Prediction method and system for coating defects in automobile paint shops

Abstract

The present application provides a method and system for predicting coating defects in an automobile paint shop, the method comprising: establishing an initial model based on the correlation between coating defects and influencing factors of coating defects; training the initial model to obtain a prediction model; based on the prediction model, probabilistically predicting defects within a predetermined time period in the future to obtain a prediction result; comparing the prediction result with a target value set by the system; in response to the prediction result being greater than the target value, generating a first optimization value based on the deviation between the prediction result and the target value through the prediction model, the optimization value is used to adjust the parameters of the man-machine-material-method-loop in the automobile paint shop. The method and system for predicting coating defects in an automobile paint shop provided by the present application can predict the occurrence of defects in advance, and minimize the occurrence of coating defects by adjusting parameters, thereby preventing the occurrence of defects in large quantities and reducing the number of returned vehicles.

Description

Technical Field

[0001] The present application relates to the technical field of automobile painting, and in particular to a method for predicting painting defects in an automobile painting shop. Background Art

[0002] With the development of science and technology and the progress of society, people's demand for automobiles is increasing. As consumers have higher and higher requirements for the appearance of automobiles, the quality control of automobile painting is particularly important. However, various defects often occur in the process of automobile painting, which affects the beauty and service life of the car, and also brings huge economic losses to the company. Therefore, how to effectively prevent and reduce the occurrence of automobile painting defects is an urgent problem to be solved. At present, the quality defects of automobile painting workshops are mostly under fire-fighting quality management. After the batch quality defects are produced, the process engineers will conduct a systematic analysis, find out the causes and formulate measures to verify the effectiveness of the measures. Or adopt a checklist-based preventive quality management, and take preventive quality control measures in advance through standardized operations, control lists, control plans and other means.

[0003] However, in most cases, the real cause cannot be quickly identified, the investigation cycle of the real cause of batch defects is long, and because the coating process is long, by the time batch defects are discovered during the rework process, a large batch has already been caused. The defects cannot be discovered in time, resulting in a large number of reworked vehicles. Summary of the invention

[0004] To at least partially solve the above-mentioned problems, according to the first aspect of the present application, an embodiment of the present application provides a method for predicting paint defects in an automobile paint shop, comprising: establishing an initial model based on the correlation between paint defects and influencing factors of paint defects; training the initial model to obtain a prediction model; based on the prediction model, probabilistically predicting defects within a predetermined time period in the future to obtain prediction results; comparing the prediction results with a target value set by the system; in response to the prediction result being greater than the target value, generating a first optimization value based on the deviation between the prediction result and the target value through the prediction model, and the optimization value is used to adjust the parameters of the man-machine-material-method loop in the automobile paint shop.

[0005] In some embodiments, the prediction method also includes: after a single car body passes through the topcoat production, real-time collection of man-machine-material-method-loop parameters related to the single car body, and predicting the defect information of the single car body based on the man-machine-material-method-loop parameters and the prediction model; in response to abnormal defect information of the single car body, a reminder is issued to the wet film inspector after the topcoat spraying.

[0006] In some embodiments, issuing a reminder to a wet film inspector after topcoat spraying includes: displaying a defective portion of a single car body through a single car body butterfly diagram presented on a display in a car paint shop.

[0007] In some embodiments, the prediction method also includes: recording the defect type, the vehicle model causing the defect, and the time when the defect occurred; in response to the number of occurrences of a predetermined type of defect exceeding a threshold, determining the defect occurrence range based on the vehicle model causing the defect of the predetermined type of defect and the time when the defect occurred; calculating a second optimized value within the defect occurrence range through a prediction model, and the second optimized value is used to adjust the man-machine-material-method-environment parameters within the defect occurrence range in the automobile paint shop.

[0008] In some embodiments, the future predetermined time period includes the next 1 hour, 4 hours, and 12 hours.

[0009] In some embodiments, the prediction method further includes: the prediction model performs deviation correction based on the prediction result of the next hour and the actual defect occurrence value after one hour.

[0010] In some embodiments, the coating defects include paint residue, shrinkage cavities, fiber hair, insufficient paint, and sagging.

[0011] In some embodiments, the factors affecting the sagging defect include robot air pressure, spray molding air volume, rotary cup speed, ambient temperature, ambient humidity, and material viscosity.

[0012] In some embodiments, training the initial model to obtain the prediction model includes: establishing machine learning for quality warning through a regression model and a linear and random forest algorithm to obtain the prediction model.

[0013] According to the second aspect of the present application, an embodiment of the present application also provides a system for predicting paint defects in an automobile paint shop, and the prediction system is used to execute the method for predicting paint defects in an automobile paint shop provided in any embodiment of the first aspect of the present application.

[0014] The embodiments of the present application provide a method and system for predicting paint defects in an automobile paint shop. A prediction model is established and used to predict the probability of defects within a predetermined time period in the future. The prediction results are used to guide the adjustment of the man-machine-material-method-environment parameters in the current automobile paint shop, thereby predicting the occurrence of defects in advance and minimizing the occurrence of paint defects by adjusting the parameters, thereby preventing the occurrence of defects in large quantities and reducing the number of returned vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] Figure 1 It is a flow chart of a method for predicting painting defects in an automobile paint shop provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The technical scheme of the preferred embodiment of the present application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection claimed in the present application.

[0018] The words "first", "second" and similar words used in this application do not indicate any order, quantity or importance, but are only used to distinguish different parts. The words "include" or "comprise" and similar words mean that the elements before the word include the elements listed after the word, and do not exclude the possibility of including other elements. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] The present invention proposes a new prediction method and prediction system for coating defects in automobile paint shops, aiming to build a comprehensive and reliable prediction model through technical means such as data mining and statistical analysis, and provide process engineers with more accurate and effective decision-making basis through pre-prediction, in-process judgment and post-analysis, thereby realizing automatic prediction of defects in the automobile painting process and avoiding fire-fighting quality management.

[0020] According to the first aspect of the present application, an embodiment of the present application provides a method for predicting paint defects in an automobile paint shop, including: establishing an initial model based on the correlation between paint defects and influencing factors of paint defects; training the initial model to obtain a prediction model; based on the prediction model, probabilistically predicting defects within a predetermined time period in the future to obtain prediction results; comparing the prediction results with a target value set by the system; in response to the prediction result being greater than the target value, generating a first optimization value based on the deviation between the prediction result and the target value through the prediction model, and the optimization value is used to adjust the parameters of the man-machine-material-method loop in the automobile paint shop.

[0021] In this embodiment, the initial model refers to a basic model for predicting future trends based on historical data, which includes the correlation and weight coefficients of various influencing factors. Through in-depth analysis and research on various data collected in the early stage, multiple key influencing factors can be obtained, such as paint formula, spray gun angle, spray speed, etc. These factors interact and restrict each other, and jointly determine the quality and performance level of the final product. After the initial model is obtained through preliminary modeling, a series of training and testing work can be carried out on the obtained initial model in order to further improve and optimize it to obtain a prediction model. In some embodiments, the initial model is trained to obtain a prediction model, including: establishing a machine learning for quality warning through a regression model and a linear and random forest algorithm to obtain a prediction model.

[0022] In this embodiment, after obtaining the prediction model, it can be applied to the actual situation for defect prediction. Specifically, the current or required human-machine-material-method-environment parameters can be input into the model, and then the prediction results of the coating defects can be obtained. The prediction results can reflect the probability of a certain defect that may occur in the future, and the probability value can be visualized for reference and decision-making by relevant personnel. In this embodiment, under normal circumstances, an expected defect rate threshold is set in advance to measure whether the prediction results meet expectations. If the predicted probability is higher than this threshold, it means that there may be a greater quality risk; otherwise, it indicates that the human-machine-material-method-environment parameter control measures on the current production line are relatively appropriate and do not require excessive intervention.

[0023] If the prediction results show that there is a high probability of defects occurring in a certain period of time in the future, some measures can be taken to reduce this risk. Specifically, the production process and equipment status can be improved by adjusting certain parameters of the human-machine-material-method-environment, thereby reducing the output rate of defective products. For example, if a prediction result shows that the angle of paint spraying is not appropriate, it may cause problems such as uneven paint film thickness. At this time, the angle parameters of the spray gun can be appropriately adjusted to make it closer to the theoretical optimal value. More specifically, if the predicted value exceeds the target value, it will be pushed to the process engineer through a task message. After receiving the message, the process engineer can click the analysis or optimization button to enter the specific analysis interface. In the analysis interface, the TOP factors and factor scores, factor abnormal directions, and factor optimization directions that affect the defect rate can be seen. The simulation optimization button can be used to automatically simulate and calculate the first optimization value of the TOP factor, and the comparison of the defect rate before and after the simulation can be seen. The process engineer can provide on-site guidance based on the simulated optimization values.

[0024] The embodiments of the present application provide a method for predicting paint defects in an automobile paint shop. A prediction model is established and used to predict the probability of defects within a predetermined time period in the future. The prediction results are used to guide the adjustment of the man-machine-material-method-environment parameters in the current automobile paint shop, so that the occurrence of defects can be predicted in advance, and the occurrence of paint defects can be minimized by adjusting the parameters, thereby preventing the occurrence of defects in large quantities and reducing the number of returned vehicles.

[0025] In some embodiments, the prediction method also includes: after a single car body passes through the topcoat production, real-time collection of man-machine-material-method-loop parameters related to the single car body, and predicting the defect information of the single car body based on the man-machine-material-method-loop parameters and the prediction model; in response to abnormal defect information of the single car body, a reminder is issued to the wet film inspector after the topcoat spraying.

[0026] In this embodiment, in addition to making advance predictions for the entire production line, we can also make in-process judgments for individual vehicle bodies on the production line to discover local problems earlier and take targeted remedial measures to avoid large-scale repairs or even scrapping. The specific approach is to conduct data analysis on various parameters such as man-machine-material-method-environment on each vehicle body through real-time monitoring after each vehicle body passes the topcoat spraying process to obtain real-time prediction results. Once the real-time prediction results of a single vehicle body are found to be abnormal, the wet film inspector will be notified immediately to go to the site to check whether there are any visible defects (such as color difference, bubbles, particles, etc.).

[0027] Optionally, a reminder is issued to the wet film inspector after the topcoat is sprayed, including: displaying the location of defects in the single body through a single body butterfly diagram presented on a display in the automobile paint shop. In this embodiment, by presenting a single body butterfly diagram on a display in the automobile paint shop, the location of defects on the body can be intuitively displayed, which is convenient for inspectors to quickly locate and take corresponding measures. Specifically, a special function module can be added to the software interface, and a visual butterfly diagram can be automatically generated based on the prediction data generated by these prediction models. The butterfly diagram can intuitively show the location of defects in the body, which is convenient for manual review.

[0028] In some embodiments, the prediction method also includes: recording the defect type, the vehicle model causing the defect, and the time when the defect occurred; in response to the number of occurrences of a predetermined type of defect exceeding a threshold, determining the defect occurrence range based on the vehicle model causing the defect of the predetermined type of defect and the time when the defect occurred; calculating a second optimized value within the defect occurrence range through a prediction model, and the second optimized value is used to adjust the man-machine-material-method-environment parameters within the defect occurrence range in the automobile paint shop.

[0029] In this embodiment, when the number of occurrences of a predetermined type of defect exceeds a threshold, a post-analysis can be performed in the above manner. Several classification standards can be set according to actual conditions, such as color difference, dirty spot, crack, etc. according to the nature of the defect; sedan, SUV, MPV, etc. according to the vehicle model. Specifically, when a defect exceeds the standard, the process engineer can filter out the defect information by time period, defect type, vehicle model and other information, and can automatically calculate the TOP factors and factor scores, factor abnormal directions, factor optimization directions, and historical trends and abnormal information of this factor that affect the defects within this range through the regression model and linear, random forest and other algorithms. The process engineer can automatically calculate the optimal value of the TOP factor (i.e., the second optimized value) through the simulation optimization function to guide the on-site parameter adjustment.

[0030] In some embodiments, the future predetermined time period includes the next 1 hour, 4 hours, and 12 hours. In this embodiment, the defect prediction for the above time period is neither too long to make a timely response nor too short to capture sufficient information. Generally speaking, the closer the selected interval is to the current time point, the more conducive it is for us to make accurate predictions.

[0031] In some embodiments, the prediction method also includes: the prediction model is based on the prediction results of the next hour and the actual defect occurrence value after 1 hour to make deviation corrections. In this embodiment, in order to avoid the prediction results from deviating too much from the actual value due to external interference and other factors, we need to retrain and tune the model regularly. Specifically, a part of new samples can be extracted from the historical data set in the recent period to supplement the training set, and the defect probability distribution within the next hour can be predicted based on the latest version of the model. Subsequently, the prediction results are compared and analyzed with the actual occurrence values ​​to observe the degree of difference between the two. If the gap is large, it means that the model is overfitting or underfitting, and the parameters need to be readjusted or other compensation strategies need to be introduced.

[0032] In some embodiments, coating defects include paint residue, shrinkage, fiber hair, insufficient paint, and sagging. In some embodiments, factors affecting sagging defects include robot air pressure, spray molding air volume, cup speed, ambient temperature, ambient humidity, and material viscosity.

[0033] According to the second aspect of the present application, an embodiment of the present application also provides a system for predicting paint defects in an automobile paint shop, and the prediction system is used to execute the method for predicting paint defects in an automobile paint shop provided in any embodiment of the first aspect of the present application.

[0034] The prediction method and prediction system of coating defects in automobile paint shops provided by the embodiments of the present application, through the methods of prior prediction, in-process judgment, and post-analysis, based on the association between coating defects and influencing factors, big data analysis and early warning model construction are carried out, and with the help of regression models and linear, random forest and other algorithms, a method and system for predicting automobile coating defect quality are established. Single vehicle defect information can be predicted in real time, and the wet film inspectors of the topcoat can be reminded to focus on it through the large-screen butterfly pattern style to reduce batch outflow. When batch defects exceed the standard, the present invention can filter out defect information through information such as time period, defect type, and vehicle model, and can automatically calculate the TOP factors and factor scores, factor abnormal directions, factor optimization directions, and historical trends and abnormal information of this factor that affect defects within this range through regression models and linear, random forest and other algorithms. Process engineers can automatically calculate the optimal value of the TOP factor through the simulation optimization function to guide on-site parameter adjustment.

[0035] Based on the above-mentioned embodiments of the present application, in the absence of explicit negation or conflict, the technical features of one embodiment may be beneficially combined with one or more other embodiments.

[0036] Although some specific embodiments of the present application have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are only for illustration and not for limiting the scope of the present application. It should be understood by those skilled in the art that the above embodiments may be modified or some technical features may be replaced by equivalents without departing from the scope and spirit of the present application. The scope of the present application is defined by the appended claims.

Claims

1. A method for predicting coating defects in an automobile coating shop, characterized in that: include: Establishing an initial model based on the association between coating defects and influencing factors of the coating defects; Training the initial model to obtain a prediction model; Based on the prediction model, a probability prediction of defects within a predetermined time period in the future is performed to obtain a prediction result; Comparing the prediction result with the target value set by the system; In response to the prediction result being greater than the target value, a first optimization value is generated by the prediction model based on a deviation between the prediction result and the target value, and the optimization value is used to adjust parameters of a man-machine-material-method loop in an automobile paint shop.

2. The method for predicting coating defects in an automobile coating shop according to claim 1, characterized in that: Also includes: After a single body passes through the topcoat production, real-time collection of man-machine-material-method-loop parameters related to the single body, and prediction of defect information of the single body based on the man-machine-material-method-loop parameters and the prediction model; In response to the abnormal defect information of the single vehicle body, a reminder is sent to the wet film inspector after the topcoat is sprayed.

3. The method for predicting coating defects in an automobile coating shop according to claim 2, characterized in that: Alerts to wet film inspectors after topcoat spraying, including: Display of defective areas in a single body through a single body butterfly diagram presented on a monitor in the automotive paint shop.

4. The method for predicting coating defects in an automobile coating shop according to claim 2, characterized in that: Also includes: Record the defect type, vehicle model causing the defect, and time of occurrence; In response to the number of occurrences of a predetermined type of defect exceeding a threshold, determining a defect occurrence range based on the vehicle model on which the predetermined type of defect occurs and the time when the defect occurs; A second optimization value within the defect occurrence range is calculated by the prediction model, and the second optimization value is used to adjust the man-machine-material-method-environment parameters within the defect occurrence range in the automobile paint shop.

5. The method for predicting coating defects in an automobile coating shop according to claim 1, characterized in that: The future scheduled time period includes the future 1 hour, 4 hours and 12 hours.

6. The method for predicting coating defects in an automobile coating shop according to claim 5, characterized in that: Also includes: The prediction model performs deviation correction based on the prediction result for the next hour and the actual defect occurrence value after one hour.

7. The method for predicting coating defects in an automobile coating shop according to claim 1, characterized in that: The coating defects include paint residue, shrinkage cavities, fiber hair, insufficient paint and sagging.

8. The method for predicting coating defects in an automobile coating shop according to claim 7, characterized in that: The factors that affect sagging defects include robot air pressure, spray molding air volume, rotary cup speed, ambient temperature, ambient humidity, and material viscosity.

9. The method for predicting coating defects in an automobile coating shop according to claim 1, characterized in that: Training the initial model to obtain a prediction model includes: Through regression model and linear and random forest algorithms, machine learning for quality warning is established to obtain the prediction model.

10. A system for predicting coating defects in an automobile coating shop, characterized in that: The prediction system is used to execute the method for predicting painting defects in an automobile paint shop as described in any one of claims 1-9.

Images