A method for early warning of vehicle size deviation and prediction of deformation trend

Abstract

The present invention relates to a method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, and the method includes: obtaining the deviation between the actual value and the theoretical value of the three-dimensional coordinates of the characteristic points of the white body to form a data set; performing an overall deformation AI analysis based on the deviation data in the data set to determine the overall dimension deformation of the vehicle; performing a local out-of-tolerance AI analysis based on the deviation data in the data set to determine the local out-of-tolerance area. Compared with the prior art, the present invention can timely discover the dimensional problems in the production process, make rectifications in a timely manner, and avoid the problems of difficult vehicle repair, long problem-solving cycle, and large consumption of human and material resources after the event.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle dimension deformation analysis, and in particular to a method for extracting vehicle dimension deformation trends and early warning of out-of-tolerance situations. Background Art

[0002] In traditional manufacturing, the way to handle dimension problems is to solve problems after they occur. It has the following defects: 1. It is difficult to repair the whole vehicle after the problem is solved; 2. The problem-solving cycle is long, and the cost of process communication and inspection is high; 3. The determined solution needs to be verified again, consuming a large amount of manpower and material resources.

[0003] During the automobile production process, the overall deformation of the whole vehicle dimension at the general assembly station has a particularly great impact on the four doors, two covers, and the front and rear windshields. Therefore, it is very significant to identify the overall deformation at the general assembly station for dimension control. On the other hand, for vehicles in on-line production, dimension deformations caused by reasons such as parts, tooling, or operations need to be identified as early as possible and rectified in a timely manner. Therefore, the above-mentioned way of handling dimension problems in traditional manufacturing is no longer applicable. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for extracting vehicle dimension deformation trends and early warning of out-of-tolerance situations to overcome the defects existing in the above-mentioned prior art.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A method for extracting vehicle dimension deformation trends and early warning of out-of-tolerance situations, the method comprising:

[0007] Obtaining the deviation between the actual value and the theoretical value of the three-dimensional coordinates of the white body feature points to form a data set;

[0008] Performing overall deformation AI analysis based on the deviation data in the data set to determine the overall dimension deformation of the vehicle;

[0009] Performing local out-of-tolerance AI analysis based on the deviation data in the data set to determine the local out-of-tolerance area.

[0010] Preferably, when the method is executed, overall deformation AI analysis is first performed. If there is overall dimension deformation of the vehicle, local out-of-tolerance AI analysis is performed. If there is no overall dimension deformation of the vehicle, the process ends.

[0011] Preferably, the overall deformation AI analysis includes longitudinal deformation analysis of the white body side panels and transverse deformation analysis of the white body side panels.

[0012] Preferably, the longitudinal deformation analysis of the white body side panels specifically includes:

[0013] (a1) Divide the side panel of the body-in-white longitudinally into upper and lower regions, respectively count the proportions of the number of feature points with deviation greater than 0 and less than 0 in the upper and lower regions among all the feature points in the corresponding regions. If it exceeds the set threshold, preliminarily determine that there is longitudinal deformation and execute the (b1) process;

[0014] (b1) For the plane to be analyzed, determine the top envelope line formed by the feature points in this plane as the highest feature point band, and determine the bottom envelope line formed by the feature points in this plane as the lowest feature point band. For any feature point in the lowest feature point band, take it as the feature point to be compared. Obtain the feature points falling within the highest feature point band within a certain width range on the left and right of the feature point to be compared as the comparison feature points. Calculate one by one whether the deviation data between the comparison feature points and the feature point to be compared is greater than the tolerance band. Count the number of all feature points with deviation data greater than the tolerance band, and calculate the proportion of the feature points greater than the tolerance band in the total number of feature points in the highest feature point band and the lowest feature point band. If it is greater than the set threshold, determine that there is longitudinal deformation.

[0015] Preferably, the transverse deformation analysis of the side panel of the body-in-white specifically includes:

[0016] (a2) Divide the side panel of the body-in-white transversely into left and right regions, respectively count the proportions of the number of feature points with deviation greater than 0 and less than 0 in the left and right regions among all the feature points in the corresponding regions. If it exceeds the set threshold, preliminarily determine that there is transverse deformation and execute the (b2) process;

[0017] (b2) For the plane to be analyzed, determine the left envelope line formed by the feature points in this plane as the left feature point band, and determine the right envelope line formed by the feature points in this plane as the right feature point band. For any feature point in the left feature point band, take it as the feature point to be compared. Obtain the feature points falling within the right feature point band within a certain width range above and below the feature point to be compared as the comparison feature points. Calculate one by one whether the deviation data between the comparison feature points and the feature point to be compared is greater than the tolerance band. Count the number of all feature points with deviation data greater than the tolerance band, and calculate the proportion of the feature points greater than the tolerance band in the total number of feature points in the left feature point band and the right feature point band. If it is greater than the set threshold, determine that there is transverse deformation.

[0018] Preferably, the local out-of-tolerance AI analysis includes:

[0019] Divide the deviation data in the dataset into two parts according to the positive and negative values, namely the positive deviation set and the negative deviation set;

[0020] Perform density-based clustering on the data in the positive deviation set and the negative deviation set respectively to obtain clustering clusters, and each clustering cluster represents a local out-of-tolerance region.

[0021] Preferably, the method further includes: for n continuously detected vehicles, analyzing the change trends of the overall vehicle size deformation and the local out-of-tolerance regions of the continuous samples. If the size deformations and out-of-tolerance regions of the continuous samples tend to be consistent, a warning is given.

[0022] Preferably, if there is overall vehicle size deformation in all continuous samples, the coincidence degree of the out-of-tolerance regions of the continuous samples is determined. If the coincidence degree is greater than the set value, a warning is given.

[0023] Preferably, the coincidence degree of the out-of-tolerance regions is expressed as the proportion of the feature points with overlapping positions in the out-of-tolerance region to the total number of feature points in the out-of-tolerance region.

[0024] Preferably, the overall vehicle size deformation condition and / or the local out-of-tolerance region are visually displayed in the form of a 3D scatter plot.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] (1) The present invention uses AI technology for deformation and out-of-tolerance warning, solves the vehicle body size problems caused by the deformation of the general assembly station and the local out-of-tolerance of dimensions during the production process, establishes clear deformation rules and local out-of-tolerance rules, and the AI system tracks and gives warning feedback based on these rules, can timely detect the dimensional problems in the production process, and timely rectify them, avoiding the problems of difficult vehicle body repair, long problem-solving cycle, and large consumption of manpower and material resources after the event;

[0027] (2) The present invention identifies and warns of the change trends of the overall size deformation and local out-of-tolerance regions of n consecutive vehicles, and timely discovers problems in the mass production process;

[0028] (3) The warning system of the present invention is combined with the AI system, and the warning problems can be intuitively displayed in a visual way, realizing the systematic integration and visual management of AI intelligent dimension solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is the overall process block diagram of a method for extracting the vehicle size deformation trend and out-of-tolerance warning of the present invention;

[0030] Figure 2 is the specific process schematic diagram of the present invention integrating overall deformation AI analysis, local out-of-tolerance AI analysis and continuous vehicle analysis;

[0031] Figure 3 is the process block diagram of the clustering method during the local out-of-tolerance AI analysis of the present invention;

[0032] Figure 4 is the schematic diagram of the visualization display result after out-of-tolerance analysis in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Note that the following description of the embodiments is only illustrative in nature, and the present invention is not intended to limit the objects to which it applies or its uses, and the present invention is not limited to the following embodiments.

[0034] Embodiment

[0035] As Figure 1 shown, this embodiment provides a method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, and the method includes:

[0036] Obtain the deviation between the actual value and the theoretical value of the three-dimensional coordinates of the feature points of the body-in-white to form a data set;

[0037] Based on the deviation data in the data set, perform an overall deformation AI analysis to determine the overall dimension deformation of the vehicle;

[0038] Based on the deviation data in the data set, perform a local out-of-tolerance AI analysis to determine the local out-of-tolerance area.

[0039] As a preferred embodiment, when the method is executed, an overall deformation AI analysis is first performed. If there is an overall dimension deformation of the vehicle, a local out-of-tolerance AI analysis is performed. If there is no overall dimension deformation of the vehicle, the process ends.

[0040] In addition, the method further includes: for n continuously detected vehicles, analyze the change trends of the overall dimension deformation and the local out-of-tolerance areas of the continuous samples. If the dimension deformations and the out-of-tolerance areas of the continuous samples tend to be consistent, a warning is issued. If there is an overall dimension deformation of the vehicle in all continuous samples, determine the coincidence degree of the out-of-tolerance areas of the continuous samples. If the coincidence degree is greater than the set value, a warning is issued. The coincidence degree of the out-of-tolerance areas is expressed as the proportion of the feature points with overlapping positions in the out-of-tolerance area to the total number of feature points in the out-of-tolerance area. The specific execution process is as Figure 2 shown.

[0041] As other embodiments, the above overall deformation AI analysis and local out-of-tolerance AI analysis can also be executed separately.

[0042] The overall deformation AI analysis includes the longitudinal deformation analysis of the body-in-white side panel and the transverse deformation analysis of the body-in-white side panel.

[0043] The longitudinal deformation analysis of the body-in-white side panel specifically includes:

[0044] (a1) Divide the body-in-white side panel into upper and lower regions longitudinally, and respectively count the proportions of the number of feature points with deviations greater than 0 and less than 0 in the upper and lower regions to all feature points in the corresponding regions. If it exceeds the set threshold (the set threshold selected here in this embodiment is 80%), it is preliminarily determined that there is longitudinal deformation, and the process of (b1) is executed;

[0045] (b1) For the plane to be analyzed, determine the top envelope line formed by the feature points in this plane as the highest feature point band, and determine the bottom envelope line formed by the feature points in this plane as the lowest feature point band. For any feature point in the lowest feature point band, take it as the feature point to be compared. Obtain the feature points within a certain width (200 mm is selected in this embodiment) on both the left and right sides of the feature point to be compared that fall within the highest feature point band as the comparison feature points. Calculate one by one whether the deviation data between the comparison feature points and the feature point to be compared is greater than the tolerance band (tolerance band calculation method: 0.75 times the sum of the positive tolerance thresholds of the corresponding two feature points). Count the number of all feature points whose deviation data is greater than the tolerance band, and calculate the proportion of the feature points greater than the tolerance band in the total number of feature points in the highest feature point band and the lowest feature point band. If it is greater than the set threshold (the set threshold here is selected as 80% in this embodiment), it is determined that there is longitudinal deformation.

[0046] The analysis of the lateral deformation of the body-in-white side panel specifically includes:

[0047] (a2) Divide the body-in-white side panel into left and right two regions along the transverse direction. Respectively count the proportion of the number of feature points with deviation greater than 0 and less than 0 in the left and right two regions in all feature points in the corresponding regions. If it exceeds the set threshold (the set threshold here is selected as 80% in this embodiment), it is initially determined that there is lateral deformation, and execute the (b2) process;

[0048] (b2) For the plane to be analyzed, determine the left envelope line formed by the feature points in this plane as the left feature point band, and determine the bottom envelope line formed by the feature points in this plane as the right feature point band. For any feature point in the left feature point band, take it as the feature point to be compared. Obtain the feature points within a certain width (200 mm is selected in this embodiment) above and below the feature point to be compared that fall within the right feature point band as the comparison feature points. Calculate one by one whether the deviation data between the comparison feature points and the feature point to be compared is greater than the tolerance band (tolerance band calculation method: 0.75 times the sum of the positive tolerance thresholds of the corresponding two feature points). Count the number of all feature points whose deviation data is greater than the tolerance band, and calculate the proportion of the feature points greater than the tolerance band in the total number of feature points in the left feature point band and the right feature point band. If it is greater than the set threshold (the set threshold here is selected as 80% in this embodiment), it is determined that there is lateral deformation.

[0049] The local out-of-tolerance AI analysis includes:

[0050] Divide the deviation data in the dataset into two parts according to the positive and negative values, namely the positive deviation set and the negative deviation set;

[0051] Perform density-based clustering on the data in the positive deviation set and the negative deviation set respectively to obtain clustering clusters, and each clustering cluster represents a local out-of-tolerance area.

[0052] In this embodiment, the clustering algorithm uses the DBSCAN clustering algorithm. As Figure 3 shown in the following is the specific execution process of this algorithm, which is as follows:

[0053] Input:

[0054] X: A dataset containing n objects

[0055] ε: Radius parameter

[0056] Output: A set of density-based clusters

[0057] Process:

[0058] Mark all objects as unvisited and execute the following:

[0059] Randomly select an unvisited object p from X. N is the set of objects in the ε neighborhood of p; mark p as visited;

[0060] If p is a core point, create a new cluster C and add p to C. For each point p' in the object set N, if p' is an unvisited point, mark p' as visited. If p' is a core point, add p' to N. If p' is not a member point of any cluster, add p' to C and output C;

[0061] If p is not a core point, mark p as a noise point;

[0062] Loop until there are no unmarked objects in X.

[0063] The overall vehicle size deformation situation and / or local out-of-tolerance areas are visually displayed in the form of a 3D scatter plot. Figure 4 Shown in the following is a schematic diagram of the visual display result after the out-of-tolerance analysis of the present invention. Figure 4 The points in it are the body-in-white feature points, and the area enclosed by the square frame is the local out-of-tolerance area.

[0064] To sum up, the present invention integrates the AI output with the system. The specific AI algorithm principles for the overall deformation analysis and local out-of-tolerance analysis are as described above: The overall deformation analysis first performs data processing to process the original data into data that conforms to the input specifications, and identifies the global dimension assembly deviation according to the rules. The local out-of-tolerance analysis uses DBSCAN to identify the regional dimension assembly deviation and visually displays it in the form of a 3D scatter plot, outputs the deviation warning of a single sample. After performing the deviation warning of a single sample for multiple vehicles, then analyze the deviation of several consecutive samples and output the deviation trend warning of the consecutive samples.

[0065] The above-described embodiments are merely examples and do not represent a limitation on the scope of the present invention. These embodiments can also be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the technical idea of the present invention.

Claims

1. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, characterized in that, The method includes: Obtaining the actual values and theoretical values of the three-dimensional coordinates of the characteristic points of the body-in-white to form a data set of deviations; Performing overall deformation AI analysis based on the deviation data in the data set to determine the overall dimensional deformation of the vehicle; Performing local out-of-tolerance AI analysis based on the deviation data in the data set to determine the local out-of-tolerance areas; The local out-of-tolerance AI analysis includes: Dividing the deviation data in the data set into two parts according to the positive and negative values of the numerical values, namely the positive deviation set and the negative deviation set; Performing density-based clustering on the data in the positive deviation set and the negative deviation set respectively to obtain clustering clusters, and each clustering cluster represents a local out-of-tolerance area.

2. The vehicle size deformation trend extraction and out-of-tolerance warning method according to claim 1, wherein, When the method is executed, overall deformation AI analysis is first performed. If there is overall dimensional deformation of the vehicle, local out-of-tolerance AI analysis is performed. If there is no overall dimensional deformation of the vehicle, the process ends.

3. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, according to claim 1, characterized in that The overall deformation AI analysis includes longitudinal deformation analysis of the side panels of the body-in-white and transverse deformation analysis of the side panels of the body-in-white.

4. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, according to claim 3, characterized in that The longitudinal deformation analysis of the side panels of the body-in-white specifically includes: (a1) Dividing the side panels of the body-in-white longitudinally into upper and lower regions, respectively counting the proportions of the number of characteristic points with deviations greater than 0 and less than 0 in the upper and lower regions among all the characteristic points in the corresponding regions. If it exceeds the set threshold, it is preliminarily determined that there is longitudinal deformation, and the (b1) process is executed; (b1) For the plane to be analyzed, determine the top envelope line formed by the characteristic points in the plane as the highest characteristic point band, and determine the bottom envelope line formed by the characteristic points in the plane as the lowest characteristic point band. For any characteristic point in the lowest characteristic point band, use it as the comparison characteristic point, obtain the characteristic points falling within the highest characteristic point band within a certain width range on the left and right of the comparison characteristic point as the comparison characteristic points, calculate one by one whether the deviation data between the comparison characteristic points and the comparison characteristic point is greater than the tolerance band, count the number of all characteristic points with deviation data greater than the tolerance band, and calculate the proportion of the characteristic points greater than the tolerance band in the total number of characteristic points in the highest characteristic point band and the lowest characteristic point band. If it is greater than the set threshold, it is determined that there is longitudinal deformation.

5. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, according to claim 3, characterized in that The transverse deformation analysis of the side panels of the body-in-white specifically includes: (a2) Dividing the side panels of the body-in-white transversely into left and right regions, respectively counting the proportions of the number of characteristic points with deviations greater than 0 and less than 0 in the left and right regions among all the characteristic points in the corresponding regions. If it exceeds the set threshold, it is preliminarily determined that there is transverse deformation, and the (b2) process is executed; (b2) For the plane to be analyzed, determine the left envelope line formed by the characteristic points in the plane as the left characteristic point band, and determine the right envelope line formed by the characteristic points in the plane as the right characteristic point band. For any characteristic point in the left characteristic point band, use it as the comparison characteristic point, obtain the characteristic points falling within the right characteristic point band within a certain width range above and below the comparison characteristic point as the comparison characteristic points, calculate one by one whether the deviation data between the comparison characteristic points and the comparison characteristic point is greater than the tolerance band, count the number of all characteristic points with deviation data greater than the tolerance band, and calculate the proportion of the characteristic points greater than the tolerance band in the total number of characteristic points in the left characteristic point band and the right characteristic point band. If it is greater than the set threshold, it is determined that there is transverse deformation.

6. The vehicle size deformation trend extraction and out-of-tolerance warning method according to claim 1, characterized in that The method further includes: for n continuously detected vehicles, analyzing the change trends of the overall vehicle size deformation and the local out-of-tolerance regions of the continuous samples, and if the size deformation and the out-of-tolerance regions of the continuous samples tend to be consistent, a warning is given.

7. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, according to claim 6, characterized in that If there is overall vehicle size deformation in all the continuous samples, the coincidence degree of the out-of-tolerance regions of the continuous samples is determined, and if the coincidence degree is greater than a set value, a warning is given.

8. A method for extracting the deformation trend of vehicle dimensions and early warning of out-of-tolerance, according to claim 6, characterized in that, The coincidence degree of the out-of-tolerance regions is expressed as the proportion of the feature points with coincident positions in the out-of-tolerance regions to the total number of feature points in the out-of-tolerance regions.

9. A method for extracting the vehicle size deformation trend and early warning of out-of-tolerance, according to claim 1, characterized in that The overall vehicle size deformation situation and / or the local out-of-tolerance regions are visually displayed in the form of a 3D scatter plot.

Images