Edge rolling quality abnormity diagnosis method and system based on space-time consistency

Abstract

The invention discloses an edge rolling quality abnormity diagnosis method and system based on space-time consistency, and the method comprises the steps: collecting the space-time data in an edge rolling technology in real time, and enabling the space-time data to comprise a robot number, a process number, an edge rolling pressure value, and the three-dimensional coordinates of an edge rolling point position; acquiring a historical benchmark pressure curve and a dynamic early warning line corresponding to the current rolloff pressure value based on the spatio-temporal data; comparing the current binding pressure value with the historical benchmark pressure curve and the dynamic early warning line point by point; if the binding pressure values of M continuous data points exceed the dynamic early warning line, it is judged that binding is abnormal; if the current rolloff pressure value does not exceed the dynamic early warning line and the current rolloff pressure value is inconsistent with the trend of the historical benchmark pressure curve, judging that the rolloff is abnormal; and according to the spatio-temporal data associated with the binding pressure value corresponding to the binding abnormity, a specific robot number, a process number and a three-dimensional coordinate of a binding point location are positioned. According to the method and the system, the production efficiency of the edge rolling process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile manufacturing, and particularly relates to a method and system for diagnosing abnormal hemming quality based on spatio-temporal consistency. Background Art

[0002] The hemming process is widely used in the hemming processing of body panels (such as car doors and hoods), and its system usually consists of hemming tools, die fixtures, and robot control. In the prior art, the monitoring of hemming force is mainly achieved through laboratory theoretical models. For example, force signals are collected by sensors and their correlations with waves and hemming height are analyzed (see "Experimental Research on Hemming Force Monitoring" by Lin Juguang et al.). However, the prior art has the following defects: there are significant differences between the laboratory environment and the production site, and in the actual process, the same robot needs to perform multiple process operations (such as hemming processes in different panel areas), resulting in a large deviation between the theoretical model and the actual data. The existing methods can only judge the overall quality of the panel and cannot locate abnormal hemming equipment (such as roller wear, spring failure, die damage, etc.). The prior art takes the whole panel as the analysis unit and cannot be refined to specific hemming points, making it difficult to achieve refined monitoring. Summary of the Invention

[0003] To solve the technical problems existing in the background art, the present invention proposes a method and system for diagnosing abnormal hemming quality based on spatio-temporal consistency.

[0004] A method for diagnosing abnormal hemming quality based on spatio-temporal consistency proposed by the present invention includes:

[0005] Real-time collecting spatio-temporal data in the hemming process, where the spatio-temporal data includes robot number, process number, hemming pressure value, and three-dimensional coordinates of the hemming point;

[0006] Obtaining a historical benchmark pressure curve and a dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data;

[0007] Comparing the current hemming pressure value with the historical benchmark pressure curve and the dynamic warning line point by point;

[0008] If the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined that hemming is abnormal;

[0009] If the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that hemming is abnormal;

[0010] Locating the specific robot number, process number, and three-dimensional coordinates of the hemming point according to the spatio-temporal data associated with the hemming pressure value corresponding to the abnormal hemming.

[0011] Preferably, the generation process of the historical benchmark pressure curve and the dynamic warning line specifically includes:

[0012] Real-time collect the hemming pressure data of the robot hemming process. The hemming pressure data includes the robot number, panel identification, hemming angle value, and the corresponding pressure value.

[0013] Group the hemming pressure data by robot number, and sort the panel identifications of the panels hemmed by each robot in chronological order.

[0014] Classify the panel identifications of each robot according to the hemming angle value to form multiple angle value groups.

[0015] Within each angle value group, mark the serial numbers for the panel identifications in chronological order, and extract the corresponding hemming pressure data according to the serial numbers and angle values.

[0016] Sort the hemming data of each angle value group in descending order of pressure value, and calculate the 25th percentile Q1 and the 75th percentile Q2.

[0017] Define the upper edge as Q1+(Q2-Q1), and the lower edge as Q2-(Q2-Q1).

[0018] Clear the abnormal data that is greater than the upper edge or less than the lower edge in the same serial number and angle value group.

[0019] Calculate the pressure mean μ and the standard deviation σ for the cleaned data. Take μ as the reference point, and μ±Nσ as the upper and lower warning points. Sort the reference points and warning points of each angle value of each robot in chronological order to generate the historical benchmark pressure curve and the dynamic warning curve of the robot, where N is a preset coefficient (2≤N≤4).

[0020] Preferably, the hemming abnormalities include hemming head abnormalities, single-point quality abnormalities, die abnormalities, spring abnormalities, acquisition abnormalities, and panel abnormalities.

[0021] Preferably, if the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined as a hemming abnormality. Specifically, it includes:

[0022] If the hemming pressure values of M consecutive data points exceed the dynamic warning line, record the corresponding process number, where 3≤M≤5.

[0023] Obtain the periodic trend of exceeding the warning line corresponding to the process number.

[0024] If there is no periodic trend, the hemming abnormality is specifically a hemming head abnormality.

[0025] If there is a periodic trend, query the number of abnormalities P at this process number point in history, and compare the number of abnormalities P with a preset threshold. When the number of abnormalities P is greater than the threshold, the hemming abnormality is specifically a die abnormality, otherwise the hemming abnormality is specifically a single-point quality abnormality.

[0026] Preferably, if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal, specifically including:

[0027] When the historical benchmark pressure curve fluctuates and the real-time hemming pressure value is constant, the hemming abnormality is specifically the abnormality of the hemming head or the spring;

[0028] When the historical benchmark pressure curve is constant and the real-time hemming pressure value fluctuates, the hemming abnormality is specifically the abnormality of the spring or the acquisition abnormality;

[0029] When the real-time hemming pressure value is always above or below the historical benchmark pressure curve, the hemming abnormality is specifically the abnormality of the panel or the spring;

[0030] When the slope of the rising or falling real-time hemming pressure value is greater than that of the historical benchmark pressure curve, the hemming abnormality is specifically the abnormality of the hemming head or the spring;

[0031] When the real-time hemming pressure value fluctuates periodically and the historical benchmark pressure curve fluctuates aperiodically, the hemming abnormality is specifically the abnormality of the hemming head.

[0032] Preferably, it further includes:

[0033] If the cumulative number of abnormal times P at the same hemming point within 24 hours is ≥ 5 times, a device maintenance instruction is triggered.

[0034] Preferably, it further includes:

[0035] After each production batch is completed, the historical benchmark pressure curve and the dynamic warning line are dynamically updated through an incremental learning algorithm.

[0036] A hemming quality abnormality diagnosis system based on spatio-temporal consistency proposed by the present invention includes:

[0037] A data acquisition module, configured to collect spatio-temporal data in the hemming process in real time, where the spatio-temporal data includes the robot number, the process number, the hemming pressure value, and the three-dimensional coordinates of the hemming point;

[0038] A first processing module, configured to obtain the historical benchmark pressure curve and the dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data;

[0039] A second processing module, configured to compare the current hemming pressure value with the historical benchmark pressure curve and the dynamic warning line point by point; if the hemming pressure values of consecutive M data points exceed the dynamic warning line, it is determined that the hemming is abnormal; if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal;

[0040] Anomaly location module, which is used to locate the specific robot number, process number, and three-dimensional coordinates of the hemming point according to the spatio-temporal data associated with the hemming pressure value corresponding to the hemming anomaly.

[0041] In the present invention, the proposed method and system for diagnosing hemming quality anomalies based on spatio-temporal consistency collect spatio-temporal data in the hemming process in real time, construct a benchmark pressure curve and a dynamic warning line for spatio-temporal grouping, and combine trend consistency analysis and multi-source data fusion technology to achieve precise classification and location of hemming quality anomalies and equipment failures. The present invention does not require additional hardware investment, supports incremental learning to dynamically optimize the model, and improves the reliability and production efficiency of the hemming process in automobile manufacturing. Brief Description of the Drawings

[0042] Figure 1 It is a schematic diagram of the working process of a method for diagnosing hemming quality anomalies based on spatio-temporal consistency proposed by the present invention;

[0043] Figure 2 It is a schematic diagram of the generation process of the historical benchmark pressure curve and the dynamic warning line of a method for diagnosing hemming quality anomalies based on spatio-temporal consistency proposed by the present invention;

[0044] Figure 3 It is a schematic diagram of the diagnostic implementation process of a method for diagnosing hemming quality anomalies based on spatio-temporal consistency proposed by the present invention;

[0045] Figure 4 It is a schematic diagram of the system architecture of a system for diagnosing hemming quality anomalies based on spatio-temporal consistency proposed by the present invention. Detailed Embodiment

[0046] Refer to Figures 1-4 A method for diagnosing hemming quality anomalies based on spatio-temporal consistency proposed by the present invention includes the following steps:

[0047] S1. Collect spatio-temporal data in the hemming process in real time. The spatio-temporal data includes the robot number, process number, hemming pressure value, and three-dimensional coordinates of the hemming point.

[0048] S2. Obtain the historical benchmark pressure curve and the dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data.

[0049] In this embodiment, as Figure 2 shown, the generation process of the historical benchmark pressure curve and the dynamic warning line specifically includes:

[0050] Collect the hemming pressure data of the robot hemming process in real time. The hemming pressure data includes the robot number, panel identification, hemming angle value, and the corresponding pressure value;

[0051] Group the hemming pressure data by robot number and sort the panel identifiers of each robot's hemming in chronological order;

[0052] Classify the panel identifiers of each robot according to the hemming angle value to form multiple angle value groups;

[0053] Within each angle value group, mark the serial numbers for the panel identifiers in chronological order, and extract the corresponding hemming pressure data according to the serial numbers and angle values;

[0054] Sort the hemming data of each angle value group in descending order of pressure value, and calculate the 25% quantile Q1 and the 75% quantile Q2;

[0055] Define the upper edge as Q1+(Q2-Q1) and the lower edge as Q2-(Q2-Q1);

[0056] Clear the abnormal data in the same serial number and angle value group that is greater than the upper edge or less than the lower edge;

[0057] Calculate the pressure mean μ and the standard deviation σ for the cleaned data. Taking μ as the reference point, μ±Nσ are the upper and lower warning points; sort the reference points and warning points of each angle value of each robot in chronological order to generate the historical benchmark pressure curve and the dynamic warning curve of the robot, where N is a preset coefficient (2≤N≤4).

[0058] S3. Compare the current hemming pressure value with the historical benchmark pressure curve and the dynamic warning line point by point.

[0059] If the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined that the hemming is abnormal.

[0060] If the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal.

[0061] In this embodiment, as Figure 3 shown, the hemming abnormalities include hemming head abnormalities, single-point quality abnormalities, die abnormalities, spring abnormalities, acquisition abnormalities, and panel abnormalities.

[0062] In this embodiment, if the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined that the hemming is abnormal, specifically including:

[0063] If the hemming pressure values of M consecutive data points exceed the dynamic warning line, record the corresponding process number, where 3≤M≤5;

[0064] Obtain the periodic trend of exceeding the warning line corresponding to the process number;

[0065] If there is no periodic trend, the hemming abnormality is specifically a hemming head abnormality;

[0066] If there is a periodic trend, query the number P of abnormalities at the process number point in the history, and compare the number P of abnormalities with a preset threshold. When the number P of abnormalities is greater than the threshold, the hemming abnormality is specifically a diaphragm abnormality; otherwise, the hemming abnormality is specifically a single-point quality abnormality.

[0067] In this embodiment, if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined as a hemming abnormality, specifically including:

[0068] When the historical benchmark pressure curve fluctuates and the real-time hemming pressure value is constant, the hemming abnormality is specifically a hemming head or spring abnormality;

[0069] When the historical benchmark pressure curve is constant and the real-time hemming pressure value fluctuates, the hemming abnormality is specifically a spring abnormality or a collection abnormality;

[0070] When the real-time hemming pressure value is constantly above or below the historical benchmark pressure curve, the hemming abnormality is specifically a panel abnormality or a spring abnormality;

[0071] When the slope of the rise or fall of the real-time hemming pressure value is greater than that of the historical benchmark pressure curve, the hemming abnormality is specifically a hemming head or spring abnormality;

[0072] When the real-time hemming pressure value fluctuates periodically and the historical benchmark pressure curve fluctuates non-periodically, the hemming abnormality is specifically a hemming head abnormality.

[0073] S4. Locate the specific robot number, process number, and three-dimensional coordinates of the hemming point according to the spatio-temporal data associated with the hemming pressure value corresponding to the hemming abnormality.

[0074] In this embodiment, it further includes:

[0075] If the cumulative number of abnormalities P at the same hemming point within 24 hours is ≥ 5 times, trigger an equipment maintenance instruction.

[0076] In this embodiment, it further includes:

[0077] After the end of each production batch, dynamically update the historical benchmark pressure curve and the dynamic warning line through an incremental learning algorithm.

[0078] Refer to Figures 1-4 , A hemming quality abnormality diagnosis system based on spatio-temporal consistency proposed by the present invention includes:

[0079] A data acquisition module for real-time acquisition of spatio-temporal data in the hemming process, where the spatio-temporal data includes robot number, process number, hemming pressure value, and three-dimensional coordinates of the hemming point;

[0080] The first processing module is used to obtain the historical benchmark pressure curve and the dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data;

[0081] The second processing module is used to compare the current hemming pressure value with the historical benchmark pressure curve and the dynamic warning line point by point; if the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined that the hemming is abnormal; if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal;

[0082] The abnormal positioning module is used to locate the specific robot number, process number, and three-dimensional coordinates of the hemming point according to the spatio-temporal data associated with the hemming pressure value corresponding to the hemming abnormality.

[0083] As mentioned above, it is only the preferred specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A method for diagnosing the abnormality of hemming quality based on spatio-temporal consistency, characterized in that, Including: Real-time collection of spatio-temporal data in the hemming process, where the spatio-temporal data includes robot number, process number, hemming pressure value, and three-dimensional coordinates of the hemming points; Obtaining the historical benchmark pressure curve and dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data; Point-by-point comparison of the current hemming pressure value with the historical benchmark pressure curve and dynamic warning line; If the hemming pressure values of consecutive M data points exceed the dynamic warning line, it is determined that the hemming is abnormal; If the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal; Locating the specific robot number, process number, and three-dimensional coordinates of the hemming points according to the spatio-temporal data associated with the hemming pressure value corresponding to the hemming abnormality.

2. The method for diagnosing abnormal hemming quality based on spatio-temporal consistency according to claim 1, wherein The generation process of the historical benchmark pressure curve and dynamic warning line specifically includes: Real-time collection of hemming pressure data of the robot hemming process, where the hemming pressure data includes robot number, panel identification, hemming angle value, and corresponding pressure value; Grouping the hemming pressure data by robot number and sorting the panel identifications of each robot's hemming in chronological order; Classifying the panel identifications of each robot according to the hemming angle value to form multiple angle value groups; Within each angle value group, marking the serial numbers for the panel identifications in chronological order and extracting the corresponding hemming pressure data according to the serial numbers and angle values; Sorting the hemming data of each angle value group in descending order of pressure value, and calculating the 25% quantile Q1 and 75% quantile Q2; Defining the upper edge as Q1+(Q2-Q1) and the lower edge as Q2-(Q2-Q1); Clearing the abnormal data greater than the upper edge or less than the lower edge in the same serial number and angle value group; Calculating the pressure mean μ and standard deviation σ for the cleaned data, using μ as the reference point, μ±Nσ as the upper and lower warning points; sorting the reference points and warning points of each angle value of each robot in chronological order to generate the historical benchmark pressure curve and dynamic warning curve of the robot, where N is a preset coefficient (2≤N≤4).

3. The method for diagnosing abnormal hemming quality based on spatio-temporal consistency according to claim 1, wherein, The hemming abnormalities include hemming head abnormality, single-point quality abnormality, die abnormality, spring abnormality, acquisition abnormality, and panel abnormality.

4. The method for diagnosing abnormal hemming quality based on spatio-temporal consistency according to claim 3, wherein The case where if the hemming pressure values of consecutive M data points exceed the dynamic warning line, it is determined that the hemming is abnormal, specifically includes: If the hemming pressure values of consecutive M data points exceed the dynamic warning line, record the corresponding process number, where 3≤M≤5; Obtaining the periodic trend of exceeding the warning line corresponding to the process number; If there is no periodic trend, the hemming abnormality is specifically a hemming head abnormality; If there is a periodic trend, query the number P of abnormal points of this process number in history and compare the number P of abnormalities with a preset threshold. When the number P of abnormalities is greater than the threshold, the hemming abnormality is specifically a die abnormality, otherwise the hemming abnormality is specifically a single-point quality abnormality.

5. The method for diagnosing the abnormal hemming quality based on spatio-temporal consistency according to claim 3, wherein The case where if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with the historical benchmark pressure curve, it is determined that the hemming is abnormal, specifically includes: When the historical benchmark pressure curve fluctuates and the real-time hemming pressure value is constant, the hemming abnormality is specifically a hemming head or spring abnormality; When the constant value of the historical benchmark pressure curve remains unchanged and the real-time hemming pressure value fluctuates, the hemming abnormality is specifically a spring abnormality or a collection abnormality; When the real-time hemming pressure value is constantly above or below the historical benchmark pressure curve, the hemming abnormality is specifically a panel abnormality or a spring abnormality; When the slope of the rise or fall of the real-time hemming pressure value is greater than that of the historical benchmark pressure curve, the hemming abnormality is specifically a hemming head or a spring abnormality; When the real-time hemming pressure value fluctuates periodically and the historical benchmark pressure curve fluctuates aperiodically, the hemming abnormality is specifically a hemming head abnormality.

6. The method for diagnosing abnormal hemming quality based on spatio-temporal consistency according to claim 4, characterized in that It also includes: If the cumulative number of abnormal times P at the same hemming point within 24 hours is greater than or equal to 5 times, a device maintenance instruction is triggered.

7. The edge rolling quality anomaly diagnosis method based on spatio-temporal consistency according to claim 1, wherein It also includes: After each production batch is completed, the historical benchmark pressure curve and the dynamic warning line are dynamically updated through an incremental learning algorithm.

8. A spatio-temporal consistency-based hemming quality anomaly diagnosis system, characterized in that, It includes: A data acquisition module for real-time collecting spatio-temporal data in the hemming process, where the spatio-temporal data includes the robot number, the process number, the hemming pressure value, and the three-dimensional coordinates of the hemming point; A first processing module for obtaining the historical benchmark pressure curve and the dynamic warning line corresponding to the current hemming pressure value based on the spatio-temporal data; A second processing module for comparing the current hemming pressure value with the historical benchmark pressure curve and the dynamic warning line point by point; if the hemming pressure values of M consecutive data points exceed the dynamic warning line, it is determined that there is a hemming abnormality; if the current hemming pressure value does not exceed the dynamic warning line and the trend of the current hemming pressure value is inconsistent with that of the historical benchmark pressure curve, it is determined that there is a hemming abnormality; An abnormality positioning module for positioning the specific robot number, the process number, and the three-dimensional coordinates of the hemming point according to the spatio-temporal data associated with the hemming pressure value corresponding to the hemming abnormality.

Images