget diameter after spot welding to obtain spot welding diameter data. The data cleaning module is used to merge and divide the spot welding current data and spot welding diameter data, and perform data de-identification processing to obtain valid spot welding sample data. The optimization analysis module performs current optimization analysis based on effective sample data of spot welding, obtains the correspondence between welding current and welding number under the ideal weld nugget diameter, and obtains current number reference information; The control optimization module optimizes the welding current of the robot based on the current frequency reference information and the number of welding operations performed by the robot.

2. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 1, characterized in that, The first acquisition unit is configured with a first acquisition strategy, which includes: In the production of automobile side panels, any resistance spot welding point is designated as the first spot welding point; the robots of the same model that weld the first spot welding point are designated as spot welding robot 1 to spot welding robot n, where n is the total number of spot welding robots. The magnitude of the current applied to the welding area through the electrode when the spot welding robot welds the first weld point is recorded as the welding current, and the magnitude of the welding current set by the spot welding robot is recorded as the preset current; the number of times the spot welding robot welds the first weld point from the time the electrode is ground or replaced is recorded as the welding count. For spot welding robot 1, during each spot welding, the corresponding welding current and welding number are collected and recorded as the spot welding parameter information of spot welding robot 1. The spot welding parameter information of all spot welding robots is collected repeatedly and merged to obtain spot welding current data. The preset current corresponding to each welding number is obtained, and the preset current when the welding number is 0 is recorded as WI.

3. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 2, characterized in that, The second acquisition unit is configured with a second acquisition strategy, which includes: For spot welding robot 1, after each spot welding is completed, the diameter of the weld nugget is collected and recorded as the spot welding diameter information of spot welding robot 1; the spot welding parameter information of all spot welding robots is collected repeatedly to obtain the spot welding diameter data. Obtain the ideal weld nugget diameter of the first weld joint, denoted as the ideal weld nugget diameter AL.

4. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 3, characterized in that, The data cleaning module is configured with data cleaning strategies, which include: The range of welding times is obtained based on the spot welding current data and is sequentially recorded as welding times 1 to welding times m, where m is the maximum value of welding times in the welding current data; for welding times 1, based on the spot welding current data and spot welding diameter data, the welding current and weld nugget diameter corresponding to all welding times 1 are recorded as the original current diameter set; The welding current and weld nugget diameter collected in the same time in the original current-diameter set are combined into a data pair, denoted as welding data pair (AI, AD). After completion, the first welding dataset is obtained, where AI represents welding current and AD represents weld nugget diameter. Calculate the median ME of the welding current in the first welding dataset and calculate the absolute deviation of the median MA. Welding currents that are not located in [ME-k1×1.4826×MA, ME+k1×1.4826×MA] are recorded as outliers and marked as abnormal currents, where k1 is a set scaling factor. Repeatedly obtain the outliers corresponding to the weld nugget diameter in the first welding dataset and mark them as abnormal diameters. For any welding data pair in the first welding dataset, if it contains abnormal current or abnormal diameter, it is marked as an abnormal data pair and removed. After completion, the second welding dataset is obtained.

5. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 4, characterized in that, Data cleaning strategies also include: Calculate the mean values ​​of welding current and weld nugget diameter in the second welding dataset, denoted as IU and DU respectively, and form a column vector, denoted as mean vector U0=[IU, DU]. T ;Any welding data pair in the second welding dataset is denoted as the first data pair (BI, BD); Based on the welding current and weld nugget diameter in the second welding dataset, calculate the sample variance of the welding current and the sample variance of the weld nugget diameter, denoted as CII and CDD respectively, and calculate the sample covariance of the welding current and the weld nugget diameter, denoted as CID. The covariance matrix of welding current and weld nugget diameter is constructed and denoted as the first covariance matrix, which is as follows: , where M represents the first covariance matrix.

6. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 5, characterized in that, Data cleaning strategies also include: Convert the first data pair (BI, BD) into a column vector, denoted as [BI, BD]. T Then calculate [BI, BD] based on M. T Calculate the Mahalanobis distance BG to the mean vector U0; repeat the calculation of the Mahalanobis distances for all welding data pairs in the second welding dataset to obtain the Mahalanobis distance set. Outliers in the Mahalanobis distance set are filtered according to the 3σ principle and recorded as outlier distances. Welding data pairs corresponding to outlier distances are marked as outlier data pairs and removed. After completion, the effective welding dataset with welding number 1 is obtained. Repeatedly obtain the valid welding dataset corresponding to all welding times to obtain valid spot welding sample data.

7. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 6, characterized in that, The optimization analysis module is configured with optimization analysis strategies, which include: Based on the valid welding dataset of welding number 1, obtain the sum of the corresponding Mahalanobis distances for all welding data pairs, denoted as the weighted basis AQ; for any welding current in the valid welding dataset, denoted as EI, and the corresponding Mahalanobis distance denoted as EG, calculate EI×EG / AQ, denoted as the weighted current of EI; Repeatedly calculate the weighted current of all welding currents in the effective welding dataset, and sum them to obtain the representative value of the welding current, which is denoted as the representative current of welding number 1. Repeatedly obtain the representative value of the welding current to obtain the representative diameter of welding number 1. Repeatedly obtain the representative current and representative diameter for all welding cycles, and form a data pair with each welding cycle and the corresponding representative current and representative diameter, denoted as sample point YN=(RN, NI, ND), where RN, NI and ND represent the welding cycle, representative current and representative diameter, respectively.

8. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 7, characterized in that, Optimization analysis strategies also include: Based on the preset current corresponding to the number of welding, the sample points with the same preset current are grouped into one category and called the sample set. They are then sequentially named from sample set 1 to sample set v in descending order of the number of welding, where v is the total number of sample sets. For sample set 1, the range of welding times in sample set 1 is denoted as the number interval [AN1, BN1]; and the welding times and representative diameter in the sample set are linearly fitted to obtain the ND relationship model of sample set 1, denoted as ND1=ek1×RN1+b1, where ek1 and b1 represent the slope and intercept, respectively. Arrange the sample points in sample set 1 in ascending order of welding number, and denote any sample point as YN1j=(RN1j, NI1j, ND1j), where j represents the position number; obtain the representative current M1I and representative diameter M1D of the sample point at the end position, and denote the end current diameter 1. Repeat the process of obtaining the representative current Q2I and representative diameter Q2D of the sample point at the beginning position in sample set 2, and denote it as the beginning current diameter 2.

9. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 8, characterized in that, Optimization analysis strategies also include: Calculate (Q2I-M1I) / (Q2D-M1D), denoted as the slope IK1 of sample set 1; substitute RN1j into the ND relationship model of sample set 1, and denot the result as PD1j; calculate (AL-PD1j)×IK1, denoted as the target current corresponding to the welding number RN1j; Repeatedly obtain the target current corresponding to all welding times in sample set 1, and perform nonlinear fitting on the welding times and target current to obtain the relationship curve between welding times and target current corresponding to the interval [AN1, BN1], which is denoted as the target current curve; Repeat the number of times for all sample sets and obtain the target current curve for all number of times, which is recorded as the current number reference information.

10. The multi-robot collaborative automotive side panel production line parameter control and optimization system according to claim 9, characterized in that, The control optimization module is configured with control optimization strategies, which include: For any spot welding robot, after the previous spot welding is completed, the corresponding welding count is obtained and recorded as the current welding count; Based on the current welding cycle range, the corresponding target current is obtained using the corresponding target current curve, recorded as the next target current, and the welding current for the next spot weld is adjusted to the next target current; the welding current of all spot welding robots is controlled repeatedly.

Images