Aluminum alloy cutting adaptive control system and method based on visual detection

Abstract

The application discloses an aluminum alloy cutting adaptive control system and method based on visual detection, relates to the technical field of visual detection, and processes a to-be-cut region into a plurality of detection sub-regions, creates respective visual detection intelligent agents, executes visual detection of the aluminum alloy workpiece by the respective detection sub-regions, registers the visual detection intelligent agents to a honeycomb Internet grid, executes data analysis on each visual detection intelligent agent, packs the analysis result into a cutting control file, arranges data in the honeycomb Internet grid, sets a target cutting path to upload the honeycomb Internet grid, selects cutting control files at different positions in the honeycomb Internet grid, generates an execution file for executing adaptive cutting on the aluminum alloy workpiece, obtains real-time cutting data of each cutting region, judges whether there is an abnormal operation in the current cutting region, decides whether to generate feedback correction data of the cutting region based on the judgment result and synchronizes the feedback correction data to the honeycomb Internet grid, and updates the corresponding visual detection intelligent agent.

Description

Technical Field

[0001] This invention relates to the field of machine vision technology, specifically to an adaptive control system and method for aluminum alloy cutting based on visual inspection. Background Technology

[0002] The demand for intelligent and collaborative control systems in aluminum alloy cutting is increasing. In existing aluminum alloy cutting control systems, there is a lack of effective linkage between visual inspection and cutting execution. Inspection data is difficult to efficiently support cutting control decisions, and there is no unified distributed data management architecture to achieve overall scheduling and sharing of inspection, control, and execution data. The real-time data feedback mechanism of the cutting process is imperfect, making it impossible to correct operations and optimize the visual inspection process in a timely manner based on the cutting site conditions. This results in poor cutting adaptability, easy occurrence of operational abnormalities, and difficulty in guaranteeing cutting accuracy and processing efficiency, making it difficult to meet the high-precision and adaptive cutting requirements of aluminum alloy workpieces. Summary of the Invention

[0003] The purpose of this invention is to provide an adaptive control system and method for aluminum alloy cutting based on vision detection, so as to solve the problems in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an adaptive control system for aluminum alloy cutting based on vision detection, the system comprising: The visual inspection module is used to process the area to be cut into several inspection sub-regions, create and deploy a corresponding visual inspection agent in each inspection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding inspection sub-region. The visual inspection agent is registered to the honeycomb grid. The data processing module performs data parsing once for each visual detection agent based on the honeycomb grid, packages the parsing results into the corresponding cutting control file, and arranges the cutting control file in the honeycomb grid. The cutting execution module is used to set the target cutting path and upload the target cutting path to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions of itself, integrates them to generate the execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. The feedback correction module obtains real-time cutting data of the target cutting path in each cutting area, determines whether there is any abnormal operation in the current cutting area based on the real-time cutting data, and decides whether to generate feedback correction data for the cutting area based on the judgment result. The feedback correction data is synchronized to the honeycomb grid to update the corresponding visual detection agent.

[0005] Furthermore, the area to be cut is divided into several inspection sub-regions. A corresponding visual inspection agent is created and deployed in each inspection sub-region. The process of visual inspection of the aluminum alloy workpiece in the corresponding inspection sub-region by the visual inspection agent includes: The aluminum alloy workpiece is identified as the target object that needs to be cut. Determine the area to be cut of the target object, deploy an industrial vision camera, divide the area to be cut into several detection sub-regions based on the field of view, initialize a model architecture to control the industrial vision camera, and allocate a database address to store the model architecture for connection operations with the database. Based on the number of detection sub-regions, create a corresponding number of index addresses for connecting to the database address. Operate the index addresses and database addresses to deploy a visual inspection agent for each detection sub-region. Based on the visual inspection agent, control their respective industrial vision cameras to complete a visual inspection of the aluminum alloy workpiece and obtain the visual inspection data of their respective detection sub-regions.

[0006] Furthermore, the process of manipulating the index address and database address to deploy the visual detection agent for each detection sub-region includes: Each detection sub-region calls the database address based on its own index address, obtains the initial model architecture stored in the database through the database address, sets the process script for performing vision and stores it in the database, and associates the process script with the database address; Based on the process script processing of the initial model architecture, the initial model architecture is converted into a visual inspection agent by filling in the script content after the process script is compiled. The database address after the visual inspection agent is created is updated, and the association between the updated database address and other index addresses is established synchronously. The visual inspection agent is synchronized to several other index addresses.

[0007] Furthermore, the process of registering visual detection agents to the cellular internet grid includes: The cellular network consists of a number of cellular cells. Adjacent cellular cells communicate with each other. Each cellular cell is assigned a cellular address, and a registration pool for cellular cells is created based on the cellular addresses. A listening thread is established for each registration pool to monitor all operations performed within the registration pool and determine whether any abnormal behavior occurs within the registration pool based on the data monitoring. If abnormal behavior occurs in the registration pool, any registration behavior that operates on the registration pool will be prohibited. When there is no abnormal behavior in the registration pool, the system receives a registration request from any visual detection agent, binds the index address of the corresponding visual detection agent to the cellular address of the currently unassigned cellular grid, constructs an associated address pair, and completes the registration of the visual detection agent on the cellular internet grid.

[0008] Furthermore, the process of performing data parsing once for each visual detection agent based on the honeycomb internet grid, and packaging the parsing results into the corresponding cutting control file includes: A data parsing queue is constructed between the visual inspection agent and the honeycomb grid. The data parsing script is edited and input into the data parsing queue. The data parsing script includes several parsing fields that execute data parsing on the visual inspection agent in an order of execution. Each parsing field is used to extract a subset of visual inspection data of the aluminum alloy workpiece by the visual inspection agent. For any visual inspection agent, based on all the assigned parsing fields, perform a complete data parsing of the corresponding visual inspection agent once. A complete data parsing includes several batch parsings in execution order. Integrate the visual inspection subset data from all batch parsings as the final parsing result. Set packaging parameters to use the parsing results of each data analysis as the cutting control file for the corresponding visual detection agent.

[0009] Furthermore, the process of arranging the control file data within the cellular grid includes: Each cutting control file is initially arranged into the honeycomb grid of its respective visual detection agent. A direct honeycomb path is constructed between honeycomb grids in adjacent positions. A honeycomb grid is selected as the retrieval source. Honeycomb grids that match the retrieval source are retrieved and set as the same source grids. Set the correspondence between the retrieval source and the same source grid, and set the honeycomb grid as the path jump point. When the retrieval source matches all the same source grids related to itself, it indirectly connects the honeycomb direct path of its own position with the path jump point of each same source grid to build a data chain in the same cutting path. Integrate all the cutting control files corresponding to the data chain, build a control file set, bind each retrieval source to the control file set, and store the control file set in the honeycomb grid of the retrieval source.

[0010] Furthermore, a target cutting path is set and uploaded to the cellular grid. The cellular grid selects cutting control files arranged at different positions within itself, integrates them to generate an execution file that executes the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. This process includes: The target cutting path consists of several sub-target cutting paths. A data channel is established, and an electronic fence is deployed. The electronic fence consists of several sub-fence areas. A behavior situation set is created in each sub-fence area. The behavior situation set is used to define several abnormal access behaviors. When an access behavior occurs in any sub-fence area of ​​the electronic fence, it is determined whether the access behavior is an abnormal access behavior that matches the behavior status set of the current sub-fence area. If it is, a behavior point is occupied in the sub-fence area; otherwise, no operation is performed. Set up communication lines between corresponding behavior points in different sub-fence areas, connect the behavior points in the sub-fence areas through the communication lines, and build a cross-fence message channel to synchronize messages between different sub-fence areas. When all the sub-target cutting paths are uploaded to the honeycomb grid, each sub-target cutting path is adapted to the honeycomb grids arranged at different positions on the honeycomb grid, and the control file set of the successfully adapted honeycomb grids is selected to obtain the corresponding complete cutting control files. Set the cutting control file as the execution file for the target cutting path, and perform adaptive cutting on the aluminum alloy workpiece based on the execution file.

[0011] Furthermore, the process of obtaining real-time cutting data of the target cutting path in each cutting region, and determining whether there are any abnormal operations in the current cutting region based on the real-time cutting data, includes: The cutting range of the sub-target cutting path on the target object is determined as the corresponding cutting area, and all operation data performed by the cutting device in the cutting area is recorded as real-time cutting data. Set up a standard cutting dataset, which includes standard cutting data for several categories of cutting behaviors under standard execution procedures. Input the real-time cutting data and the standard cutting dataset into a pre-built bag-of-words model. The bag-of-words model compares and determines whether there are any abnormal operations in each cutting region. When the comparison results are inconsistent, it is determined that there is an abnormal operation in the corresponding cutting area; when the comparison results are consistent, it is determined that there is no abnormal operation.

[0012] Furthermore, the process of determining whether to generate feedback correction data for the segmented region based on the judgment result, and synchronizing the feedback correction data to the cellular network grid to update the corresponding visual detection agent includes: If no abnormal operation is detected in the cutting area, no operation is performed. When an abnormal operation is detected in the cutting area, corresponding feedback correction data is generated based on the real-time cutting data and standard cutting data of the corresponding cutting area. In the segmented area where abnormal operations occur, a data synchronization point is set up, and a synchronization operation queue is established between the data synchronization point and a corresponding cellular grid. The feedback correction data is synchronized to the cellular grid through the synchronization operation queue. Based on the feedback correction data, the process script at the current location of the honeycomb grid is expanded and compiled, the visual inspection agent stored at the honeycomb grid is updated, the update result of the visual inspection agent is sent to other honeycomb grids via data link, and the visual inspection agent at other honeycomb grids is optimized based on the update result.

[0013] This invention also provides an adaptive control method for an adaptive control system for aluminum alloy cutting based on vision detection, comprising the following steps: Step S1: Divide the area to be cut into several detection sub-regions, create and deploy a corresponding visual inspection agent in each detection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding detection sub-region. Register the visual inspection agent to the honeycomb grid. Step S2: Perform data parsing once for each visual detection agent based on the honeycomb internet grid, and package the parsing results into the corresponding cutting control file, and arrange the cutting control file in the honeycomb internet grid; Step S3: Set the target cutting path and upload it to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions, integrates them to generate an execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. Step S4: Obtain real-time cutting data of the target cutting path in each cutting area, determine whether there is abnormal operation in the current cutting area based on the real-time cutting data, decide whether to generate feedback correction data for the cutting area based on the judgment result, and synchronize the feedback correction data to the honeycomb Internet grid to update the corresponding visual detection agent.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. The visual inspection module in this invention deploys dedicated visual inspection agents according to the inspection sub-regions and registers them with the honeycomb grid, realizing targeted visual inspection of the areas to be cut on aluminum alloy workpieces, providing accurate data support for cutting control, and completing the parsing of visual inspection data, packaging and arranging of cutting control files based on the honeycomb grid, realizing the orderly management of various cutting-related data, and laying a data foundation for the high efficiency of cutting execution.

[0015] 2. The cutting execution module in this invention generates an execution file by integrating the matching cutting control file through a honeycomb grid, realizing adaptive cutting of aluminum alloy workpieces based on visual inspection data. This improves the accuracy and adaptability of the cutting path execution. The feedback correction module obtains real-time data of the cutting area and judges operational anomalies. It generates feedback correction data as needed and synchronizes it to the honeycomb grid to update the visual inspection agent, forming a closed-loop control of visual inspection, cutting execution, feedback correction, and inspection optimization. While correcting cutting anomalies in a timely manner, it also optimizes the visual inspection process in reverse, continuously improving the cutting control accuracy and ensuring the stability of aluminum alloy cutting processing. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a system block diagram of the present invention.

[0018] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, an adaptive control system for aluminum alloy cutting based on vision detection is disclosed. The system includes: The visual inspection module is used to process the area to be cut into several inspection sub-regions, create and deploy a corresponding visual inspection agent in each inspection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding inspection sub-region. The visual inspection agent is registered to the honeycomb grid. The data processing module performs data parsing once for each visual detection agent based on the honeycomb grid, packages the parsing results into the corresponding cutting control file, and arranges the cutting control file in the honeycomb grid. The cutting execution module is used to set the target cutting path and upload the target cutting path to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions of itself, integrates them to generate the execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. The feedback correction module obtains real-time cutting data of the target cutting path in each cutting area, determines whether there is any abnormal operation in the current cutting area based on the real-time cutting data, and decides whether to generate feedback correction data for the cutting area based on the judgment result. The feedback correction data is synchronized to the honeycomb grid to update the corresponding visual detection agent.

[0021] It should be further explained that, in the specific implementation process, the area to be cut is divided into several inspection sub-regions. A corresponding visual inspection agent is created and deployed in each inspection sub-region. The process of visual inspection of the aluminum alloy workpiece under the corresponding inspection sub-region by the visual inspection agent includes: Identify the target object to be cut; the target object is an aluminum alloy workpiece. The area to be cut, which is the aluminum alloy workpiece to be used as the cutting plane, is determined. An industrial vision camera is deployed at the corresponding position on the cutting plane to obtain the field of view of the industrial vision camera. Based on the coverage area of ​​the field of view, the area to be cut is divided into several detection sub-regions. Several detection sub-regions are labeled, denoted as i, where i = 1, 2, 3, ..., n, and n is a natural number greater than 0. The coordinates of the region center corresponding to each detection sub-region are obtained, and the coordinates of the region center of the detection sub-region labeled i are marked as... ; in, = (x, y), with a corner point of the cutting plane containing the area to be cut as the origin to establish the coordinate axis. x is the horizontal distance of the corresponding detection sub-region relative to the origin, and y is the vertical distance of the corresponding detection sub-region relative to the origin.

[0022] Initialize a model architecture for controlling industrial vision cameras and allocate a database address to store the current model architecture. Based on the database address, complete the connection operation with the database and operate the model architecture to access and store data from the database. Based on the number of detection sub-regions, a number of index addresses are created for connecting to the database address. Each detection sub-region calls the database address based on its own index address, and then obtains the corresponding initial model architecture stored in the database through the database address. Set up several operation procedures for performing visual inspection, compile these operation procedures into corresponding operation scripts, store the operation scripts in the database, and associate the operation scripts with the database address; The initial model architecture is processed based on the process script. The initial model architecture is filled into the script content compiled from the process script and converted into the corresponding visual detection agent. The database address after the visual detection agent is created is updated, and the updated database address is synchronously associated with other index addresses. Based on the association between the updated database address and several index addresses, the visual detection agent is synchronized to several other index addresses, and then the corresponding visual detection agent is deployed for each detection sub-region. Each inspection sub-region is set up with an inspection time period. During the inspection time period, the corresponding industrial vision camera is controlled by the deployed vision inspection agent to complete a visual inspection of the aluminum alloy workpiece and obtain the visual inspection data of the corresponding inspection sub-region.

[0023] It should be further explained that, in the specific implementation process, the process of registering the visual inspection agent to the cellular internet grid includes: The cellular network consists of a number of cellular cells. Cells in adjacent positions communicate with each other. Each cellular cell is assigned a cellular address, and a registration pool for the corresponding cellular cell is created based on the cellular address. A corresponding listening thread is established for each registration pool. The listening thread is used to monitor all operations performed in the registration pool and determine whether any abnormal behavior occurs in the registration pool based on the data monitoring. The determination of abnormal behavior is based on the preset behavior operation procedure steps. Specifically, abnormal behavior includes a number of behavioral steps. Each behavioral step is compared with the operation procedure steps based on the execution order of the steps. If any behavioral step is inconsistent, it is determined that there is abnormal behavior in the registration pool; otherwise, it is determined that there is no abnormal behavior. If abnormal behavior occurs in the registration pool, any registration behavior that operates on the registration pool will be prohibited. When there is no abnormal behavior in the registration pool, the registration request initiated by any visual detection agent is received, the index address of the corresponding visual detection agent is bound to the honeycomb address of the currently unassigned honeycomb grid, the corresponding association address pair is constructed, and the registration of the visual detection agent on the honeycomb Internet grid is completed. The format of the associated address pairs is as follows: The associated address pair = <index address of the visual detection agent, honeycomb address of the honeycomb grid>; Each visual inspection agent interacts with the cellular grid through associated address pairs. The direction of the data interaction is from the location of the visual inspection agent to a certain cellular grid, or from the cellular grid to the location of a certain visual inspection agent.

[0024] It should be further explained that, in the specific implementation process, data parsing is performed once for each visual detection agent based on the honeycomb internet grid, and the parsing results are packaged into a corresponding cutting control file. The process of arranging the cutting control file in the honeycomb internet grid includes: Based on each associated address pair, a data parsing queue is constructed between each visual detection agent and the honeycomb grid. The associated address pair serves as the unique identifier of the data parsing queue. The data parsing script is edited and input into the data parsing queue. The data parsing script includes parsing fields corresponding to several execution sequences for performing data parsing on the visual inspection agent. Each parsing field is used to extract a subset of visual inspection data of the aluminum alloy workpiece from the visual inspection agent. For any visual inspection agent, based on all the parsing fields of the corresponding assigned data parsing script, perform a complete data parsing of the corresponding visual inspection agent. A complete data parsing includes several batch parsings in execution order. The visual inspection subset data corresponding to all batch parsings are integrated as the final parsing result of the corresponding data parsing. Set the packaging parameters, which consist of the packaging format, the packaging file path, and the packaging identifier. Use the parsing result of each data analysis as the cutting control file for the corresponding visual inspection agent. The cutting control file includes several cutting path segments when performing cutting operations on the aluminum alloy workpiece, the cutting frequency, cutting pressure, and cutting depth of each cutting segment, as well as the cutting direction corresponding to the connection of adjacent cutting path segments.

[0025] Each cutting control file is initially arranged in the honeycomb grid at the location of its respective visual detection agent until all cutting control files are initially arranged in the honeycomb grid of their respective honeycomb network. Construct corresponding direct honeycomb paths between adjacent honeycomb grids, select a honeycomb grid as the retrieval source, retrieve several honeycomb grids that match the current retrieval source, and set the honeycomb grids that match the current retrieval source as the same source grids. The correspondence between the honeycomb grid that serves as the retrieval source and the grid of the same source is defined as follows: the two can perform a complete cut on the aluminum alloy workpiece, that is, starting from the retrieval source as the path start point, the cut on the aluminum alloy workpiece is started until the path end point is reached, forming a corresponding cutting path. The grid of the same source is distributed on the corresponding cutting path, and the grid of the same source and the retrieval source are in adjacent or non-adjacent positions. Each cell other than the retrieval source and other cells directly adjacent to the retrieval source is treated as a path jump point. When the retrieval source matches all cells of the same source related to itself, it indirectly connects the direct honeycomb path of its own position with the path jump points of each cell of the same source to build a data chain on the same cutting path. All the cutting control files corresponding to the data chain are integrated to construct a corresponding control file set. Each retrieval source is bound to the corresponding control file set, and the control file set is stored in the location of the honeycomb grid corresponding to the retrieval source.

[0026] It should be further explained that, in the specific implementation process, the target cutting path is set and uploaded to the cellular grid. The cellular grid selects the cutting control files arranged at different positions within itself, integrates them to generate an execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. This process includes: Set the target cutting path, which consists of several sub-target cutting paths; Set the location of the target cutting path as the channel start point, set the location of the cellular grid as the channel end point, construct a data channel between the channel start point and the channel end point, and deploy an electronic fence for the data channel. The electronic fence consists of several sub-fence areas. A behavior situation set is created in each sub-fence area. The behavior situation set is used to define a number of abnormal access behaviors. Each abnormal access behavior occupies a behavior point in the corresponding sub-fence area. When an access behavior occurs in any sub-fence area of ​​the electronic fence, it is determined whether the access behavior conforms to an abnormal access behavior defined by the behavior status set corresponding to the current sub-fence area. If yes, then select a behavior point in the sub-fence area, and the behavior point corresponds to a region coordinate of the corresponding layout in the sub-fence area; if no, then do nothing. Each sub-fence area has the same area and is set to the same rectangular shape. Connecting lines are set between the corresponding behavior points of different sub-fence areas. The lower left corner of each sub-fence area is set as the origin of the coordinate system. The left and lower fence edges of the sub-fence are used as the y-axis and x-axis, respectively, to establish the fence coordinate axis of each sub-fence area. The behavior points in each sub-fence area that are at the same coordinate position relative to their respective fence coordinate axis are connected by a connecting line to build a corresponding cross-fence message channel. The cross-fence message channel is used to synchronize messages between different sub-fence areas. When a new behavior point is added to a sub-fence area, the abnormal access behavior of the newly added behavior point is synchronized to other sub-fence areas through the cross-fence message channel until all behavior points in each sub-fence area have been occupied. A number of sub-target cutting paths are assigned to different sub-fence areas. When the target cutting path is uploaded to the cellular network grid via the data channel, the electronic fence is activated. Each sub-fence area monitors the access environment of the corresponding assigned sub-target cutting path, locates the sub-fence areas with abnormalities, and performs repair operations on the abnormal access behavior of the corresponding sub-fence areas based on preset correction parameters. When all sub-target cutting paths are successfully uploaded to the honeycomb grid, each sub-target cutting path is adapted to the honeycomb grids arranged at different positions on the honeycomb grid. The control file set corresponding to the successfully adapted honeycomb grid is selected to obtain all the cutting control files included in the control file set. All obtained cutting control files are set as execution files for the target cutting path. Adaptive cutting is then performed on the aluminum alloy workpiece based on these files. Specifically, after the execution files are sent to the pre-deployed cutting equipment, the equipment initiates the adaptive cutting process, monitors the cutting conditions and workpiece status in real time, and dynamically adjusts cutting parameters to ensure cutting quality and adapt to the material characteristics of the aluminum alloy. This includes pre-cutting preparation and initialization: after receiving the execution files, the cutting equipment automatically completes initialization, adjusting initial parameters such as the cutting head position, laser power (or water jet pressure), and auxiliary gas flow rate according to the parameters in the execution files; simultaneously, it performs adaptive cutting on the aluminum alloy workpiece. The alloy workpiece undergoes secondary positioning. An industrial vision camera performs visual recognition and laser contour positioning to detect the workpiece's actual position and flatness. If any deviation exists, the cutting path is automatically adjusted to ensure alignment with the workpiece's actual position. This includes real-time monitoring of the cutting start process: the cutting equipment starts the cutting operation according to the path and parameters recorded in the execution file. During the cutting process, various sensors, such as temperature sensors, displacement sensors, and light signal sensors, monitor key data in real time, including the temperature of the cutting area, workpiece deformation, the distance between the cutting head and the workpiece, the flatness of the cut surface, and laser reflection signals, and provide real-time feedback of the monitoring data. It also includes adaptive parameter adjustments during the cutting process, such as temperature adaptive adjustment: if the temperature of the cutting area is detected to be higher than the preset temperature range, the cutting speed is automatically reduced and the auxiliary gas flow rate is increased to accelerate heat dissipation and avoid problems such as workpiece deformation and burrs on the cutting surface; if the temperature is lower than the preset temperature range, resulting in incomplete cutting, the laser power is increased to ensure that the molten metal can be blown away from the kerf in time; position adaptive adjustment: if slight deformation of the workpiece is detected, or the distance between the cutting head and the workpiece deviates from the preset value, the cutting head position and tool path are automatically adjusted to compensate for deformation errors and ensure that the cutting path always conforms to the preset contour of the workpiece; for focal point offset during the cutting process, the focal point position is adjusted in real time to ensure that the cutting head is always at the optimal focal length and improve the smoothness of the cut; parameter adaptive optimization: the cutting parameters are dynamically adjusted according to the actual thickness deviation and surface condition changes of the aluminum alloy workpiece. For example, when encountering areas with uneven workpiece thickness, the matching ratio of laser power and cutting speed is automatically adjusted; if slag is detected on the cutting surface, the auxiliary gas pressure and focal point position are adjusted to reduce slag formation.

[0027] It should be further explained that, in the specific implementation process, the process of obtaining real-time cutting data of the target cutting path in each cutting area, and determining whether there are any abnormal operations in the current cutting area based on the real-time cutting data, includes: Obtain the cutting range of each sub-target cutting path corresponding to the target cutting path on the target object, and determine them as their respective cutting areas. Record all operation data executed by the cutting device in each cutting area as their respective real-time cutting data. Set up a cutting standard dataset, which includes cutting standard data corresponding to several categories of cutting behaviors under the standard execution procedure. Input the real-time cutting data of each sub-target cutting path and the cutting standard dataset into the pre-built bag-of-words model. The bag-of-words model compares and determines whether there are abnormal operations in each cutting region. The cutting behavior executed by the cutting device in each cutting area is determined, and the real-time cutting data of the current cutting behavior is compared with the cutting standard data. The comparison is based on the bag-of-words model. The initial bag-of-words model is constructed based on historical cutting data generated in the historical cutting process, and the initial bag-of-words model is trained and optimized through big data analysis technology. When the comparison results between the real-time cutting data and the cutting standard data are inconsistent, it is determined that there is an abnormal operation in the cutting area corresponding to the real-time cutting data; when the comparison results between the real-time cutting data and the cutting standard data are consistent, it is determined that there is no abnormal operation.

[0028] It should be further explained that, in the specific implementation process, the process of determining whether to generate feedback correction data for the segmented region based on the judgment result, and synchronizing the feedback correction data to the cellular grid to update the corresponding visual detection agent includes: If no abnormal operation is detected in the cutting area, no operation is performed in the corresponding cutting area; if an abnormal operation is detected in the cutting area, corresponding feedback correction data is generated based on the real-time cutting data and standard cutting data of the corresponding cutting area. Set a data synchronization point at the location of the cut area where abnormal operation occurs, establish a synchronization operation queue between the current data synchronization point and a corresponding honeycomb grid, and synchronize the current feedback correction data to the honeycomb grid through the synchronization operation queue. Based on the feedback correction data, the process script corresponding to the current location of the honeycomb grid is expanded and compiled, thereby updating the visual inspection agent stored at the location of the honeycomb grid. The update result of the current visual inspection agent is sent to other locations of the honeycomb grid via data link, and the visual inspection agents at other locations of the honeycomb grid are optimized based on the update result.

[0029] It should be noted that the updated and optimized vision inspection agent overcomes the abnormal operations that occur when the cutting equipment performs cutting on the target cutting path of the aluminum alloy workpiece. Through continuous iteration of the vision inspection agent, the cutting operation of the aluminum alloy workpiece can be completed more accurately, realizing adaptive cutting control of aluminum alloy.

[0030] Please see Figure 2 As shown, an adaptive control method for an aluminum alloy cutting adaptive control system based on vision detection includes the following steps: Step S1: Divide the area to be cut into several detection sub-regions, create and deploy a corresponding visual inspection agent in each detection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding detection sub-region. Register the visual inspection agent to the honeycomb grid. Step S2: Perform data parsing once for each visual detection agent based on the honeycomb internet grid, and package the parsing results into the corresponding cutting control file, and arrange the cutting control file in the honeycomb internet grid; Step S3: Set the target cutting path and upload it to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions, integrates them to generate an execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. Step S4: Obtain real-time cutting data of the target cutting path in each cutting area, determine whether there is abnormal operation in the current cutting area based on the real-time cutting data, decide whether to generate feedback correction data for the cutting area based on the judgment result, and synchronize the feedback correction data to the honeycomb Internet grid to update the corresponding visual detection agent.

[0031] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A vision-based adaptive control system for aluminum alloy cutting, characterized in that, The system includes: The visual inspection module is used to process the area to be cut into several inspection sub-regions, create and deploy a corresponding visual inspection agent in each inspection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding inspection sub-region. The visual inspection agent is registered to the honeycomb grid. The data processing module performs data parsing once for each visual detection agent based on the honeycomb grid, packages the parsing results into the corresponding cutting control file, and arranges the cutting control file in the honeycomb grid. The cutting execution module is used to set the target cutting path and upload the target cutting path to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions of itself, integrates them to generate the execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. The feedback correction module obtains real-time cutting data of the target cutting path in each cutting area, determines whether there is any abnormal operation in the current cutting area based on the real-time cutting data, and decides whether to generate feedback correction data for the cutting area based on the judgment result. The feedback correction data is synchronized to the honeycomb grid to update the corresponding visual detection agent.

2. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 1, characterized in that, The area to be cut is divided into several inspection sub-regions. A corresponding visual inspection agent is created and deployed in each inspection sub-region. The process of visual inspection of the aluminum alloy workpiece in the corresponding inspection sub-region by the visual inspection agent includes: The aluminum alloy workpiece is identified as the target object that needs to be cut. Determine the area to be cut of the target object, deploy an industrial vision camera, divide the area to be cut into several detection sub-regions based on the field of view, initialize a model architecture to control the industrial vision camera, and allocate a database address to store the model architecture for connection operations with the database. Based on the number of detection sub-regions, create a corresponding number of index addresses for connecting to the database address. Operate the index addresses and database addresses to deploy a visual inspection agent for each detection sub-region. Based on the visual inspection agent, control their respective industrial vision cameras to complete a visual inspection of the aluminum alloy workpiece and obtain the visual inspection data of their respective detection sub-regions.

3. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 2, characterized in that, The process of deploying a visual detection agent for each detection sub-region by manipulating the index address and database address includes: Each detection sub-region calls the database address based on its own index address, obtains the initial model architecture stored in the database through the database address, sets the process script for performing vision and stores it in the database, and associates the process script with the database address; Based on the process script processing of the initial model architecture, the initial model architecture is converted into a visual inspection agent by filling in the script content after the process script is compiled. The database address after the visual inspection agent is created is updated, and the association between the updated database address and other index addresses is established synchronously. The visual inspection agent is synchronized to several other index addresses.

4. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 3, characterized in that, The process of registering a visual inspection agent to a cellular internet grid includes: The cellular network consists of a number of cellular cells. Adjacent cellular cells communicate with each other. Each cellular cell is assigned a cellular address, and a registration pool for cellular cells is created based on the cellular addresses. A listening thread is established for each registration pool to monitor all operations performed within the registration pool and determine whether any abnormal behavior occurs within the registration pool based on the data monitoring. If abnormal behavior occurs in the registration pool, any registration behavior that operates on the registration pool will be prohibited. When there is no abnormal behavior in the registration pool, the system receives a registration request from any visual detection agent, binds the index address of the corresponding visual detection agent to the cellular address of the currently unassigned cellular grid, constructs an associated address pair, and completes the registration of the visual detection agent on the cellular internet grid.

5. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 4, characterized in that, The process of performing data parsing once for each visual detection agent based on a honeycomb internet grid, and packaging the parsing results into the corresponding cutting control file includes: A data parsing queue is constructed between the visual inspection agent and the honeycomb grid. The data parsing script is edited and input into the data parsing queue. The data parsing script includes several parsing fields that execute data parsing on the visual inspection agent in an order of execution. Each parsing field is used to extract a subset of visual inspection data of the aluminum alloy workpiece by the visual inspection agent. For any visual inspection agent, based on all the assigned parsing fields, perform a complete data parsing of the corresponding visual inspection agent once. A complete data parsing includes several batch parsings in execution order. Integrate the visual inspection subset data from all batch parsings as the final parsing result. Set packaging parameters to use the parsing results of each data analysis as the cutting control file for the corresponding visual detection agent.

6. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 5, characterized in that, The process of arranging the control file within a honeycomb grid includes: Each cutting control file is initially arranged into the honeycomb grid of its respective visual detection agent. A direct honeycomb path is constructed between honeycomb grids in adjacent positions. A honeycomb grid is selected as the retrieval source. Honeycomb grids that match the retrieval source are retrieved and set as the same source grids. Set the correspondence between the retrieval source and the same source grid, and set the honeycomb grid as the path jump point. When the retrieval source matches all the same source grids related to itself, it indirectly connects the honeycomb direct path of its own position with the path jump point of each same source grid to build a data chain in the same cutting path. Integrate all the cutting control files corresponding to the data chain, build a control file set, bind each retrieval source to the control file set, and store the control file set in the honeycomb grid of the retrieval source.

7. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 6, characterized in that, The process of setting a target cutting path and uploading it to the cellular grid, where the cellular grid selects cutting control files arranged at different locations within itself, integrates them to generate an execution file that executes the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece includes: The target cutting path consists of several sub-target cutting paths. A data channel is established, and an electronic fence is deployed. The electronic fence consists of several sub-fence areas. A behavior situation set is created in each sub-fence area. The behavior situation set is used to define several abnormal access behaviors. When an access behavior occurs in any sub-fence area of ​​the electronic fence, it is determined whether the access behavior is an abnormal access behavior that matches the behavior status set of the current sub-fence area. If it is, a behavior point is occupied in the sub-fence area; otherwise, no operation is performed. Set up communication lines between corresponding behavior points in different sub-fence areas, connect the behavior points in the sub-fence areas through the communication lines, and build a cross-fence message channel to synchronize messages between different sub-fence areas. When all the sub-target cutting paths are uploaded to the honeycomb grid, each sub-target cutting path is adapted to the honeycomb grids arranged at different positions on the honeycomb grid, and the control file set of the successfully adapted honeycomb grids is selected to obtain the corresponding complete cutting control files. Set the cutting control file as the execution file for the target cutting path, and perform adaptive cutting on the aluminum alloy workpiece based on the execution file.

8. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 7, characterized in that, The process of obtaining real-time cutting data of the target cutting path in each cutting region, and determining whether there are abnormal operations in the current cutting region based on the real-time cutting data, includes: The cutting range of the sub-target cutting path on the target object is determined as the corresponding cutting area, and all operation data performed by the cutting device in the cutting area is recorded as real-time cutting data. Set up a standard cutting dataset, which includes standard cutting data for several categories of cutting behaviors under standard execution procedures. Input the real-time cutting data and the standard cutting dataset into a pre-built bag-of-words model. The bag-of-words model compares and determines whether there are any abnormal operations in each cutting region. When the comparison results are inconsistent, it is determined that there is an abnormal operation in the corresponding cutting area; when the comparison results are consistent, it is determined that there is no abnormal operation.

9. The adaptive control system for aluminum alloy cutting based on vision detection according to claim 8, characterized in that, The process of determining whether to generate feedback correction data for the cut region based on the judgment result, and synchronizing the feedback correction data to the cellular network grid to update the corresponding visual detection agent includes: If no abnormal operation is detected in the cutting area, no operation is performed. When an abnormal operation is detected in the cutting area, corresponding feedback correction data is generated based on the real-time cutting data and standard cutting data of the corresponding cutting area. In the segmented area where abnormal operations occur, a data synchronization point is set up, and a synchronization operation queue is established between the data synchronization point and a corresponding cellular grid. The feedback correction data is synchronized to the cellular grid through the synchronization operation queue. Based on the feedback correction data, the process script at the current location of the honeycomb grid is expanded and compiled, the visual inspection agent stored at the honeycomb grid is updated, the update result of the visual inspection agent is sent to other honeycomb grids via data link, and the visual inspection agent at other honeycomb grids is optimized based on the update result.

10. An adaptive control method for an adaptive control system for aluminum alloy cutting based on vision detection, as described in any one of claims 1 to 9, characterized in that, Includes the following steps: Step S1: Divide the area to be cut into several detection sub-regions, create and deploy a corresponding visual inspection agent in each detection sub-region, and have the visual inspection agent perform visual inspection of the aluminum alloy workpiece in the corresponding detection sub-region. Register the visual inspection agent to the honeycomb grid. Step S2: Perform data parsing once for each visual detection agent based on the honeycomb internet grid, and package the parsing results into the corresponding cutting control file, and arrange the cutting control file in the honeycomb internet grid; Step S3: Set the target cutting path and upload it to the honeycomb grid. The honeycomb grid selects the cutting control files arranged at different positions, integrates them to generate an execution file for executing the target cutting path, and performs adaptive cutting of the aluminum alloy workpiece. Step S4: Obtain real-time cutting data of the target cutting path in each cutting area, determine whether there is abnormal operation in the current cutting area based on the real-time cutting data, decide whether to generate feedback correction data for the cutting area based on the judgment result, and synchronize the feedback correction data to the honeycomb Internet grid to update the corresponding visual detection agent.

Images