Automobile roof automatic hemming control system and method

Abstract

The present invention provides a control system and method for automatic hemming of automobile top covers, including: a model building module for learning features of target images of different types of automobile top covers, building a recognition model, learning associations of different types of automobile top covers and historical hemming strategies, and building a strategy auxiliary model; a decision module for analyzing automobile top covers to be hemmed based on the recognition model and the strategy auxiliary model, generating an auxiliary strategy for hemming the automobile top covers to be hemmed; an hemming control module for generating automatic hemming control instructions according to the auxiliary strategy for hemming the automobile top covers to be hemmed, and automatically hemming the automobile top covers to be hemmed; a display module for real-time monitoring of the hemming control process based on sensors, generating a dynamic hemming picture of the automobile top covers, and dynamically displaying the dynamic hemming picture on a preset display interface. The accuracy and reliability of the hemming of automobile top covers are improved.

Description

Technical Field

[0001] The invention relates to the technical field of equipment control, and in particular to a control system and method for automatic hemming of a car roof. Background Art

[0002] At present, in the field of automobile manufacturing, the hemming process of automobile roof plays a vital role in the overall quality and appearance of the car body. The traditional method of automobile roof hemming usually relies on manual or semi-automatic equipment. However, with the rapid development of the automobile industry and the continuous improvement of consumers' requirements for automobile quality and appearance, these traditional methods have gradually exposed a series of limitations;

[0003] Manual hemming is a common operation method. Workers rely on their experience and hand tools to hem the car roof. However, this method is not only inefficient, but the quality of hemming depends largely on the workers' skill level and working status. It is difficult to ensure the consistency and high precision of the hemming of the car roof. Moreover, although semi-automatic hemming equipment has improved efficiency to a certain extent, it still has many problems. For example, the degree of automation is not high enough, and frequent manual intervention and adjustment are required, resulting in unstable production rhythm. Moreover, its control system is relatively simple, and it is impossible to accurately monitor and adjust various parameters in the hemming process in real time, making it difficult to meet the increasingly stringent quality standards of car roofs.

[0004] Therefore, in order to overcome the above-mentioned defects, the present invention provides a control system and method for automatic hemming of a car roof. Summary of the invention

[0005] The present invention provides a control system and method for automatic hemming of automobile roofs, which are used to accurately and effectively construct a recognition model by performing feature learning on target images of automobile roofs of different types, and at the same time, to perform association learning on different types of automobile roofs and historical hemming strategies, and to accurately and reliably construct a strategy auxiliary model, so as to facilitate accurate and effective determination of the type and hemming strategy of the automobile roof to be hemmed. Secondly, the automobile roof to be hemmed is analyzed by the constructed recognition model and strategy auxiliary model, so as to effectively formulate the auxiliary strategy, and generate corresponding automatic hemming control instructions according to the obtained auxiliary strategy, so as to achieve accurate and reliable hemming control of the automobile roof. Finally, the hemming control process is monitored in real time, and a corresponding dynamic hemming picture is generated according to the monitoring result for display, so as to facilitate real-time and effective understanding of the hemming process of the automobile roof, and to facilitate timely abnormal locking and alarming when the hemming of the automobile roof is abnormal according to the displayed result, thereby improving the accuracy and reliability of the hemming of the automobile roof.

[0006] The present invention provides an automatic hemming control system for a car roof, comprising:

[0007] The model building module is used to obtain target images of different types of automobile roofs, perform feature learning on the target images, and build a recognition model. At the same time, it performs association learning on different types of automobile roofs and historical hemming strategies to build a strategy-assisted model.

[0008] A decision module, for analyzing the automobile roof to be hemmed based on the recognition model and the strategy auxiliary model, and generating an auxiliary strategy for hemming the automobile roof to be hemmed;

[0009] An edge wrapping control module is used to generate an automatic edge wrapping control instruction according to an auxiliary strategy for wrapping the top cover of the automobile to be wrapped, and to perform automatic edge wrapping control on the top cover of the automobile to be wrapped based on the automatic edge wrapping control instruction;

[0010] The display module is used to monitor the edge wrapping control process in real time based on the sensor, and to generate a dynamic edge wrapping picture of the automobile roof based on the monitoring result, and to dynamically display the dynamic edge wrapping picture on a preset display interface.

[0011] Preferably, a vehicle top cover automatic hemming control system, a model building module, comprises:

[0012] An image acquisition unit, used to acquire image acquisition requirements and to retrieve target images corresponding to different types of automobile roofs from a preset target library according to the image acquisition requirements;

[0013] An image processing unit, used for preprocessing target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers;

[0014] A feature extraction unit, used to extract features from a standard image based on a preset neural network to determine key features of a car roof;

[0015] Identification model generation unit, used to:

[0016] Performing a first learning on the type of the automobile roof to obtain a first model element, and performing a second learning on the key features of the automobile roof to obtain a second model element;

[0017] The first model element and the second model element are integrated to construct a recognition model.

[0018] Preferably, a vehicle top cover automatic hemming control system, a model building module, comprises:

[0019] A strategy retrieval unit, used to retrieve the historical hemming strategy corresponding to each type of automobile top cover, and at the same time, determine the historical hemming parameter set corresponding to the historical hemming strategy;

[0020] A learning dimension determination unit is used to read a historical edge parameter set, determine a parameter type corresponding to the historical edge parameter set, and determine a learning dimension according to the parameter type;

[0021] A classification unit is used to classify the historical edge parameter set according to the parameter type, and obtain the corresponding sub-historical edge parameter set under each learning dimension;

[0022] Strategy-assisted model building unit for:

[0023] Read the sub-history edge-wrapping parameter set corresponding to each learning dimension to determine the data change characteristics of the sub-history edge-wrapping parameter set corresponding to each learning dimension;

[0024] The data change characteristics of the corresponding sub-history edge parameter set under each learning dimension are learned to obtain the initial strategy auxiliary model;

[0025] The type of the automobile roof is associated with the initial strategy assistance model to obtain the strategy assistance model.

[0026] Preferably, in an automatic hemming control system for a car roof, parameter types include: an hemming starting point, an hemming ending point, an hemming path, and an hemming force.

[0027] Preferably, a decision model of an automatic hemming control system for a car top cover comprises:

[0028] A type recognition unit, used for reading the top cover image of the automobile top cover to be hemmed, inputting the top cover image of the automobile top cover to be hemmed into the recognition model for recognition, and outputting the type of the automobile top cover to be hemmed according to the recognition result;

[0029] The analysis unit is used to input the type of the automobile roof to be hemmed into the strategy auxiliary model for analysis, and obtain an auxiliary strategy for hemming the automobile roof to be hemmed.

[0030] Preferably, a decision model of an automatic hemming control system for a car top cover comprises:

[0031] Package unit for:

[0032] Before analyzing the hemmed automobile roof, the recognition model is used as the superordinate model, and the strategy auxiliary model is used as the subordinate model;

[0033] The recognition model and the strategy auxiliary model are associated in execution order based on the execution order of the upper model and the lower model;

[0034] The recognition model and strategy-assisted model are encapsulated based on the association results.

[0035] Preferably, an automatic hemming control system for a car top cover, an hemming control module, comprises:

[0036] A strategy reading unit, used for reading the auxiliary strategy and determining the execution steps for hemming the top cover of the automobile to be hemmed;

[0037] A parameter adjustment unit is used to obtain the edge-wrapping requirements of the customer, adjust the parameters of the execution steps according to the edge-wrapping requirements, and obtain the target execution steps according to the adjustment results;

[0038] An instruction generation unit is used to read the target execution steps, determine the edge wrapping execution action corresponding to each execution step, and generate an automatic edge wrapping control instruction according to the edge wrapping execution action corresponding to each execution step;

[0039] The control unit is used for automatically controlling the edge wrapping of the automobile roof to be edge wrapped according to the automatic edge wrapping control instruction.

[0040] Preferably, a vehicle top cover automatic hemming control system, display module, comprises:

[0041] Monitoring layout unit, used to:

[0042] Acquire an edge wrapping scene of a car roof, and determine a monitoring point for real-time monitoring of an edge wrapping control process of the car roof based on the edge wrapping scene;

[0043] At the same time, the monitoring dimensions of the edge control process are determined based on the monitoring requirements, and the multi-dimensional sensors are laid out based on the monitoring points and monitoring dimensions;

[0044] The dynamic display unit is used for:

[0045] Based on the monitoring layout results, the multi-dimensional sensors are linked and controlled, and based on the linkage control results, the edge wrapping control process of the car roof is monitored in real time to obtain the real-time edge wrapping parameters corresponding to different times;

[0046] Acquire a target shape of a car top cover, and generate a thumbnail simulation image of the car top cover based on the target shape;

[0047] The real-time hemming parameters at different moments are serialized, and the multi-dimensional quantitative hemming index values ​​at each moment are obtained based on the serialization processing results;

[0048] Determine the mapping position of the multi-dimensional quantitative edge-wrapping index value at each moment in the thumbnail simulation diagram based on the real-time monitoring result, and associate and display the multi-dimensional quantitative edge-wrapping index value at each moment at the corresponding mapping position;

[0049] Abnormal warning unit, used for:

[0050] Generate a dynamic bordering picture based on the associated display result, and at the same time, obtain configuration parameters of a preset display interface, and perform parameter adaptation on the dynamic bordering picture based on the configuration parameters;

[0051] Dynamically display the dynamic edge wrapping screen based on the parameter adaptation result, and connect the edge wrapping control process verification mechanism with the preset display interface based on the dynamic display result;

[0052] Based on the docking result and the edge control process verification mechanism, the multi-dimensional quantitative edge indicator values ​​at different mapping positions at each moment are verified, and when there are multi-dimensional quantitative edge indicator values ​​that do not meet the preset edge requirements, the current mapping position is locked;

[0053] Generate an abnormal report on the car top cover edge wrapping based on the locking result, and transmit the abnormal report to the management terminal for early warning notification.

[0054] The present invention provides a method for controlling automatic hemming of a car roof, comprising:

[0055] Step 1: Obtain target images of different types of car roofs, perform feature learning on the target images, and build a recognition model. At the same time, perform association learning on different types of car roofs and historical hemming strategies to build a strategy-assisted model.

[0056] Step 2: Analyze the automobile roof cover to be hemmed based on the recognition model and the strategy auxiliary model, and generate an auxiliary strategy for hemming the automobile roof cover to be hemmed;

[0057] Step 3: Generate an automatic edge wrapping control instruction according to the auxiliary strategy for edge wrapping the automobile roof cover to be edged, and perform automatic edge wrapping control on the automobile roof cover to be edged based on the automatic edge wrapping control instruction;

[0058] Step 4: The edge wrapping control process is monitored in real time based on the sensor, and a dynamic edge wrapping picture of the automobile roof is generated based on the monitoring result, and the dynamic edge wrapping picture is dynamically displayed on a preset display interface.

[0059] Preferably, in a method for controlling the automatic hemming of a car roof, in step 1, target images of different types of car roofs are obtained, and feature learning is performed on the target images to construct a recognition model, including:

[0060] Obtain image retrieval requirements, and retrieve target images corresponding to different types of car roofs from a preset target library according to the image retrieval requirements;

[0061] Preprocess the target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers;

[0062] Extract features from standard images based on a preset neural network to determine key features of the car roof;

[0063] The type of the automobile top cover is first learned to obtain a first model element, and at the same time, the key features of the automobile top cover are second learned to obtain a second model element;

[0064] The first model element and the second model element are integrated to construct a recognition model.

[0065] A recognition model for identifying the type of automobile roof is constructed based on the first learning result and the second learning result.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] By learning the features of target images of different types of automobile roof covers, the recognition model can be accurately and effectively constructed. At the same time, the different types of automobile roof covers and historical hemming strategies are associated with learning to accurately and reliably construct the strategy-assisted model, which is convenient for accurately and effectively determining the type and hemming strategy of the automobile roof cover to be hemmed. Secondly, the automobile roof cover to be hemmed is analyzed through the constructed recognition model and strategy-assisted model to effectively formulate the auxiliary strategy, and generate the corresponding automatic hemming control instructions according to the obtained auxiliary strategy to realize accurate and reliable hemming control of the automobile roof cover. Finally, the hemming control process is monitored in real time, and the corresponding dynamic hemming screen is generated according to the monitoring results for display, which is convenient for real-time and effective understanding of the hemming process of the automobile roof cover, and also convenient for timely abnormal locking and alarming when the hemming of the automobile roof cover is abnormal according to the display results, thereby improving the accuracy and reliability of the hemming of the automobile roof cover.

[0068] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0069] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0071] Figure 1 This is a structural diagram of an automatic hemming control system for a car roof according to an embodiment of the present invention;

[0072] Figure 2 It is a structural diagram of a model building module in an automatic hemming control system for a car top cover according to an embodiment of the present invention;

[0073] Figure 3The present invention is a flowchart of a method for controlling automatic hemming of a vehicle roof in an embodiment of the present invention. DETAILED DESCRIPTION

[0074] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0075] Embodiment 1:

[0076] This embodiment provides a control system for the automatic hemming of a car roof. Figure 1 As shown, including:

[0077] The model building module is used to obtain target images of different types of automobile roofs, perform feature learning on the target images, and build a recognition model. At the same time, it performs association learning on different types of automobile roofs and historical hemming strategies to build a strategy-assisted model.

[0078] A decision module, for analyzing the automobile roof to be hemmed based on the recognition model and the strategy auxiliary model, and generating an auxiliary strategy for hemming the automobile roof to be hemmed;

[0079] An edge wrapping control module is used to generate an automatic edge wrapping control instruction according to an auxiliary strategy for wrapping the top cover of the automobile to be wrapped, and to perform automatic edge wrapping control on the top cover of the automobile to be wrapped based on the automatic edge wrapping control instruction;

[0080] The display module is used to monitor the edge wrapping control process in real time based on the sensor, and to generate a dynamic edge wrapping picture of the automobile roof based on the monitoring result, and to dynamically display the dynamic edge wrapping picture on a preset display interface.

[0081] In this embodiment, the target image may be an image obtained by collecting images of different types of automobile roofs.

[0082] In this embodiment, feature learning can be to learn the characteristics of different types of automobile roof covers, so as to facilitate the construction of a recognition model based on the learned features, so as to facilitate accurate and effective identification of the type of automobile roof cover when performing automobile roof cover edging control.

[0083] In this embodiment, the historical hemming strategy is known in advance.

[0084] In this embodiment, the strategy auxiliary model can accurately and effectively generate corresponding hemming strategies for different types of automobile roofs when hemming, thereby achieving effective control of the hemming process of the automobile roof.

[0085] In this embodiment, the automobile roof to be hemmed may be an automobile roof that needs to be hemmed.

[0086] In this embodiment, the auxiliary strategy may be a solution suitable for edge-binding the top cover of the automobile to be edge-binded.

[0087] In this embodiment, the dynamic hemming screen can be a virtual screen generated by the monitoring results of the hemming control process, the purpose of which is to display the hemming process of the car roof in real time and effectively, so as to facilitate the user to understand the hemming progress and situation of the car roof in real time and effectively.

[0088] In this embodiment, the preset display interface is set in advance and is used to display the dynamic edge binding picture.

[0089] The working principle and beneficial effects of the above technical scheme are as follows: by learning the features of target images of different types of automobile roof covers, an accurate and effective construction of a recognition model is achieved; at the same time, association learning is performed on different types of automobile roof covers and historical hemming strategies, so as to achieve an accurate and reliable construction of a strategy-assisted model, which is convenient for accurately and effectively determining the type and hemming strategy of the automobile roof cover to be hemmed; secondly, the automobile roof cover to be hemmed is analyzed by the constructed recognition model and strategy-assisted model, so as to achieve effective formulation of auxiliary strategies, and generate corresponding automatic hemming control instructions according to the obtained auxiliary strategies, so as to achieve accurate and reliable hemming control of the automobile roof cover; finally, the hemming control process is monitored in real time, and a corresponding dynamic hemming screen is generated according to the monitoring results for display, so as to achieve real-time and effective understanding of the hemming process of the automobile roof cover, and to timely perform abnormal locking and alarm when the hemming of the automobile roof cover is abnormal according to the displayed results, thereby improving the accuracy and reliability of the hemming of the automobile roof cover.

[0090] Embodiment 2:

[0091] Based on Example 1, this embodiment provides an automatic hemming control system for a car roof, such as Figure 2 As shown, the model building modules include:

[0092] An image acquisition unit, used to acquire image acquisition requirements and to retrieve target images corresponding to different types of automobile roofs from a preset target library according to the image acquisition requirements;

[0093] An image processing unit, used for preprocessing target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers;

[0094] A feature extraction unit, used to extract features from a standard image based on a preset neural network to determine key features of a car roof;

[0095] Identification model generation unit, used to:

[0096] Performing a first learning on the type of the automobile roof to obtain a first model element, and performing a second learning on the key features of the automobile roof to obtain a second model element;

[0097] The first model element and the second model element are integrated to construct a recognition model.

[0098] In this embodiment, the preset target library is set in advance and is used to store target images corresponding to different types of automobile roofs, wherein the target image is the image of the automobile roof.

[0099] In this embodiment, the preprocessing may be denoising or other processing on the target image, the purpose of which is to facilitate feature extraction, wherein the standard image is the result of the preprocessing on the target image.

[0100] In this embodiment, the key feature may be information such as the shapes of different types of car roofs extracted after processing the standard image through a preset neural network.

[0101] In this embodiment, the first model element may be a model structure that can identify the type of the car roof obtained by learning the type of the car roof.

[0102] In this embodiment, the second model element may be a model structure that is obtained after learning the key features of the car roof and is capable of identifying the key features of the car roof.

[0103] The working principle and beneficial effects of the above technical solution are: first, obtaining the image retrieval requirements, and then selecting the target images corresponding to different types of car roofs from a pre-set target gallery according to the image retrieval requirements; performing preprocessing operations on the target images to obtain standard images of different types of car roofs; second, using a preset neural network to extract features from the standard images, and then clarifying the key features of the car roof; third, performing a first learning process on the type of the car roof to obtain the first model element, and also performing a second learning process on the key features of the car roof to obtain the second model element; finally, the first model element and the second model element are comprehensively sorted out to construct a recognition model that can identify the type of car roof, and by separately learning the types and key features of the car roof and comprehensively constructing the recognition model, different types of car roofs can be more accurately identified, thereby improving the recognition efficiency.

[0104] Embodiment 3:

[0105] Based on Example 1, this embodiment provides an automatic hemming control system for a car roof, and a model building module, including:

[0106] A strategy retrieval unit, used to retrieve the historical hemming strategy corresponding to each type of automobile top cover, and at the same time, determine the historical hemming parameter set corresponding to the historical hemming strategy;

[0107] A learning dimension determination unit is used to read a historical edge parameter set, determine a parameter type corresponding to the historical edge parameter set, and determine a learning dimension according to the parameter type;

[0108] A classification unit is used to classify the historical edge-wrapping parameter set according to the parameter type, and obtain the corresponding sub-historical edge-wrapping parameter set under each learning dimension;

[0109] Strategy-assisted model building unit for:

[0110] Read the sub-history edge-wrapping parameter set corresponding to each learning dimension to determine the data change characteristics of the sub-history edge-wrapping parameter set corresponding to each learning dimension;

[0111] The data change characteristics of the corresponding sub-history edge parameter set under each learning dimension are learned to obtain the initial strategy auxiliary model;

[0112] The type of the automobile roof is associated with the initial strategy assistance model to obtain the strategy assistance model.

[0113] In this embodiment, the historical hemming parameter set may be specific parameters corresponding to the historical hemming strategy, for example, parameters such as hemming strength and hemming starting point.

[0114] In this embodiment, the learning dimension is determined according to the parameter type and has a one-to-one correspondence with the parameter type.

[0115] In this embodiment, the sub-historical edge-binding parameter set may be a result obtained by classifying the historical edge-binding parameter set according to parameter type.

[0116] In this embodiment, the data change feature may be a change in data values ​​and a change in composition corresponding to the sub-history wrapping parameter set.

[0117] In this embodiment, the initial strategy auxiliary model may be a result obtained after training the data change characteristics of the sub-history edge parameter set under each learning dimension, and is a model element corresponding to each learning dimension.

[0118] The working principle and beneficial effects of the above technical solution are: by analyzing the historical hemming strategies, the parameter types and learning dimensions can be accurately and effectively determined, so that the sub-historical hemming parameter sets under each dimension can be effectively analyzed according to the learning dimensions, and the initial strategy auxiliary model under each learning dimension can be constructed according to the analysis results. Finally, the initial strategy auxiliary models under different learning dimensions are associated to achieve the effective construction of the strategy auxiliary model, which provides convenience for the automatic hemming control of the automobile roof.

[0119] Embodiment 4:

[0120] On the basis of Example 3, this embodiment provides an automatic hemming control system for a car roof, and the parameter types include: an hemming starting point, an hemming ending point, an hemming path, and an hemming force.

[0121] Embodiment 5:

[0122] Based on Example 1, this embodiment provides a vehicle roof automatic hemming control system, and the decision model includes:

[0123] A type recognition unit, used for reading the top cover image of the automobile top cover to be hemmed, inputting the top cover image of the automobile top cover to be hemmed into the recognition model for recognition, and outputting the type of the automobile top cover to be hemmed according to the recognition result;

[0124] The analysis unit is used to input the type of the automobile roof to be hemmed into the strategy auxiliary model for analysis, and obtain an auxiliary strategy for hemming the automobile roof to be hemmed.

[0125] The working principle and beneficial effects of the above technical solution are: by identifying the image of the top cover of the car top cover to be edged through the recognition model, the type of the car top cover to be edged can be effectively identified, and the recognition result is input into the strategy auxiliary model for analysis, so as to realize the auxiliary strategy for edge-binding the car top cover to be edged and quickly and effectively determine, which provides convenience for the automatic edge-binding control of the car top cover.

[0126] Embodiment 6:

[0127] Based on Example 1, this embodiment provides a vehicle roof automatic hemming control system, and the decision model includes:

[0128] Package unit for:

[0129] Before analyzing the hemmed automobile roof, the recognition model is used as the superordinate model, and the strategy auxiliary model is used as the subordinate model;

[0130] The recognition model and the strategy auxiliary model are associated in execution order based on the execution order of the upper model and the lower model;

[0131] The recognition model and strategy-assisted model are encapsulated based on the association results.

[0132] In this embodiment, the superordinate model may be a recognition model used as the first model used in the analysis process, that is, the data to be analyzed needs to be input into the recognition model for processing before subsequent processing can be performed.

[0133] In this embodiment, the subordinate model may be a representation of the order in which the strategy auxiliary model processes the data, that is, the output result of the recognition model needs to be processed.

[0134] The working principle and beneficial effect of the above technical solution are: by determining the execution order of the recognition model and the strategy-assisted model, the recognition model and the strategy-assisted model are effectively associated and packaged according to the execution order, thereby facilitating effective analysis of the car roof to be edged according to the analysis order.

[0135] Embodiment 7:

[0136] Based on Example 1, this embodiment provides an automatic hemming control system for a car roof, and an hemming control module, including:

[0137] A strategy reading unit, used for reading the auxiliary strategy and determining the execution steps for hemming the top cover of the automobile to be hemmed;

[0138] A parameter adjustment unit is used to obtain the edge-wrapping requirements of the customer, adjust the parameters of the execution steps according to the edge-wrapping requirements, and obtain the target execution steps according to the adjustment results;

[0139] An instruction generation unit is used to read the target execution steps, determine the edge wrapping execution action corresponding to each execution step, and generate an automatic edge wrapping control instruction according to the edge wrapping execution action corresponding to each execution step;

[0140] The control unit is used for automatically controlling the edge wrapping of the automobile roof to be edge wrapped according to the automatic edge wrapping control instruction.

[0141] In this embodiment, the execution step may be a specific execution operation corresponding to hemming the automobile roof to be hemmed, including the required process and the like.

[0142] In this embodiment, the edge wrapping requirement is submitted by the customer and is used to represent the personalized requirement for edge wrapping.

[0143] In this embodiment, the target execution step may be a result obtained by adaptively adjusting the execution step according to the edge wrapping requirement.

[0144] The working principle and beneficial effects of the above technical solution are: by analyzing the auxiliary strategy, the execution steps of the hemming are effectively determined; secondly, the parameters of the determined execution steps are adjusted according to the customer's hemming requirements, so as to accurately and effectively lock the target execution steps that meet the user's needs; finally, automatic hemming control instructions are generated according to the target execution steps, and reliable and effective automatic hemming control is achieved on the car roof to be hemmed according to the automatic hemming control instructions, thereby improving the effect of the hemming control.

[0145] Embodiment 8:

[0146] Based on Example 1, this embodiment provides an automatic hemming control system for a car roof, a display module, including:

[0147] Monitoring layout unit, used to:

[0148] Acquire an edge wrapping scene of a car roof, and determine a monitoring point for real-time monitoring of an edge wrapping control process of the car roof based on the edge wrapping scene;

[0149] At the same time, the monitoring dimensions of the edge control process are determined based on the monitoring requirements, and the multi-dimensional sensors are laid out based on the monitoring points and monitoring dimensions;

[0150] The dynamic display unit is used for:

[0151] Based on the monitoring layout results, the multi-dimensional sensors are linked and controlled, and based on the linkage control results, the edge wrapping control process of the car roof is monitored in real time to obtain the real-time edge wrapping parameters corresponding to different times;

[0152] Acquire a target shape of a car top cover, and generate a thumbnail simulation image of the car top cover based on the target shape;

[0153] The real-time hemming parameters at different moments are serialized, and the multi-dimensional quantitative hemming index values ​​at each moment are obtained based on the serialization processing results;

[0154] Determine the mapping position of the multi-dimensional quantitative edge-wrapping index value at each moment in the thumbnail simulation diagram based on the real-time monitoring result, and associate and display the multi-dimensional quantitative edge-wrapping index value at each moment at the corresponding mapping position;

[0155] Abnormal warning unit, used for:

[0156] Generate a dynamic border screen based on the associated display result, and at the same time, obtain configuration parameters of a preset display interface, and perform parameter adaptation on the dynamic border screen based on the configuration parameters;

[0157] Dynamically display the dynamic edge wrapping screen based on the parameter adaptation result, and connect the edge wrapping control process verification mechanism with the preset display interface based on the dynamic display result;

[0158] Based on the docking result and the edge control process verification mechanism, the multi-dimensional quantitative edge indicator values ​​at different mapping positions at each moment are verified, and when there are multi-dimensional quantitative edge indicator values ​​that do not meet the preset edge requirements, the current mapping position is locked;

[0159] Generate an abnormal report on the car top cover edge wrapping based on the locking result, and transmit the abnormal report to the management terminal for early warning notification.

[0160] In this embodiment, the hemming scene may be a working environment in which the hemming process of the automobile roof is performed.

[0161] In this embodiment, the monitoring dimension may be the type of monitoring of the hemming control process, for example, the hemming path, the hemming strength, and the sequence of hemming execution steps.

[0162] In this embodiment, the real-time hemming parameters may be the results obtained after real-time monitoring of the hemming control process of the automobile roof, including information such as the hemming strength and the starting point position of the hemming.

[0163] In this embodiment, the target shape may be the shape of a car roof or the like.

[0164] In this embodiment, the thumbnail simulation image may be a virtual image generated according to the target shape and capable of representing the edge wrapping condition of the automobile roof.

[0165] In this embodiment, the multi-dimensional quantitative hemming index value may be the specific hemming condition corresponding to the hemming of the automobile roof at each moment.

[0166] In this embodiment, the mapping position may be the specific edge position corresponding to the multi-dimensional quantitative edge index value at each moment in the thumbnail simulation image, that is, the specific edge conditions of different dimensions representing the current position.

[0167] In this embodiment, the preset display interface is set in advance, wherein the configuration parameters may be format requirements of the preset display interface for display content, etc.

[0168] In this embodiment, the verification mechanism of the hemming control process is known in advance and is a specific method for verifying the hemming condition of the automobile roof.

[0169] In this embodiment, the abnormality report is a data report for recording abnormalities in the hemming of the automobile roof.

[0170] The working principle and beneficial effects of the above technical solution are: by realizing effective monitoring layout of multi-dimensional sensors according to the hemming scene of the automobile roof, it is convenient to carry out real-time and effective monitoring of the hemming process of the automobile roof; secondly, the multi-dimensional sensors are linked and controlled to realize real-time and effective monitoring of the hemming process of the automobile roof; at the same time, a thumbnail simulation diagram is generated according to the target shape of the automobile roof, and the obtained real-time hemming parameters are serialized and displayed in association on the thumbnail simulation diagram, so that the management personnel can intuitively and effectively view the hemming situation of the automobile roof; finally, a dynamic hemming picture is generated according to the associated display result, and the dynamic hemming picture is dynamically displayed on the preset display interface; and the hemming state of the automobile roof is verified in real time and effectively through the hemming control process verification mechanism, so as to ensure the reliability of the automatic hemming control of the automobile roof; at the same time, when there is an abnormality in the hemming state, a car roof hemming abnormality report is generated in time and a corresponding early warning notification is made, so as to facilitate the management personnel to adjust the hemming strategy in time and ensure the hemming control effect of the automobile roof.

[0171] Embodiment 9:

[0172] This embodiment provides a method for controlling the automatic hemming of a car roof. Figure 3 As shown, including:

[0173] Step 1: Obtain target images of different types of car roofs, perform feature learning on the target images, and build a recognition model. At the same time, perform association learning on different types of car roofs and historical hemming strategies to build a strategy-assisted model.

[0174] Step 2: Analyze the automobile roof cover to be hemmed based on the recognition model and the strategy auxiliary model, and generate an auxiliary strategy for hemming the automobile roof cover to be hemmed;

[0175] Step 3: Generate an automatic edge wrapping control instruction according to the auxiliary strategy for edge wrapping the automobile roof cover to be edged, and perform automatic edge wrapping control on the automobile roof cover to be edged based on the automatic edge wrapping control instruction;

[0176] Step 4: The edge wrapping control process is monitored in real time based on the sensor, and a dynamic edge wrapping picture of the automobile roof is generated based on the monitoring result, and the dynamic edge wrapping picture is dynamically displayed on a preset display interface.

[0177] The working principle and beneficial effects of the above technical scheme are as follows: by learning the features of target images of different types of automobile roof covers, an accurate and effective construction of a recognition model is achieved; at the same time, association learning is performed on different types of automobile roof covers and historical hemming strategies, so as to achieve an accurate and reliable construction of a strategy-assisted model, which is convenient for accurately and effectively determining the type and hemming strategy of the automobile roof cover to be hemmed; secondly, the automobile roof cover to be hemmed is analyzed by the constructed recognition model and strategy-assisted model, so as to achieve effective formulation of auxiliary strategies, and generate corresponding automatic hemming control instructions according to the obtained auxiliary strategies, so as to achieve accurate and reliable hemming control of the automobile roof cover; finally, the hemming control process is monitored in real time, and a corresponding dynamic hemming screen is generated according to the monitoring results for display, so as to achieve real-time and effective understanding of the hemming process of the automobile roof cover, and to timely perform abnormal locking and alarm when the hemming of the automobile roof cover is abnormal according to the displayed results, thereby improving the accuracy and reliability of the hemming of the automobile roof cover.

[0178] Embodiment 10:

[0179] On the basis of Example 9, this embodiment provides a method for controlling the automatic hemming of a car roof. In step 1, target images of different types of car roofs are obtained, and feature learning is performed on the target images to construct a recognition model, including:

[0180] An image acquisition unit, used to acquire image acquisition requirements and to retrieve target images corresponding to different types of automobile roofs from a preset target library according to the image acquisition requirements;

[0181] An image processing unit, used for preprocessing target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers;

[0182] A feature extraction unit, used to extract features from a standard image based on a preset neural network to determine key features of a car roof;

[0183] Identification model generation unit, used to:

[0184] The type of the automobile top cover is first learned to obtain a first model element, and at the same time, the key features of the automobile top cover are second learned to obtain a second model element;

[0185] The first model element and the second model element are integrated to construct a recognition model.

[0186] A recognition model for identifying the type of automobile roof is constructed based on the first learning result and the second learning result.

[0187] The working principle and beneficial effects of the above technical solution are: first, obtaining the image retrieval requirements, and then selecting the target images corresponding to different types of car roofs from a pre-set target gallery according to the image retrieval requirements; performing preprocessing operations on the target images to obtain standard images of different types of car roofs; second, using a preset neural network to extract features from the standard images, and then clarifying the key features of the car roof; third, performing a first learning process on the type of the car roof to obtain the first model element, and also performing a second learning process on the key features of the car roof to obtain the second model element; finally, the first model element and the second model element are comprehensively sorted out to construct a recognition model that can identify the type of car roof, and by separately learning the types and key features of the car roof and comprehensively constructing the recognition model, different types of car roofs can be more accurately identified, thereby improving the recognition efficiency.

[0188] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An automatic hemming control system for a car roof, characterized in that: include: The model building module is used to obtain target images of different types of automobile roofs, perform feature learning on the target images, and build a recognition model. At the same time, it performs association learning on different types of automobile roofs and historical hemming strategies to build a strategy-assisted model. A decision module, for analyzing the automobile roof to be hemmed based on the recognition model and the strategy auxiliary model, and generating an auxiliary strategy for hemming the automobile roof to be hemmed; An edge wrapping control module is used to generate an automatic edge wrapping control instruction according to an auxiliary strategy for wrapping the top cover of the automobile to be wrapped, and to perform automatic edge wrapping control on the top cover of the automobile to be wrapped based on the automatic edge wrapping control instruction; A display module is used to monitor the edge wrapping control process in real time based on sensors, and to generate a dynamic edge wrapping picture of the automobile roof based on the monitoring result, and to dynamically display the dynamic edge wrapping picture on a preset display interface; Model building modules, including: An image acquisition unit, used to acquire image acquisition requirements and to retrieve target images corresponding to different types of automobile roofs from a preset target library according to the image acquisition requirements; An image processing unit, used for preprocessing target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers; A feature extraction unit, used to extract features from a standard image based on a preset neural network to determine key features of a car roof; Identification model generation unit, used to: Performing a first learning on the type of the automobile roof to obtain a first model element, and performing a second learning on the key features of the automobile roof to obtain a second model element; The first model element and the second model element are integrated to construct a recognition model.

2. The automatic hemming control system for automobile roof according to claim 1, characterized in that: Model building modules, including: A strategy retrieval unit, used to retrieve the historical hemming strategy corresponding to each type of automobile top cover, and at the same time, determine the historical hemming parameter set corresponding to the historical hemming strategy; A learning dimension determination unit is used to read a historical edge parameter set, determine a parameter type corresponding to the historical edge parameter set, and determine a learning dimension according to the parameter type; A classification unit is used to classify the historical edge parameter set according to the parameter type, and obtain the corresponding sub-historical edge parameter set under each learning dimension; Strategy-assisted model building unit for: Read the sub-history edge-wrapping parameter set corresponding to each learning dimension to determine the data change characteristics of the sub-history edge-wrapping parameter set corresponding to each learning dimension; The data change characteristics of the corresponding sub-history edge parameter set under each learning dimension are learned to obtain the initial strategy auxiliary model; The type of the automobile roof is associated with the initial strategy assistance model to obtain the strategy assistance model.

3. The automatic hemming control system for automobile roof according to claim 2, characterized in that: Parameter types include: hemming start point, hemming end point, hemming path, and hemming strength.

4. The automatic hemming control system for automobile roof according to claim 1, characterized in that: Decision-making module, including: A type recognition unit, used for reading the top cover image of the automobile top cover to be hemmed, inputting the top cover image of the automobile top cover to be hemmed into the recognition model for recognition, and outputting the type of the automobile top cover to be hemmed according to the recognition result; The analysis unit is used to input the type of the automobile roof to be hemmed into the strategy auxiliary model for analysis, and obtain an auxiliary strategy for hemming the automobile roof to be hemmed.

5. The automatic hemming control system for automobile roof according to claim 1, characterized in that: Decision-making module, including: Package unit for: Before analyzing the hemmed automobile roof, the recognition model is used as the superordinate model, and the strategy auxiliary model is used as the subordinate model; The recognition model and the strategy auxiliary model are associated in execution order based on the execution order of the upper model and the lower model; The recognition model and strategy-assisted model are encapsulated based on the association results.

6. The automatic hemming control system for automobile roof according to claim 1, characterized in that: Edge wrapping control module, including: A strategy reading unit, used for reading the auxiliary strategy and determining the execution steps for hemming the top cover of the automobile to be hemmed; A parameter adjustment unit is used to obtain the edge-wrapping requirements of the customer, adjust the parameters of the execution steps according to the edge-wrapping requirements, and obtain the target execution steps according to the adjustment results; An instruction generation unit is used to read the target execution steps, determine the edge wrapping execution action corresponding to each execution step, and generate an automatic edge wrapping control instruction according to the edge wrapping execution action corresponding to each execution step; The control unit is used for automatically controlling the edge wrapping of the automobile roof to be edge wrapped according to the automatic edge wrapping control instruction.

7. The automatic hemming control system for automobile roof according to claim 1, characterized in that: Display module, including: Monitoring layout unit, used to: Acquire an edge wrapping scene of a car roof, and determine a monitoring point for real-time monitoring of an edge wrapping control process of the car roof based on the edge wrapping scene; At the same time, the monitoring dimensions of the edge control process are determined based on the monitoring requirements, and the multi-dimensional sensors are laid out based on the monitoring points and monitoring dimensions; The dynamic display unit is used for: Based on the monitoring layout results, the multi-dimensional sensors are linked and controlled, and based on the linkage control results, the edge wrapping control process of the car roof is monitored in real time to obtain the real-time edge wrapping parameters corresponding to different times; Acquire a target shape of a car top cover, and generate a thumbnail simulation image of the car top cover based on the target shape; The real-time hemming parameters at different moments are serialized, and the multi-dimensional quantitative hemming index values ​​at each moment are obtained based on the serialization processing results; Determine the mapping position of the multi-dimensional quantitative edge-wrapping index value at each moment in the thumbnail simulation diagram based on the real-time monitoring result, and associate and display the multi-dimensional quantitative edge-wrapping index value at each moment at the corresponding mapping position; Abnormal warning unit, used for: Generate a dynamic border screen based on the associated display result, and at the same time, obtain configuration parameters of a preset display interface, and perform parameter adaptation on the dynamic border screen based on the configuration parameters; Dynamically display the dynamic edge wrapping screen based on the parameter adaptation result, and connect the edge wrapping control process verification mechanism with the preset display interface based on the dynamic display result; Based on the docking result and the edge control process verification mechanism, the multi-dimensional quantitative edge indicator values ​​at different mapping positions at each moment are verified, and when there are multi-dimensional quantitative edge indicator values ​​that do not meet the preset edge requirements, the current mapping position is locked; Generate an abnormal report on the car top cover edge wrapping based on the locking result, and transmit the abnormal report to the management terminal for early warning notification.

8. A method for controlling the automatic hemming of a car roof, characterized in that: include: Step 1: Obtain target images of different types of car roofs, perform feature learning on the target images, and build a recognition model. At the same time, perform association learning on different types of car roofs and historical hemming strategies to build a strategy-assisted model. Step 2: Analyze the automobile roof cover to be hemmed based on the recognition model and the strategy auxiliary model, and generate an auxiliary strategy for hemming the automobile roof cover to be hemmed; Step 3: Generate an automatic edge wrapping control instruction according to the auxiliary strategy for edge wrapping the automobile roof cover to be edged, and perform automatic edge wrapping control on the automobile roof cover to be edged based on the automatic edge wrapping control instruction; Step 4: Real-time monitoring of the edge wrapping control process is performed based on the sensor, and a dynamic edge wrapping picture of the automobile roof is generated based on the monitoring result, and the dynamic edge wrapping picture is dynamically displayed on a preset display interface; In step 1, target images of different types of car roofs are obtained, and feature learning is performed on the target images to build a recognition model, including: Obtain image retrieval requirements, and retrieve target images corresponding to different types of car roofs from a preset target library according to the image retrieval requirements; Preprocess the target images corresponding to different types of automobile roof covers to obtain standard images of different types of automobile roof covers; Extract features from standard images based on a preset neural network to determine key features of the car roof; Performing a first learning on the type of the automobile roof to obtain a first model element, and performing a second learning on the key features of the automobile roof to obtain a second model element; The first model element and the second model element are integrated to construct a recognition model.

Images