Sealing strip detection method and system and storage medium
The convolutional neural network detection model automatically detects the assembly quality of door seal strips, solving the problems of error detection and low efficiency of manual detection, realizing high-precision and high-efficiency seal strip detection, and improving production efficiency.
Patent Information
- Application Number
- CN202410960959.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-12-03
AI Technical Summary
In the prior art, there are errors in the assembly quality of door seal strips, which are slow, low efficiency and difficult, resulting in abnormal seal strips not being discovered in time and affecting production efficiency.
The convolutional neural network detection model is adopted to collect assembly images and vehicle configuration information on the vehicle door, automatically detect the assembly quality of the seal strip, output qualified or unqualified results, and determine the abnormal type, count the abnormal probability, and realize automatic detection.
It improves the accuracy and efficiency of seal strip assembly quality inspection, reduces manual detection errors, accurately locates abnormal problems, improves production efficiency, and reduces personnel costs.
Smart Images

Figure CN119067147A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of vehicle technology, and in particular to a sealing strip detection method, system and storage medium. Background Art
[0002] The main functions of car door sealing strips are shock absorption, waterproofing, dustproofing, sound insulation, decoration, rain diversion, etc.
[0003] The door sealing strip is critical to the sealing of the car. If the sealing strip is assembled abnormally, it will cause rainwater to enter the car, slow air conditioning cooling, dust to enter the door hinges and door locks, and air leakage from the door will occur, which will cause increased wind noise and bring various inconveniences to driving and riding. Therefore, the assembly quality of the door sealing strip is particularly important.
[0004] Currently, the quality of sealing strip assembly is usually inspected manually. However, manual inspection may lead to errors or omissions, and manual inspection is slow, inefficient, and difficult. Summary of the invention
[0005] In view of this, the present invention aims to propose a sealing strip detection method, system and storage medium, which at least solves the technical problems in the prior art of manually inspecting the sealing strip assembly quality, such as false detection and missed detection, slow manual inspection speed, low inspection efficiency and great inspection difficulty.
[0006] According to a first aspect of the present invention, a sealing strip detection method is provided, comprising:
[0007] Collecting an assembly image of a sealing strip to be inspected assembled on a vehicle door, and acquiring vehicle door configuration information corresponding to the vehicle door, wherein the vehicle door configuration information includes a vehicle identification code and vehicle model information;
[0008] Acquire standard configuration information corresponding to the vehicle identification code, and acquire a pre-established detection model corresponding to the vehicle model information;
[0009] Using the detection model to detect the assembly image according to the standard configuration information, and outputting a detection result, wherein the detection result includes qualified and unqualified;
[0010] If the detection result is unqualified, determining the abnormality type through the detection model;
[0011] Obtain the total number of times the test results are unqualified within the first preset time, and obtain the first number corresponding to each abnormal type within the first preset time;
[0012] The probability of each abnormal type is determined based on the first number corresponding to each abnormal type and the total number of times the test result is unqualified.
[0013] Optionally, if the detection result is unqualified, after determining the abnormality type through the detection model, the method further includes:
[0014] Record the occurrence time of each abnormal type, count the time interval between the occurrence time of the current abnormal type and the occurrence time of the previous abnormal type of the same type, and if a first preset number of consecutive time intervals are less than the preset interval, issue a first level alarm;
[0015] Alternatively, a second number corresponding to each abnormal type within a second preset time is recorded, and if the second number is greater than or equal to the preset number, a first level alarm is issued, wherein the second preset time is less than the first preset time.
[0016] Optionally, the standard configuration information includes a standard image, a standard sealing strip color and / or a standard door interior color;
[0017] The step of using the detection model to detect the assembly image according to the standard configuration information and outputting the detection result includes:
[0018] Using the detection model to identify the actual color of the sealing strip corresponding to the assembly image;
[0019] Using the detection model to determine whether the actual color of the sealing strip is consistent with the standard sealing strip color, and / or to determine whether the actual color of the sealing strip corresponds to the standard door interior color;
[0020] If the actual color of the sealing strip is consistent with the color of the standard sealing strip, and / or the actual color of the sealing strip corresponds to the color of the standard door interior trim, then using the detection model to compare the assembly image with the standard image;
[0021] If the assembly image is consistent with the standard image, a qualified detection result is output.
[0022] Optionally, the process of establishing the detection model includes:
[0023] Collecting abnormal image groups corresponding to different vehicle models, wherein the abnormal image group includes a plurality of abnormal images, and the plurality of abnormal images correspond to at least two abnormal types;
[0024] Through convolutional neural networks, deep learning is performed on abnormal image groups corresponding to different models to establish detection models corresponding to different models;
[0025] If the test result is unqualified, determining the abnormality type according to the detection model includes:
[0026] If the detection result is unqualified, the assembly image is compared with the abnormal image group corresponding to the vehicle model information through the detection model to determine the abnormal type corresponding to the abnormal image corresponding to the assembly image.
[0027] Optionally, before collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door, the method further includes:
[0028] Receiving an in-position signal sent by an in-position sensor, wherein the in-position sensor is used to send an in-position signal when the vehicle door reaches the detection station;
[0029] The obtaining of the door configuration information corresponding to the door includes:
[0030] The RFID reader / writer is controlled to obtain the door configuration information corresponding to the door, wherein the RFID reader / writer obtains the door configuration information by reading an RFID tag set on a door hanger, and the door hanger is used to transport the door.
[0031] Optionally, the abnormality type is divided into a missing abnormality type, a gap abnormality type, an extrusion abnormality type, a color abnormality type, and a material abnormality type;
[0032] The obtaining of the first number corresponding to each abnormal type within the first preset time includes:
[0033] Obtain the first number corresponding to the missing abnormal type, the first number corresponding to the gap abnormal type, the first number corresponding to the extrusion abnormal type, the first number corresponding to the color abnormal type, and the first number corresponding to the material abnormal type within the first preset time;
[0034] Determining the probability of each abnormal type according to the first number corresponding to each abnormal type and the total number of times the test result is unqualified includes:
[0035] The ratio of the first number corresponding to the missing abnormality type to the total number of unqualified test results is taken as the probability of the missing abnormality type, the ratio of the first number corresponding to the gap abnormality type to the total number of unqualified test results is taken as the probability of the gap abnormality type, the ratio of the first number corresponding to the extrusion abnormality type to the total number of unqualified test results is taken as the probability of the extrusion abnormality type, the ratio of the first number corresponding to the color abnormality type to the total number of unqualified test results is taken as the probability of the color abnormality type, and the ratio of the first number corresponding to the material abnormality type to the total number of unqualified test results is taken as the material abnormality type.
[0036] Optionally, after collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door, the method further includes:
[0037] Acquire a collection time corresponding to the assembly image;
[0038] After the detection model is used to detect the assembly image according to the standard configuration information and the detection result is output, the method further includes:
[0039] storing the vehicle identification code, vehicle model information, collection time, and detection results corresponding to the assembly image;
[0040] If the detection result is unqualified, after determining the abnormality type through the detection model, the method further includes:
[0041] If the detection result is unqualified, a second level alarm is issued, and / or the production line is stopped through the manufacturing system, and / or alarm information is obtained and sent to the mobile terminal.
[0042] According to a second aspect of the present invention, a sealing strip detection system is also provided, comprising:
[0043] processor;
[0044] a memory for storing instructions executable by the processor;
[0045] Wherein, the processor is configured to execute the instructions to implement the sealing strip detection method as described above.
[0046] Optionally, it further comprises a door hanger and a detection frame, wherein the door hanger is used to transport the door, an RFID tag is provided on the door hanger, and the detection frame is used to install the camera assembly and the in-place sensor;
[0047] The detection frame includes two frames spaced apart along a first direction, wherein the first direction is perpendicular to the transportation direction of the vehicle door, and the two frames form a detection station.
[0048] According to a third aspect of the present invention, a readable storage medium is further provided, wherein a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the steps of the sealing strip detection method described above are implemented.
[0049] In the embodiment of the present invention, the assembly quality of the sealing strip to be inspected assembled on the door can be automatically inspected, which can replace manual inspection, avoid the situation of misdetection and missed inspection in the manual inspection process, improve the inspection accuracy, reduce personnel configuration, thereby reducing personnel costs, and compared with manual inspection, the inspection method provided by the embodiment of the present invention has a fast inspection cycle, which can improve the inspection efficiency of the assembly quality of the sealing strip, thereby improving production efficiency. In addition, in the embodiment of the present invention, the inspection model is used to inspect the assembly image according to the standard configuration information to output the inspection result, thereby avoiding the problem of high inspection difficulty in the manual inspection process. In addition, by obtaining the vehicle identification code, the output inspection result can correspond to the vehicle identification code, thereby accurately locating the assembly problem point of the sealing strip in each vehicle. In addition, when the inspection result is unqualified, the abnormal type can also be determined, thereby accurately locating the specific assembly problem of the sealing strip. In addition, the probability of each abnormal type can be automatically counted, so as to focus on solving the assembly problem of the sealing strip corresponding to the abnormal type with a larger probability. In addition, due to the improved inspection accuracy, it can be avoided that the production and installation of the subsequent process is affected by the failure to detect the abnormality of the sealing strip, thereby affecting the production efficiency.
[0050] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0052] Figure 1 is a flow chart of steps of a sealing strip detection method provided by an embodiment of the present invention;
[0053] Figure 2 is a flowchart of another sealing strip detection method provided by an embodiment of the present invention;
[0054] Figure 3 is a practical application flow chart of the sealing strip detection method provided by an embodiment of the present invention;
[0055] Figure 4 is a schematic diagram of the three-dimensional structure of a sealing strip detection system provided by an embodiment of the present invention;
[0056] Figure 5is a front view structural schematic diagram of a sealing strip detection system provided by an embodiment of the present invention;
[0057] Figure 6 is a side structural schematic diagram of a sealing strip detection system provided by an embodiment of the present invention;
[0058] Figure 7 1 is a schematic diagram of the top structure of a sealing strip detection system provided by an embodiment of the present invention.
[0059] Reference numerals:
[0060] 1-car door, 2-car door hanger, 3-detection frame, 31-frame, 4-cabinet, 5-arrival sensor, 6-camera assembly. DETAILED DESCRIPTION
[0061] To make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the embodiments of the present invention, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can be implemented. The division of the following embodiments is for the convenience of description, and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined and referenced with each other without contradiction.
[0062] During the assembly process of the door sealing strip, there may be problems such as missing sealing strips, left and right squeezing of sealing strips, wrong color of sealing strips, abnormal sealing strip materials, etc. The assembly gap requirement of the sealing strip is usually 0mm-3mm. Exceeding the above assembly gap requirement will cause air leakage of the sealing strip and abnormal sound insulation effect. At present, the assembly quality of the sealing strip is usually inspected manually. During normal production, the door hanger drives the door forward at a constant speed, not stationary, and the inspection standards for different models are different, which greatly increases the difficulty of manual inspection. In addition, the assembly gap requirement of the door sealing strip is usually between 0mm-3mm. It is difficult to identify the gap distance at the joint of the sealing strip by the naked eye, and manual inspection is difficult. In summary, there are cases of misdetection and missed detection in manual inspection, and the manual inspection speed is slow, the inspection efficiency is low, and the inspection difficulty is high. In addition, if the abnormality of the sealing strip is not detected, it will affect the production and installation of subsequent workstations, resulting in rework and other situations, delays in work and time, and thus affecting production efficiency.
[0063] In order to solve the above problems, an embodiment of the present invention provides a sealing strip detection method, system and storage medium. The above-mentioned sealing strip detection method, system and storage medium are described in detail below.
[0064] Reference Figure 1, shows a flow chart of steps of a sealing strip detection method provided by an embodiment of the present invention, the sealing strip detection method can be applied to a processor in a sealing strip detection system, and the method may include:
[0065] Step 101 : collecting an assembly image of a sealing strip to be inspected assembled on a vehicle door, and acquiring vehicle door configuration information corresponding to the vehicle door.
[0066] After the sealing strip to be inspected is assembled on the vehicle door 1, the vehicle door hanger 2 will transport the vehicle door 1 equipped with the sealing strip to be inspected to the next process. The assembly image can be collected during the process of the vehicle door hanger 2 transporting the vehicle door 1 equipped with the sealing strip to be inspected. When collecting the assembly image, the camera assembly 6 in the sealing strip inspection system can be used to take a picture to collect the assembly image. Collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door may include: controlling the camera assembly 6 to take a picture, and obtaining the assembly image of the sealing strip to be inspected assembled on the vehicle door collected by the camera assembly 6.
[0067] The camera assembly 6 can take a picture of a sealing strip to be inspected installed on a vehicle door. Figure 7 The sealing strip detection system may include two camera assemblies 6, which are respectively mounted on two frames 31 in the sealing strip detection system. The two camera assemblies 6 can respectively take pictures of the sealing strips to be detected assembled on the two doors on the left and right sides of the door hanger 2. The camera assembly 6 may include a camera, a light source, etc. The camera is used to take pictures, and the light source is used to fill in the light when the camera takes pictures, so that the pictures taken are clearer. The sealing strip detection system also includes a cabinet 4, and the cabinet 4 includes an industrial computer. The assembly image can be stored in the industrial computer according to the acquisition time and the vehicle identification code. In step 101, China can obtain the assembly image from the industrial computer.
[0068] The door configuration information includes the vehicle identification number (VIN) and vehicle model information. VIN is a unique identification code for each vehicle. Through VIN, the relevant configuration information of the vehicle can be obtained, that is, the corresponding standard configuration information can be obtained according to VIN. The collected assembly image corresponds to the vehicle identification number and vehicle model information in the door configuration information.
[0069] Each vehicle includes a plurality of doors, for example, each vehicle includes four doors. The door hanger 2 can specifically transport four doors corresponding to one vehicle at the same time. One assembly image preferably corresponds to a sealing strip to be inspected assembled on one door, so each vehicle corresponds to four assembly images, that is, one vehicle identification code corresponds to four assembly images, and the vehicle identification codes and vehicle model information corresponding to the four assembly images are the same.
[0070] Step 102, obtaining standard configuration information corresponding to the vehicle identification code, and obtaining a pre-established detection model corresponding to the vehicle model information.
[0071] Among them, different vehicle models correspond to different detection models. According to the vehicle model information, a pre-established detection model corresponding to the vehicle model information can be obtained, so as to perform corresponding detection on different vehicle models. Different detection models can be switched according to different vehicle models. Vehicle model information may include EV (Electric Vehicle), EVR (Electric Vehicle Range Extender), etc. The detection model may be a convolutional neural network model (CNN model), which may extract image features using convolution operations and input them into a multi-layer neural network. Through the learning and training process, it may learn to classify, locate, and detect images. Standard configuration information may include standard images, standard sealing strip colors, and / or standard door interior colors. Standard configuration information is a standard for judging whether the assembly quality of the sealing strip to be tested is qualified.
[0072] Acquiring the standard configuration information corresponding to the vehicle identification code may include: sending the vehicle identification code to the production and manufacturing system (MOM system), and receiving the standard configuration information corresponding to the vehicle identification code sent by the production and manufacturing system. After receiving the vehicle identification code, the production and manufacturing system will search for the standard configuration information corresponding to the vehicle identification code, and send the standard configuration information corresponding to the vehicle identification code to the processor in the sealing strip detection system. The processor in the sealing strip detection system can be connected to the production and manufacturing system through an industrial computer, and the industrial computer can be connected to the production and manufacturing system through Ethernet for communication.
[0073] Step 103: Use the detection model to detect the assembly image according to the standard configuration information and output the detection result.
[0074] The test results include qualified and unqualified. The test results can also correspond to the vehicle identification code. The assembly image is input into the detection model, and the detection model detects the assembly image according to the standard configuration information and outputs the test results. The sealing strip detection system also includes a display. After the test results are output, the test results can also be sent to the display to display the test results on the display.
[0075] Step 104: If the detection result is unqualified, the abnormality type is determined through the detection model.
[0076] Among them, the abnormal types can be divided into missing assembly abnormalities, gap abnormalities, extrusion abnormalities, color abnormalities, material abnormalities, etc. The detection model is used to detect the assembly image according to the standard configuration information. If the output detection result is unqualified, it is necessary to further determine the abnormal type corresponding to the assembly image, so as to accurately locate the specific assembly problem of the sealing strip.
[0077] Different types of abnormalities correspond to different sealing strip assembly problems. For example, the sealing strip assembly problem corresponding to the missing assembly abnormality is the sealing strip missing assembly, the sealing strip assembly problem corresponding to the gap abnormality is the sealing strip gap being too large, the sealing strip assembly problem corresponding to the extrusion abnormality is the sealing strip extrusion, the sealing strip assembly problem corresponding to the color abnormality is the wrong color of the sealing strip, and the sealing strip assembly problem corresponding to the material abnormality is the existence of other materials on the sealing strip. Among them, when the sealing strip gap is too large, the sealing strip gap is specifically larger than the sealing strip assembly gap requirement, and the sealing strip gap specifically refers to the gap at the sealing strip joint. For example, when the color of the sealing strip is black, if other colors are detected on the sealing strip, such as yellow or white paint spots on the sealing strip, it is determined that other materials exist on the sealing strip.
[0078] Step 105, obtaining the total number of times the detection result is unqualified within the first preset time, and obtaining the first number corresponding to each abnormal type within the first preset time.
[0079] The first preset time can be set according to actual needs, such as one week, one month, one quarter, etc.
[0080] Step 106, determining the probability of each abnormal type according to the first number corresponding to each abnormal type and the total number of times the detection result is unqualified.
[0081] The probability of each abnormal type is the ratio of the first number corresponding to each abnormal type to the total number of times the test result is unqualified, and the probability of each abnormal type is between 0 and 1. After determining the probability of each abnormal type, the probability of each abnormal type can also be displayed through the display in the sealing strip detection system. In this step, the probability of each abnormal type can be automatically counted and displayed, so as to focus on solving the sealing strip assembly problem corresponding to the abnormal type with a higher probability.
[0082] In the embodiment of the present invention, the assembly quality of the sealing strip to be inspected assembled on the vehicle door can be automatically inspected, which can replace manual inspection, avoid the situation of false inspection and missed inspection in the manual inspection process, improve the inspection accuracy, reduce the staffing, thereby reducing the staff cost, and compared with manual inspection, the inspection method provided by the embodiment of the present invention has a fast inspection cycle, which can improve the inspection efficiency of the assembly quality of the sealing strip, thereby improving the production efficiency. In addition, in the embodiment of the present invention, the inspection model is used to inspect the assembly image according to the standard configuration information to output the inspection result, thereby avoiding the problem of high inspection difficulty in the manual inspection process. In addition, by obtaining the vehicle identification code, the output inspection result can correspond to the vehicle identification code, thereby accurately locating the assembly problem point of the sealing strip in each vehicle. In addition, due to the improved inspection accuracy, it can be avoided that the production and installation of the subsequent process is affected by the failure to detect the abnormality of the sealing strip, thereby affecting the production efficiency.
[0083] Reference Figure 2 In a preferred embodiment of the present invention, in step 104, if the detection result is unqualified, after determining the abnormality type through the detection model, it also includes:
[0084] Step 107, record the occurrence time corresponding to each abnormal type, count the time interval between the occurrence time corresponding to the current abnormal type and the occurrence time corresponding to the previous abnormal type of the same type, if a first preset number of time intervals are consecutively less than the preset interval, then issue a first level alarm; or, record the second number corresponding to each abnormal type within the second preset time, if the second number is greater than or equal to the preset number, then issue a first level alarm.
[0085] Among them, the occurrence time corresponding to each abnormal type can be the time when the abnormal type is determined in step 104. The first preset number can be set according to actual needs, such as 4, 5, 6, etc. The preset interval can be set according to actual needs, such as 10 minutes, 30 minutes, etc. If the first preset number of time intervals are all less than the preset interval, there are multiple first number of this abnormal type occurring in a short interval. At this time, a first level alarm is required to determine the corresponding production problem for the frequently occurring abnormal type and solve the problem to avoid the subsequent occurrence of this abnormal type, thereby minimizing losses. The first level alarm has a higher alarm level. The first level alarm can be a buzzer that lasts for a first set time, a red light that lasts for a second set time, and is prominently displayed on the display.
[0086] The second preset time is less than the first preset time, and the second preset time can be one day, two days, etc. The preset number can be set according to actual needs, such as 6, 8, 10, etc. If the second number corresponding to each abnormal type within the second preset time is greater than or equal to the preset number, it means that the abnormal type occurs multiple times within the second preset time. At this time, a first-level alarm is required to determine the corresponding production problem for the frequently occurring abnormal type and solve the problem to avoid the subsequent occurrence of the abnormal type, thereby minimizing the loss as much as possible.
[0087] In a preferred embodiment of the present invention, the abnormality types are divided into missing abnormality types, gap abnormality types, extrusion abnormality types, color abnormality types, and material abnormality types. Obtaining the first number corresponding to each abnormality type within the first preset time includes: obtaining the first number corresponding to the missing abnormality type, the first number corresponding to the gap abnormality type, the first number corresponding to the extrusion abnormality type, the first number corresponding to the color abnormality type, and the first number corresponding to the material abnormality type within the first preset time.
[0088] The probability of each abnormal type is determined according to the first number corresponding to each abnormal type and the total number of times the test result is unqualified, including: taking the ratio of the first number corresponding to the missing abnormal type to the total number of times the test result is unqualified as the probability of the missing abnormal type, taking the ratio of the first number corresponding to the gap abnormal type to the total number of times the test result is unqualified as the probability of the gap abnormal type, taking the ratio of the first number corresponding to the extrusion abnormal type to the total number of times the test result is unqualified as the probability of the extrusion abnormal type, taking the ratio of the first number corresponding to the color abnormal type to the total number of times the test result is unqualified as the probability of the color abnormal type, and taking the ratio of the first number corresponding to the material abnormal type to the total number of times the test result is unqualified as the material abnormal type.
[0089] In a preferred embodiment of the present invention, the standard configuration information includes a standard image, a standard sealing strip color and / or a standard door interior color. The detection model is used to detect the assembly image according to the standard configuration information, and the detection result is output, including: using the detection model to identify the actual color of the sealing strip corresponding to the assembly image; using the detection model to determine whether the actual color of the sealing strip is consistent with the standard sealing strip color, and / or, determining whether the actual color of the sealing strip corresponds to the standard door interior color; if the actual color of the sealing strip is consistent with the standard sealing strip color, and / or, the actual color of the sealing strip corresponds to the standard door interior color, then using the detection model to compare the assembly image with the standard image; if the assembly image is consistent with the standard image, then outputting a qualified detection result.
[0090] Specifically, if the actual color of the sealing strip is inconsistent with the standard sealing strip color, and / or the actual color of the sealing strip does not correspond to the standard door interior color, an unqualified test result is output. If the assembly image does not match the standard image, an unqualified test result is output. The standard image is an image of a sealing strip assembled on a door with qualified assembly quality corresponding to the vehicle model information. The standard sealing strip color is the color required for the sealing strip, and the standard sealing strip color can be black, brown, etc. The standard door interior color is the color of the door interior corresponding to the vehicle model information. The color of the sealing strip should also correspond to the standard door interior color. The correspondence between the standard door interior color and the color of the sealing strip can be pre-stored to determine whether the actual color of the sealing strip corresponds to the standard door interior color. The actual color of the sealing strip is the color of the sealing strip in the current assembly image. In an embodiment of the present invention, the test result can be determined based on multiple standards to improve the accuracy of the test result.
[0091] In a preferred embodiment of the present invention, the process of establishing the detection model includes: collecting abnormal image groups corresponding to different vehicle models; performing deep learning on the abnormal image groups corresponding to different vehicle models through convolutional neural networks to establish detection models corresponding to different vehicle models. The abnormal image group includes multiple abnormal images, and the multiple abnormal images correspond to at least two abnormal types. Step 104, if the detection result is unqualified, the abnormal type is determined according to the detection model, including: if the detection result is unqualified, the assembly image is compared with the abnormal image group corresponding to the vehicle model information through the detection model to determine the abnormal type corresponding to the abnormal image corresponding to the assembly image.
[0092] Among them, each abnormal image corresponds to an abnormal type, and the abnormal type may include missing assembly abnormality, gap abnormality, extrusion abnormality, color abnormality, material abnormality, etc. The number of abnormal images included in the abnormal image group is greater than or equal to the number of abnormal types. After the convolutional neural network performs deep learning on the abnormal image groups corresponding to different vehicle models, the abnormal image corresponding to the assembly image can be determined according to the assembly image, and then the abnormal type corresponding to the abnormal image can be determined. In the embodiment of the present invention, by performing deep learning on the abnormal image groups corresponding to different vehicle models, the detection model can identify different abnormal types with high recognition accuracy.
[0093] The detection model is preferably a convolutional neural network model, which may include an input layer, a convolutional layer, a pooling layer, and a fully connected layer. The input layer takes the assembly image as input and converts it into a suitable feature map, including image resizing, color space conversion, normalization or standardization. The convolutional layer performs a convolution operation on the image by using a set of learnable convolutional kernels (also called filters or filters), and the function of the convolutional kernel is to extract feature information in the image. The convolution operation can be controlled by parameters such as different strides, padding methods, and convolutional kernel sizes. Activation layer: The convolutional layer is usually followed by an activation function, such as the ReLU function, to perform a nonlinear transformation on the output of the convolutional layer. Pooling layer: It is used to reduce the spatial size of the feature map and reduce the amount of calculation. Common pooling methods include maximum pooling and average pooling. Repeat the above convolutional layer, activation layer, and pooling layer operations to extract more feature information. Fully connected layer: The feature information extracted by the convolutional layer and the pooling layer is input into the fully connected layer for classification. The fully connected layer is usually composed of multiple neurons, and the output value of each neuron represents the probability that the input belongs to that category. Output layer: The last layer is usually the Softmax layer, which normalizes the output probability value, obtains the probability distribution of each state of the sealing strip, and outputs the detection result.
[0094] Reference Figure 2 In a preferred embodiment of the present invention, before collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door in step 101, the method further includes: step 108, receiving an in-position signal sent by an in-position sensor. Acquiring the vehicle door configuration information corresponding to the vehicle door in step 101 includes: controlling an RFID (Radio-Frequency Identification) reader to acquire the vehicle door configuration information corresponding to the vehicle door, wherein the RFID reader acquires the vehicle door configuration information by reading an RFID tag set on a vehicle door hanger, and the vehicle door hanger is used to transport the vehicle door.
[0095] Specifically, the sealing strip detection system also includes a PLC (Programmable Logic Controller) system. The in-place sensor 5 is used to send an in-place signal when the vehicle door arrives at the detection station. The in-place sensor 5 can be a diffuse reflection sensor, and the diffuse reflection sensor can be electrically connected to the PLC system. The number of the in-place sensors 5 can be two, four, etc. The number of the in-place sensors 5 is preferably four, and the four in-place sensors 5 are respectively installed on the two frames 31.
[0096] The two frames 31 enclose the inspection station. When the door hanger 2 transports four doors, it will first carry two of the doors to the inspection station, and then carry the other two doors to the inspection station. After the door hanger 2 carries the two doors to the inspection station, the in-place sensor 5 is triggered, and the in-place sensor 5 sends an in-place signal to the processor and / or PLC system. After the processor receives the in-place signal sent by the in-place sensor, it controls the camera assembly 6 to work through the PLC system, so that the camera in the camera assembly 6 takes pictures.
[0097] The RFID tag is set on the door hanger 2, and the door configuration information is stored in the RFID tag. After the processor receives the in-place signal sent by the in-place sensor, it will control the RFID reader through the PLC system to read the door configuration information stored in the RFID tag. The door configuration information may include the vehicle identification code and vehicle model information. After the RFID reader is finished reading, the door configuration information will be sent to the processor. In the embodiment of the present invention, the door configuration information can be automatically read by the RFID reader without manual intervention, and the recognition efficiency and accuracy are high.
[0098] Reference Figure 2 In a preferred embodiment of the present invention, after collecting the assembly image of the sealing strip to be inspected on the vehicle door in step 101, it also includes: step 109, obtaining the acquisition time corresponding to the assembly image. Step 103, using the detection model to detect the assembly image according to the standard configuration information, after outputting the detection result, it also includes: step 110, storing the vehicle identification code, vehicle model information, acquisition time and detection result corresponding to the assembly image; step 111, if the detection result is unqualified, then a second level alarm is issued, and / or the production line is stopped through the manufacturing system, and / or the alarm information is obtained and sent to the mobile terminal.
[0099] The acquisition time can be accurate to the year, month and day. The vehicle identification code, vehicle model information, acquisition time and test results corresponding to the assembly image can be uploaded to the Internet of Things for storage. In the embodiment of the present invention, by storing the vehicle identification code, vehicle model information, acquisition time and test results corresponding to the assembly image, the test status of the assembly quality of the sealing strip is recorded, so that the traceability function can be realized, which is convenient for later querying the test status of the assembly quality of the sealing strip on the door corresponding to each vehicle.
[0100] The warning level of the second level alarm is lower than that of the first level alarm. The warning mode of the second level alarm can be a light alarm, and / or a buzzer alarm, and / or a voice alarm. After the processor outputs the test result, the test result can also be transmitted to the MOM system and the PLC system. The PLC system includes a three-color light and a buzzer. The three-color light includes a green light and a red light. If the test result is qualified, the green light will be on for 3 seconds. If the test result is unqualified, the red light will flash, and the buzzer will alarm to remind the production personnel to deal with abnormal situations. When the test result is unqualified, the MOM system will control the production line to stop, and it will be released after manual processing is completed. The display can display the alarm information, the voice broadcast system can broadcast the alarm information synchronously, and the corresponding alarm information can also be pushed to the corresponding maintenance personnel through the mobile terminal so that the maintenance personnel can handle it in time.
[0101] In the embodiment of the present invention, through alarm and real-time production line control, the current sealing strip detection problem can be handled in time, and the rework time invested in the subsequent workstations can be reduced, thereby improving production efficiency.
[0102] In order to facilitate those skilled in the art to fully understand the above sealing strip detection method, refer to Figure 3 , Figure 3 This is a practical application flow chart of the sealing strip detection method provided by an embodiment of the present invention, which specifically includes:
[0103] S1, the vehicle door reaches the inspection station; S2, reads the vehicle identification code and vehicle model information through the RFID reader; S3, obtains the corresponding standard configuration information according to the vehicle identification code; S4, triggers the camera assembly to take pictures through the in-place sensor; S5, obtains the assembly image of the sealing strip to be inspected; S6, obtains the corresponding inspection model according to the vehicle model information; S7, uses the inspection model to inspect the assembly image and outputs the inspection result; S8, if the inspection result is qualified, the vehicle door is released; S9, if the inspection result is unqualified, an alarm is issued and the production line is stopped.
[0104] An embodiment of the present invention also provides a sealing strip detection system, including a processor and a memory, the memory being used to store instructions executable by the processor; the processor being used to execute the instructions stored in the memory to implement the following steps: collecting an assembly image of the sealing strip to be detected assembled on a vehicle door, and obtaining vehicle door configuration information corresponding to the vehicle door, wherein the vehicle door configuration information includes a vehicle identification code and vehicle model information; obtaining standard configuration information corresponding to the vehicle identification code, and obtaining a pre-established detection model corresponding to the vehicle model information; using the detection model to detect the assembly image according to the standard configuration information, and outputting the detection result, wherein the detection result includes qualified and unqualified; if the detection result is unqualified, determining the abnormality type through the detection model.
[0105] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.
[0106] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0107] Reference Figures 4 to 7 The sealing strip detection system also includes a door hanger 2 and a detection frame 3. The door hanger 2 is used to transport the door 1. The door hanger 2 is provided with an RFID tag. The detection frame 3 is used to install a camera assembly 6 and an in-place sensor 5. The detection frame 3 includes two frames 31 spaced apart along a first direction, the first direction is perpendicular to the transportation direction of the door, and the two frames 31 form a detection station.
[0108] The door hanger 2 can specifically transport four doors corresponding to a vehicle at the same time. The sealing strip detection system can include two camera assemblies 6, which are respectively mounted on two frames 31 in the sealing strip detection system. The two camera assemblies 6 can respectively take pictures of the sealing strips to be detected installed on the two doors on the left and right sides of the door hanger 2.
[0109] The sealing strip detection system further comprises a cabinet 4, which comprises an industrial computer, a PLC system, a display, etc. The in-place sensor 5 may be a diffuse reflection sensor, which may be electrically connected to the PLC system.
[0110] The sealing strip detection system also includes an RFID reader / writer. The door hanger 2 is provided with an RFID tag. The RFID tag stores the door configuration information. The RFID reader / writer can read the door configuration information stored in the RFID tag.
[0111] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions. When the instructions are executed on a computer, the computer executes the sealing strip detection method described in any one of the above embodiments.
[0112] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integration. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.
[0113] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0114] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0115] The above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims
1. A sealing strip detection method, characterized in that: include: Collecting an assembly image of a sealing strip to be inspected assembled on a vehicle door, and acquiring vehicle door configuration information corresponding to the vehicle door, wherein the vehicle door configuration information includes a vehicle identification code and vehicle model information; Acquire standard configuration information corresponding to the vehicle identification code, and acquire a pre-established detection model corresponding to the vehicle model information; Using the detection model to detect the assembly image according to the standard configuration information, and outputting a detection result, wherein the detection result includes qualified and unqualified; If the detection result is unqualified, determining the abnormality type through the detection model; Obtain the total number of times the test results are unqualified within the first preset time, and obtain the first number corresponding to each abnormal type within the first preset time; The probability of each abnormal type is determined based on the first number corresponding to each abnormal type and the total number of times the test result is unqualified.
2. The sealing strip detection method according to claim 1, characterized in that: If the detection result is unqualified, after determining the abnormality type through the detection model, the method further includes: Record the occurrence time corresponding to each abnormal type, and count the time interval between the occurrence time corresponding to the current abnormal type and the occurrence time corresponding to the previous abnormal type of the same type. If a first preset number of consecutive time intervals are less than the preset interval, a first level alarm is issued; or, record the second number corresponding to each abnormal type within a second preset time. If the second number is greater than or equal to the preset number, a first level alarm is issued, wherein the second preset time is less than the first preset time.
3. The sealing strip detection method according to claim 1 or 2, characterized in that: The standard configuration information includes a standard image, a standard sealing strip color and / or a standard door interior color; The step of using the detection model to detect the assembly image according to the standard configuration information and outputting the detection result includes: Using the detection model to identify the actual color of the sealing strip corresponding to the assembly image; Using the detection model to determine whether the actual color of the sealing strip is consistent with the standard sealing strip color, and / or to determine whether the actual color of the sealing strip corresponds to the standard door interior color; If the actual color of the sealing strip is consistent with the color of the standard sealing strip, and / or the actual color of the sealing strip corresponds to the color of the standard door interior trim, then using the detection model to compare the assembly image with the standard image; If the assembly image is consistent with the standard image, a qualified detection result is output.
4. The sealing strip detection method according to claim 1, characterized in that: The process of establishing the detection model includes: Collecting abnormal image groups corresponding to different vehicle models, wherein the abnormal image group includes a plurality of abnormal images, and the plurality of abnormal images correspond to at least two abnormal types; Through convolutional neural networks, deep learning is performed on abnormal image groups corresponding to different models to establish detection models corresponding to different models; If the test result is unqualified, determining the abnormality type according to the detection model includes: If the detection result is unqualified, the assembly image is compared with the abnormal image group corresponding to the vehicle model information through the detection model to determine the abnormal type corresponding to the abnormal image corresponding to the assembly image.
5. The sealing strip detection method according to claim 1, characterized in that: Before collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door, the method further includes: Receiving an in-position signal sent by an in-position sensor, wherein the in-position sensor is used to send an in-position signal when the vehicle door reaches the detection station; The obtaining of the door configuration information corresponding to the door includes: The RFID reader / writer is controlled to obtain the door configuration information corresponding to the door, wherein the RFID reader / writer obtains the door configuration information by reading an RFID tag set on a door hanger, and the door hanger is used to transport the door.
6. The sealing strip detection method according to claim 1, characterized in that: The abnormal types are divided into missing abnormal type, gap abnormal type, extrusion abnormal type, color abnormal type, and material abnormal type; The obtaining of the first number corresponding to each abnormal type within the first preset time includes: Obtain the first number corresponding to the missing abnormal type, the first number corresponding to the gap abnormal type, the first number corresponding to the extrusion abnormal type, the first number corresponding to the color abnormal type, and the first number corresponding to the material abnormal type within the first preset time; Determining the probability of each abnormal type according to the first number corresponding to each abnormal type and the total number of times the test result is unqualified includes: The ratio of the first number corresponding to the missing abnormality type to the total number of unqualified test results is taken as the probability of the missing abnormality type, the ratio of the first number corresponding to the gap abnormality type to the total number of unqualified test results is taken as the probability of the gap abnormality type, the ratio of the first number corresponding to the extrusion abnormality type to the total number of unqualified test results is taken as the probability of the extrusion abnormality type, the ratio of the first number corresponding to the color abnormality type to the total number of unqualified test results is taken as the probability of the color abnormality type, and the ratio of the first number corresponding to the material abnormality type to the total number of unqualified test results is taken as the material abnormality type.
7. The sealing strip detection method according to claim 1, characterized in that: After collecting the assembly image of the sealing strip to be inspected assembled on the vehicle door, the method further includes: Acquire a collection time corresponding to the assembly image; After the detection model is used to detect the assembly image according to the standard configuration information and the detection result is output, the method further includes: storing the vehicle identification code, vehicle model information, collection time, and detection results corresponding to the assembly image; If the detection result is unqualified, a second level alarm is issued, and / or the production line is stopped through the manufacturing system, and / or alarm information is obtained and sent to the mobile terminal.
8. A sealing strip detection system, characterized in that: include: processor; a memory for storing instructions executable by a processor; The processor is configured to execute the instructions to implement the sealing strip detection method according to any one of claims 1 to 7.
9. The sealing strip detection system according to claim 8, characterized in that: It also includes a door hanger and a detection frame, wherein the door hanger is used to transport the door, an RFID tag is provided on the door hanger, and the detection frame is used to install a camera assembly and an in-place sensor; The detection frame includes two frames spaced apart along a first direction, wherein the first direction is perpendicular to the transportation direction of the vehicle door, and the two frames form a detection station.
10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the sealing strip detection method according to any one of claims 1 to 7 is implemented.
