Welding data automatic processing system and method

Abstract

The invention belongs to the technical field of welding data processing, and discloses a welding data automatic processing system and method. The method comprises the steps that semaphore mechanisms are set for N welding robots within unit time, interaction is conducted, and a semaphore mechanism interaction result set is obtained; constructing an interaction failure matrix based on the semaphore mechanism interaction result set; based on the interaction failure matrix, interaction feature data of the N welding robots are obtained; collecting network operation data in the working environment of the welding robot in unit time; processing the network operation data to obtain network feature data; inputting the interaction feature data and the network feature data of the N welding robots into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result; performing intelligent self-adaptive processing on the welding data based on a network fluctuation diagnosis result; according to the scheme, a welding data dynamic adjustment mechanism based on the network fluctuation state is provided, and the self-adaptability of the network fluctuation of the welding data is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of welding data processing, more particularly, to a welding data automatic processing system and method. BACKGROUND

[0002] As a key process in modern manufacturing, welding technology plays a vital role in industrial production. In recent years, with the rapid development of automation and intelligent technology, intelligent welding processing technology based on point cloud data has emerged as the key means to improve welding quality and efficiency.

[0003] In actual industrial application scenarios, such as the manufacturing process of electric vehicle related accessories, welding data intelligent processing faces many challenges. With the wide application of industrial internet of things and cloud computing technology in the welding field, the transmission and processing of welding data increasingly rely on networks. However, network fluctuations inevitably exist in industrial network environments. When network fluctuations occur, the welding data being transmitted will be interrupted, out of order or lost, making it impossible for the receiving end to obtain complete and accurate data, thereby affecting the subsequent point cloud data processing flow and ultimately negatively impacting welding quality and production efficiency.

[0004] Currently, the existing solutions to the above problems have obvious deficiencies. In response to network fluctuations, traditional methods mostly use simple retransmission mechanisms. This approach not only has low efficiency when network fluctuations are frequent, but also can cause long-term interruption of data processing, making it difficult to meet the strict requirements of industrial production for real-time and stability. Therefore, there is an urgent need for an innovative welding data automatic processing method that can effectively overcome the point cloud data processing challenges in complex environments while having strong network fluctuation adaptive ability to improve the intelligent level of the welding process and meet the needs of modern manufacturing for high-quality and efficient welding. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purposes, the present application provides the following technical solutions: a welding data automatic processing method, comprising:

[0006] Setting up a semaphore mechanism for N welding robots in a unit of time and interacting to obtain a semaphore mechanism interaction result set;

[0007] Constructing an interaction failure matrix based on the semaphore mechanism interaction result set;

[0008] Obtaining interaction feature data of the N welding robots based on the interaction failure matrix;

[0009] Collecting network running data in the welding robot working environment in a unit of time;

[0010] Processing the network operation data to obtain network feature data;

[0011] Inputting the interaction feature data of the N welding robots and the network feature data into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result;

[0012] Constructing the welding data into a welding data bounding box; obtaining the octree recursion number corresponding to the network fluctuation diagnosis result from a pre-constructed network fluctuation diagnosis result-octree recursion number matching table; the network fluctuation diagnosis result-octree recursion number matching table comprises network fluctuation diagnosis results and corresponding octree recursion numbers; performing octree recursive division on the welding data bounding box according to the octree recursion number to obtain the recursively divided welding data; and the welding robot performs data transmission according to the recursively divided welding data.

[0013] Further, the method for constructing the welding data into a welding data bounding box comprises:

[0014] Obtaining three-dimensional coordinate data of the welding data, and obtaining the X-axis minimum coordinate, the X-axis maximum coordinate, the Y-axis minimum coordinate, the Y-axis maximum coordinate, the Z-axis minimum coordinate and the Z-axis maximum coordinate from the three-dimensional coordinate data;

[0015] Obtaining the length of the welding data bounding box by subtracting the X-axis minimum coordinate from the X-axis maximum coordinate, obtaining the width of the welding data bounding box by subtracting the Y-axis minimum coordinate from the Y-axis maximum coordinate, and obtaining the height of the welding data bounding box by subtracting the Z-axis minimum coordinate from the Z-axis maximum coordinate; and constructing the mean value of the X-axis maximum coordinate and the X-axis minimum coordinate, the mean value of the Y-axis maximum coordinate and the Y-axis minimum coordinate, and the mean value of the Z-axis maximum coordinate and the Z-axis minimum coordinate into the center coordinate of the welding data bounding box;

[0016] Constructing the welding data bounding box according to the center coordinate, the length, the width and the height of the welding data bounding box.

[0017] Further, the method for obtaining the recursively divided welding data comprises:

[0018] S300: Let the octree recursion number be R, and the initial value of r be 1, and the value range of r be 1 to R;

[0019] S301: If r is equal to 1, create a recursive result set, evenly divide the welding data bounding box into eight welding data sub-bounding boxes, and add the divided welding data sub-bounding boxes to the recursive result set; if r is greater than 1, evenly octree divide each welding data sub-bounding box in the recursive result set, i.e., evenly divide each welding data sub-bounding box into eight new welding data sub-bounding boxes again, and cover the divided new welding data sub-bounding boxes to the recursive result set;

[0020] S302: Let r=r+1, if r is less than or equal to R, continue to execute S301; if r is greater than R, the recursive result set is the welding data after recursive division, and the current process is ended.

[0021] Further, the acquisition method of the semaphore mechanism interaction result set comprises:

[0022] Step one: preset a communication interval time, the number of times of semaphore mechanism interaction in a unit time according to the communication interval time is recorded as CL; let the initial value of cl be 1, and the value range of cl be 1 to CL;

[0023] Step two: for N welding robots, a corresponding semaphore set is created for each robot; the semaphore set is initialized; the semaphore set includes an interaction request semaphore, an interaction response semaphore and an interaction failure semaphore;

[0024] Step three: the N welding robots are constructed into a set of robots to be interacted; based on the set of robots to be interacted, semaphore mechanism interaction is carried out to obtain the semaphore mechanism interaction result of the clth time of semaphore mechanism interaction;

[0025] Step four: let cl=cl+1, if cl is less than or equal to CL, continue to execute steps two to three; if cl is greater than CL, the CL semaphore mechanism interaction results are constructed into a semaphore mechanism interaction result set, and the current process is ended.

[0026] Further, the acquisition method of the semaphore mechanism interaction result of the clth time of semaphore mechanism interaction comprises:

[0027] Let the initial value of i be 1, and the value range of i be 1 to ; Indicates rounding down;

[0028] A welding robot is randomly selected from the set of robots to be interacted as the ith interaction request robot, and the ith interaction request robot is removed from the set of robots to be interacted; a welding robot is randomly selected from the set of robots to be interacted as the semaphore interaction object of the ith interaction request robot, and the semaphore interaction object is removed from the set of robots to be interacted;

[0029] The welding robot number, semaphore initiation timestamp and randomly generated semaphore verification code of the ith interaction request robot are constructed into semaphore communication content; the semaphore communication content is sent to the semaphore interaction object, and the semaphore verification code and the robot number of the semaphore interaction object are uploaded to the sentinel server; the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 1;

[0030] When the semaphore interaction object of the ith interaction request robot receives the semaphore communication content of the ith interaction request robot, the interaction completion time is recorded, and the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 1;

[0031] Based on the interaction completion time and the semaphore verification code, the corresponding semaphore set is updated, and the updated semaphore set of the ith interaction request robot and the semaphore interaction object is constructed into an interaction semaphore set;

[0032] Let i = i + 1, and repeat the above process, when i is greater than , the interaction semaphore set is constructed into the semaphore mechanism interaction result of the clth semaphore mechanism interaction, and the current process is ended.

[0033] Further, the method for updating the corresponding semaphore set based on the interaction completion time and the semaphore verification code comprises:

[0034] If the interaction completion time of the ith interaction request robot and the semaphore interaction object is greater than or equal to the preset time interval threshold, it is determined that the semaphore transmission fails this time, the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 2;

[0035] If the interaction completion time of the ith interaction request robot and the semaphore interaction object is less than the preset time interval threshold, the semaphore interaction object obtains the corresponding semaphore verification code from the sentinel server, and compares the semaphore verification code with the semaphore verification code in the semaphore communication content; if the comparison result is consistent, it is determined that the semaphore transmission succeeds this time, the semaphore value of the interaction request semaphore of the corresponding semaphore interaction object is set to 2, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 2, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 1; if the comparison result is inconsistent, it is determined that the semaphore transmission fails this time, the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 2.

[0036] Further, the method for constructing the interaction failure matrix comprises:

[0037] S100: setting an initial value of cl as 1, a value range of cl as 1 to CL, CL as a number of times of interacting with the semaphore mechanism, pre-constructing an interaction failure matrix, the interaction failure matrix having N rows, CL columns, a row as a number of the welding robot, and a column as a number of times of interacting with the semaphore mechanism, and initializing all values in the interaction failure matrix as 0;

[0038] S101: obtaining a cl-th semaphore mechanism interaction result from the set of semaphore mechanism interaction results, and setting an initial value of sl as 1, a value range of sl as 1 to SL, SL as a number of the set of interaction semaphores in the cl-th semaphore mechanism interaction result;

[0039] S102: obtaining a sl-th set of interaction semaphores from the semaphore mechanism interaction result, extracting a number of the welding robot and a corresponding interaction failure semaphore from the sl-th set of interaction semaphores, and updating a matrix element in the interaction failure matrix at a position corresponding to the number of the welding robot in a row and cl in a column to a semaphore value of the interaction failure semaphore;

[0040] S103: setting sl = sl + 1, if sl is less than or equal to SL, continuing to execute S102, if sl is greater than SL, setting cl = cl + 1, if cl is less than or equal to CL, continuing to execute S101 to S102, if cl is greater than CL, obtaining the interaction failure matrix, and ending the current process.

[0041] Further, the method for obtaining the interaction feature data comprises:

[0042] S200: setting an initial value of n as 1, a value range of n as 1 to N;

[0043] S201: obtaining n-th matrix data from the interaction failure matrix, counting a number of semaphore values of the interaction failure semaphore being 3 in the n-th matrix data as an interaction failure number, and calculating the interaction failure rate, the interaction failure rate dispersion degree, and the time sequence interaction fluctuation index of the n-th welding robot according to the interaction failure number and the number of times of interacting with the semaphore mechanism;

[0044] S202: constructing the interaction failure rate, the interaction failure rate dispersion degree, and the time sequence interaction fluctuation index of the n-th welding robot into the interaction feature data of the n-th welding robot;

[0045] S203: setting n = n + 1, if n is less than or equal to N, continuing to execute S201 to S202, if n is greater than N, obtaining the interaction feature data of the N welding robots, and ending the current process.

[0046] Further, the method for obtaining the network feature data comprises:

[0047] Pre-set network interval time, divide the unit time into JG network operation data sampling points according to the network interval time;

[0048] According to the bandwidth utilization rate of the JG network operation data sampling points, the bandwidth utilization rate fluctuation degree is calculated; according to the network delay of the JG network operation data sampling points, the average network delay and the network delay standard deviation are calculated; according to the TCP retransmission rate and the data transmission rate of the JG network operation data sampling points, the link stability index is calculated;

[0049] The bandwidth utilization rate fluctuation degree, the average network delay, the network delay standard deviation and the link stability index are constructed into network feature data.

[0050] A welding data automatic processing system for realizing the welding data automatic processing method, comprising:

[0051] The signal interaction module sets up the semaphore mechanism for N welding robots in a unit time and interacts to obtain a semaphore mechanism interaction result set;

[0052] The matrix construction module constructs an interaction failure matrix based on the semaphore mechanism interaction result set;

[0053] The feature extraction module obtains the interaction feature data of the N welding robots based on the interaction failure matrix;

[0054] The data acquisition module is used for acquiring network operation data in the welding robot working environment in a unit time;

[0055] The data processing module is used for processing the network operation data to obtain network feature data;

[0056] The fluctuation diagnosis module is used for inputting the interaction feature data of the N welding robots and the network feature data into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result;

[0057] The intelligent adjustment module is used for constructing the welding data into a welding data bounding box; obtaining the octree recursion number corresponding to the network fluctuation diagnosis result from a pre-constructed network fluctuation diagnosis result-octree recursion number matching table; the network fluctuation diagnosis result-octree recursion number matching table comprises a network fluctuation diagnosis result and a corresponding octree recursion number; according to the octree recursion number, the welding data bounding box is octree recursively divided to obtain the recursively divided welding data; the welding robot performs data transmission according to the recursively divided welding data.

[0058] Compared with the prior art, the welding data automatic processing system and method have the following technical effects and advantages:

[0059] The scheme combines the semaphore interaction mechanism and network feature data collection to construct an intelligent network fluctuation diagnosis system, which can accurately identify network fluctuations of different degrees, including no fluctuation, slight fluctuation, moderate fluctuation and severe fluctuation. By comprehensively analyzing the interaction failure rate of the welding robot, the interaction failure rate dispersion degree and the time sequence interaction fluctuation index, and combining with network features such as bandwidth utilization rate, network delay, TCP retransmission rate and data transmission rate, the method can accurately diagnose the network state and ensure the accuracy and robustness of network fluctuation prediction.

[0060] The application adopts octree data structure to adaptively and recursively divide welding data, and dynamically adjusts the recursion times of octree in combination with network fluctuation diagnosis results, so as to adapt the data division precision to the network state. When there is no network fluctuation, the welding robot receives complete and high-precision point cloud data; when the network fluctuation intensifies, the layer number of octree division is reduced to reduce the data amount, and the welding task is ensured not to be interrupted. This welding data dynamic adjustment mechanism based on the network fluctuation state can ensure the welding precision while reducing unnecessary data transmission, improve the adaptability of welding data to network fluctuation, and effectively improve the stability of data transmission, avoiding the decrease of welding precision caused by unstable network.

[0061] In summary, the application provides a welding data processing scheme with adaptability, intelligence and high efficiency, which can ensure efficient transmission and processing of data in a complex industrial network environment, improve the operation stability of the welding robot under different network conditions, and provide technical support for efficient operation of the intelligent welding system. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 It is a welding data automatic processing system schematic diagram of embodiment 1 of the application;

[0063] Figure 2 It is a welding data automatic processing method flow chart of embodiment 3 of the application;

[0064] Figure 3 It is a welding data automatic processing system schematic diagram of embodiment 2 of the application;

[0065] Figure 4 It is a method flow chart for constructing welding data into a welding data bounding box;

[0066] Figure 5 It is a welding robot network connection schematic diagram based on semaphore interaction;

[0067] Figure 6 It is a network performance trend schematic diagram. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present application will be described in detail, clearly and completely below with reference to the drawings in the embodiments of the present application. It should be particularly noted that the specific embodiments described below are only used to better illustrate and describe the technical solutions of the present application, and are intended to enable those skilled in the art to better understand and implement the present application, and should not be understood as limiting the protection scope of the present application. Those skilled in the art can modify, adjust or equivalently replace the present application according to the content disclosed in the present application without departing from the spirit and essence of the present application, and these should be regarded as the protection scope of the present application.

[0069] Embodiment 1

[0070] Please refer to Figure 1 As shown in the figure, the embodiment discloses a welding data automatic processing system, which comprises a signal interaction module, a matrix construction module, a feature extraction module, a data acquisition module, a data processing module, a fluctuation diagnosis module and an intelligent adjustment module, each module is connected through wired and / or wireless connection to realize data transmission.

[0071] The signal interaction module sets up a semaphore mechanism for N welding robots in a unit time and interacts to obtain a semaphore mechanism interaction result set.

[0072] It should be noted that the welding robot network connection schematic diagram based on semaphore interaction is as shown in Figure 5 , wherein the first welding robot 001, the second welding robot 002 and the third welding robot 003 transmit welding data to the computer terminal 006 through the first router 004 and the second router 005.

[0073] The acquisition method of the semaphore mechanism interaction result set comprises:

[0074] Step one: pre-set a communication interval time, the number of times of semaphore mechanism interaction in a unit time according to the communication interval time is recorded as CL; let the initial value of cl be 1, and the value range of cl be 1 to CL;

[0075] For example, in the technical solutions of the present application, under the condition of factory gigabit Ethernet, the value range of the communication interval time is [20ms, 80ms], and the communication interval time can be set to 60ms. Under the condition of industrial Wi-Fi, the value range of the communication interval time is [80ms, 100ms], and the communication interval time can be set to 90ms.

[0076] Step two: for N welding robots, create a corresponding semaphore set for each robot; initialize the semaphore set; the semaphore set contains an interaction request semaphore, an interaction response semaphore, and an interaction failure semaphore; initializing the semaphore set means setting the initial values of the interaction request semaphore, the interaction response semaphore, and the interaction failure semaphore to 0;

[0077] It should be noted that the interaction request semaphore is used to mark the state of the welding robot initiating an interaction request, the initial value of the interaction request semaphore is 0, indicating that there is currently no interaction request; when the welding robot interacts, the semaphore value of the interaction request semaphore is 1; when the welding robot interacts successfully, the semaphore value of the interaction request semaphore is 2; when the welding robot fails to interact, the semaphore value of the interaction request semaphore is 3.

[0078] The interaction response semaphore is used to record the response state of the welding robot after receiving an interaction request, the initial value of the interaction response semaphore is 0, indicating that there is currently no interaction request; when the welding robot receives an interaction request, the semaphore value of the interaction response semaphore is 1; when the welding robot receives an interaction request and interacts successfully, the semaphore value of the interaction response semaphore is 2; when the welding robot receives an interaction request and fails to interact, the semaphore value of the interaction response semaphore is 3.

[0079] The interaction failure semaphore is used to record whether the semaphore mechanism interaction fails, the initial value of the interaction failure semaphore is 0, indicating that there is currently no interaction request; 1 indicates that the interaction is successful, and 2 indicates that the interaction fails.

[0080] Step three: construct N welding robots into a set of robots to be interacted; based on the set of robots to be interacted, perform semaphore mechanism interaction to obtain the semaphore mechanism interaction result of the cl-th semaphore mechanism interaction;

[0081] Step four: let cl = cl + 1, if cl is less than or equal to CL, continue to execute steps two to three; if cl is greater than CL, construct the CL semaphore mechanism interaction results into a semaphore mechanism interaction result set, and end the current process.

[0082] The method for obtaining the semaphore mechanism interaction result of the cl-th semaphore mechanism interaction includes:

[0083] Let the initial value of i be 1, and the value range of i be 1 to ; Indicates rounding down;

[0084] Randomly select a welding robot from the set of robots to be interacted with, and use it as the i-th interaction request robot. Remove the i-th interaction request robot from the set of robots to be interacted with. Randomly select a welding robot from the set of robots to be interacted with, and use it as the semaphore interaction object of the i-th interaction request robot. Remove the semaphore interaction object of the i-th interaction request robot from the set of robots to be interacted with.

[0085] Construct the semaphore communication content from the welding robot number of the i-th interactive request robot, the semaphore initiation timestamp, and the randomly generated semaphore verification code; send the semaphore communication content to the selected semaphore interaction object, and upload the semaphore verification code and the robot number of the semaphore interaction object to the sentinel server; set the semaphore value of the interactive request semaphore of the i-th interactive request robot to 1;

[0086] When the semaphore interaction object of the i-th interaction request robot receives the semaphore communication content of the i-th interaction request robot, the interaction completion time is recorded. The interaction completion time refers to the total time elapsed from the semaphore initiation timestamp to the time when the semaphore interaction object receives the semaphore communication content. The semaphore value of the interaction response semaphore of the semaphore interaction object is set to 1.

[0087] The corresponding semaphore set is updated based on the interaction completion time and semaphore verification code. The updated semaphore set of the i-th interaction request robot and the semaphore interaction object is constructed into an interaction semaphore set.

[0088] Let i = i + 1, and repeat the above process until i is greater than 1. At that time, the cl-th semaphore mechanism interaction is completed, and The set of interactive semaphores is used to construct the semaphore mechanism interaction result of the cl-th semaphore mechanism interaction, and the current process ends.

[0089] Methods for updating the corresponding semaphore set based on interaction completion time and semaphore verification code include:

[0090] If the interaction completion time between the i-th interaction request robot and the semaphore interaction object is greater than or equal to the preset time interval threshold, then the semaphore transmission is determined to have failed. The semaphore value of the interaction request semaphore of the i-th interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the i-th interaction request robot and the semaphore interaction object is set to 2.

[0091] If the interaction completion time of the ith interaction request robot and the semaphore interaction object is less than the preset time interval threshold, the semaphore interaction object obtains the corresponding semaphore verification code from the sentinel server, and compares the semaphore verification code with the semaphore verification code in the semaphore communication content; if the comparison result is consistent, it is determined that the semaphore transmission is successful, the semaphore value of the interaction request semaphore of the corresponding semaphore interaction object is set to 2, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 2, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 1; if the comparison result is inconsistent, it is determined that the semaphore transmission fails, the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 2.

[0092] For example, in the technical solution of the present application, under the condition of the factory gigabit Ethernet, the time interval threshold value is in the range of [40ms, 120ms], and the time interval threshold value can be set to 80ms. Under the condition of industrial Wi-Fi, the time interval threshold value is in the range of [150ms, 300ms], and the time interval threshold value can be set to 200ms.

[0093] It should be noted that in the industrial network environment, there are external malicious attacks, for example, hackers try to interfere with the communication of the welding robot to destroy the production process or steal sensitive information. If there is a fixed rule in the selection of the communication object between the welding robots, the attacker can predict the next communication object by analyzing the communication mode, and then carry out targeted attacks, such as implementing man-in-the-middle attacks, interfering with communication data, etc. Randomly selecting the communication object makes the communication mode unpredictable, and it is difficult for the attacker to find out the rule, greatly increasing the difficulty of the attack, thereby improving the security of the entire communication system. For some welding robot communication involving critical production data and process parameters, if the communication object selection has a rule, it may be easier for the lawbreaker to locate the key node of data transmission, thereby increasing the risk of data leakage. Randomly selecting the communication object can make the data transmitted between different robots, reducing the possibility of a specific node being long-term focused and attacked, effectively protecting the security and privacy of the data.

[0094] In addition, different welding robots are distributed in different areas in terms of physical location, and the communication paths of the welding robots are also different. Randomly selecting the communication object can make the communication cover more network paths, thereby more comprehensively detecting the status of the entire network. Through random communication between multiple robots, fluctuations and faults in different areas of the network can be discovered in a timely manner, avoiding missing some potential network problems due to focusing on fixed communication paths.

[0095] a matrix construction module, which constructs an interaction failure matrix based on the interaction result set of the semaphore mechanism.

[0096] The method for constructing the interaction failure matrix comprises:

[0097] S100: setting the initial value of cl as 1, the value range of cl being 1 to CL, CL being the number of times of semaphore mechanism interaction; pre-constructing an interaction failure matrix; the interaction failure matrix having N rows and CL columns, the row being the number of the welding robot and the column being the number of times of semaphore mechanism interaction; initializing all values in the interaction failure matrix as 0;

[0098] S101: obtaining the cl-th semaphore mechanism interaction result from the interaction result set of the semaphore mechanism; recording the number of the interaction semaphore set in the cl-th semaphore mechanism interaction result as SL; setting the initial value of sl as 1, the value range of sl being 1 to SL;

[0099] S102: obtaining the sl-th interaction semaphore set from the semaphore mechanism interaction result, extracting the number of the welding robot and the corresponding interaction failure semaphore from the sl-th interaction semaphore set; updating the matrix element in the interaction failure matrix at the position corresponding to the number of the welding robot in the row and cl in the column to the semaphore value of the interaction failure semaphore;

[0100] S103: setting sl = sl + 1, if sl is less than or equal to SL, continuing to execute S102; if sl is greater than SL, setting cl = cl + 1, if cl is less than or equal to CL, continuing S101 to S102, if cl is greater than CL, obtaining the interaction failure matrix, and ending the current process.

[0101] a feature extraction module, which obtains the interaction feature data of the N welding robots based on the interaction failure matrix.

[0102] The method for obtaining the interaction feature data comprises:

[0103] S200: setting the initial value of n as 1, the value range of n being 1 to N;

[0104] S201: obtaining the n-th row of matrix data from the interaction failure matrix, counting the number of the semaphore values of the interaction failure semaphore being 3 in the n-th row of matrix data as the interaction failure times; calculating the interaction failure rate, the interaction failure rate dispersion degree and the time sequence interaction fluctuation index of the n-th welding robot according to the interaction failure times and the number of times of semaphore mechanism interaction;

[0105] The method for calculating the interaction failure rate comprises:

[0106] ;

[0107] wherein, is the interaction failure rate of the nth welding robot, is the number of interaction failures of the nth row of matrix data. The interaction failure rate reflects the overall failure of the robot.

[0108] The method for calculating the dispersion degree of the interaction failure rate comprises:

[0109] ;

[0110] wherein, is the dispersion degree of the interaction failure rate of the nth welding robot, is the semaphore value of the interaction failure semaphore of the clth signal quantity mechanism interaction. The dispersion degree of the interaction failure rate reflects the fluctuation degree of the interaction failure rate. The greater the dispersion degree of the interaction failure rate, the greater the network fluctuation degree.

[0111] The method for obtaining the time sequence interaction fluctuation index comprises:

[0112] ;

[0113] wherein, is the time sequence interaction fluctuation index of the nth welding robot, and e is a constant. The time sequence interaction fluctuation index reflects the change trend of the interaction failure rate of the welding robot in the time dimension, measures the fluctuation amplitude and stability of the interaction failure rate of the welding robot over time, and can reveal the short-time abnormality, long-term trend and sudden fluctuation of the interaction failure rate of the welding robot, thereby assisting in identifying potential network fluctuation problems.

[0114] S202: constructing the interaction feature data of the nth welding robot by using the interaction failure rate, the dispersion degree of the interaction failure rate and the time sequence interaction fluctuation index of the welding robot.

[0115] S203: setting n=n+1, if n is less than or equal to N, then continuing to execute S201 to S202; if n is greater than N, then obtaining the interaction feature data of the N welding robots, and ending the current process.

[0116] The data acquisition module is configured to acquire network running data in the welding robot working environment within a unit time; the network running data comprises bandwidth utilization, network delay, TCP retransmission rate and data transmission rate.

[0117] It should be noted that the network performance trend diagram is as shown in Figure 6 , that is, the change trend of the network response time corresponding to different times, wherein the horizontal axis represents the time interval, and the vertical axis represents the response time, and the unit is millisecond (ms).

[0118] Bandwidth utilization and network latency are link layer characteristic data, TCP retransmission rate is transport layer characteristic data, and data transmission rate is application layer characteristic data. Link layer (bandwidth utilization and network latency): reflects the carrying capacity and transmission efficiency of the physical link, and can be used to identify link congestion, signal attenuation or interference, etc., to ensure the stability of data transmission. Transport layer (TCP retransmission rate): characterizes the end-to-end reliable transmission, which can be used to detect data loss and network jitter, etc., to improve the perception ability of network transmission quality. Application layer (data transmission rate): measures the data interaction performance at the application level, which can be used to evaluate the network adaptability in different business scenarios, and ensure the stable operation of the welding robot under different communication demands.

[0119] Through comprehensive analysis of the characteristic data of the link layer, the transport layer and the application layer, the influencing factors of network fluctuations can be fully described from different network protocol stack levels, the comprehensive evaluation ability of network status can be improved, the identification ability of network fluctuations can be improved, the reliability of abnormal diagnosis can be enhanced, and more targeted decision basis can be provided for network optimization.

[0120] The data processing module is configured to process the network operation data to obtain network characteristic data.

[0121] The method for obtaining the network characteristic data comprises:

[0122] The network interval time is preset, and the unit time is divided into JG network operation data sampling points according to the network interval time;

[0123] For example, in the technical scheme of the present application, under the condition of the factory gigabit Ethernet, the value range of the network interval time is [60ms, 120ms], and the network interval time can be set to 100ms. Under the condition of industrial Wi-Fi, the value range of the network interval time is [80ms, 150ms], and the network interval time can be set to 100ms.

[0124] The bandwidth utilization fluctuation degree is calculated according to the bandwidth utilization of the JG network operation data sampling points; the average network latency and the network latency standard deviation are calculated according to the network latency of the JG network operation data sampling points; and the link stability index is calculated according to the TCP retransmission rate and the data transmission rate of the JG network operation data sampling points.

[0125] The bandwidth utilization fluctuation degree, the average network latency, the network latency standard deviation and the link stability index are constructed into network characteristic data.

[0126] The method for obtaining the bandwidth utilization fluctuation degree comprises:

[0127] ;

[0128] wherein, is a bandwidth utilization fluctuation degree, is a bandwidth utilization of the jgth network operation data sampling point, is a logarithmic function.

[0129] The method for obtaining the average network latency comprises:

[0130] ;

[0131] wherein, is an average network latency, is a network latency of the jgth network operation data sampling point.

[0132] The method for obtaining the network latency standard deviation comprises:

[0133] ;

[0134] wherein, is a network latency standard deviation.

[0135] The method for obtaining the link stability index comprises:

[0136] ;

[0137] wherein, is a link stability index, is a TCP retransmission rate of the jgth network operation data sampling point, is a TCP retransmission rate of the jgth network operation data sampling point, is a TCP retransmission rate of the jgth network operation data sampling point, is a data transmission rate of the jgth network operation data sampling point, and e is a constant.

[0138] It should be noted that, The TCP retransmission rate is exponentially attenuated, which has a nonlinear effect on the link stability index. When the TCP retransmission rate is high, the link stability index decreases, and when the TCP retransmission rate is low, the link stability index is relatively smooth. The data transmission rate represents the data transmission capability. Even if there is a certain retransmission in the network, the link stability index can remain high when the data transmission rate remains stable. As a dynamic adjustment factor, if the TCP retransmission rate suddenly increases in a short time, the decrease of the link stability index increases, so that transient network fluctuations can be sensitively captured. As the average of data transmission rate, it reflects the network stability and helps to identify the link instability. The composite feature formed by the above features can accurately identify short-term sudden abnormalities. It takes into account both long-term trends and short-term fluctuations, and can not only reflect the current network state, but also perceive the changing trend of network performance. Compared with directly using TCP retransmission rate or data transmission rate, the link stability index is more stable and robust, and is suitable for different network environments.

[0139] The fluctuation diagnosis module is configured to input the interaction feature data of the N welding robots and the network feature data into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result. The network fluctuation diagnosis result includes no fluctuation, slight fluctuation, moderate fluctuation, and severe fluctuation.

[0140] It should be noted that the network fluctuation diagnosis result can be numerically represented, for example, 0 represents no fluctuation, 1 represents slight fluctuation, 2 represents moderate fluctuation, and 3 represents severe fluctuation.

[0141] The training method of the fluctuation diagnosis model includes:

[0142] The pre-constructed fluctuation diagnosis data set includes P groups of fluctuation diagnosis data and network fluctuation diagnosis results corresponding to the P groups of fluctuation diagnosis data, P is a positive integer greater than 0, and the fluctuation diagnosis data includes interaction feature data and network feature data. The fluctuation diagnosis data set is divided into a fluctuation diagnosis data training set and a fluctuation diagnosis data validation set, wherein the fluctuation diagnosis data training set is used for parameter learning of the fluctuation diagnosis model, and the fluctuation diagnosis data validation set is used for real-time evaluation of the generalization ability of the fluctuation diagnosis model.

[0143] In the training process of the fluctuation diagnosis model, a deep neural network structure based on a multilayer perceptron is used to convert the fluctuation diagnosis data into a feature vector as input, extract nonlinear features in the data through a hidden layer, and finally generate a probability distribution of the network fluctuation diagnosis result in the output layer using a softmax activation function. The network fluctuation diagnosis result corresponding to the maximum probability is output as the final prediction result. The training process aims to minimize the cross-entropy loss function, and an early stopping strategy is introduced to monitor the performance of the fluctuation diagnosis data validation set. When the prediction accuracy on the fluctuation diagnosis data validation set reaches a preset prediction accuracy threshold, it is considered that the fluctuation diagnosis model has converged, and the training is stopped. For example, in the technical solution of the present application, the prediction accuracy threshold can be set to 95%.

[0144] It should be noted that the interaction feature data mainly reflects the interaction performance of the welding robot end, and can capture abnormal situations caused by network fluctuations. The network feature data reflects the state of the underlying network, and can provide core information such as network congestion, jitter, and packet loss that directly leads to fluctuations. By combining the interaction feature data and the network feature data, the causes of network fluctuations can be more comprehensively described, the modeling accuracy of network state changes can be improved, and thus the accuracy and stability of network fluctuation prediction can be improved.

[0145] The intelligent adjustment module is configured to construct the welding data into a welding data bounding box, obtain the octree recursion number corresponding to the network fluctuation diagnosis result from a pre-constructed network fluctuation diagnosis result-octree recursion number matching table, wherein the network fluctuation diagnosis result-octree recursion number matching table includes network fluctuation diagnosis results and corresponding octree recursion numbers, and perform octree recursive division on the welding data bounding box according to the octree recursion number to obtain the recursively divided welding data. The welding robot performs data transmission according to the recursively divided welding data to adapt to network fluctuations of different degrees.

[0146] As shown in Figure 4 , the method for constructing the welding data into a welding data bounding box includes:

[0147] obtaining three-dimensional coordinate data of the welding data, and obtaining an X-axis minimum coordinate, an X-axis maximum coordinate, a Y-axis minimum coordinate, a Y-axis maximum coordinate, a Z-axis minimum coordinate, and a Z-axis maximum coordinate from the three-dimensional coordinate data;

[0148] obtaining the length of the welding data bounding box by subtracting the X-axis minimum coordinate from the X-axis maximum coordinate, obtaining the width of the welding data bounding box by subtracting the Y-axis minimum coordinate from the Y-axis maximum coordinate, and obtaining the height of the welding data bounding box by subtracting the Z-axis minimum coordinate from the Z-axis maximum coordinate; and constructing the center coordinate of the welding data bounding box from the average of the X-axis maximum coordinate and the X-axis minimum coordinate, the average of the Y-axis maximum coordinate and the Y-axis minimum coordinate, and the average of the Z-axis maximum coordinate and the Z-axis minimum coordinate;

[0149] constructing the welding data bounding box according to the center coordinate, the length, the width, and the height of the welding data bounding box.

[0150] The method for obtaining the center coordinate of the welding data bounding box includes:

[0151] ;

[0152] wherein, the center coordinate is the X-axis maximum coordinate is the X-axis minimum coordinate is the Y-axis maximum coordinate is the Y-axis minimum coordinate is is the maximum coordinate of the Z axis, is the minimum coordinate of the Z axis.

[0153] The network fluctuation diagnosis result-octree recursion number matching table is shown in Table 1:

[0154] Table 1 Network fluctuation diagnosis result-octree recursion number matching table

[0155] Network fluctuation diagnosis result Octree recursion count No fluctuation 6 Mild fluctuation 5 Moderate fluctuation 4 Severe fluctuation 2

[0156] The method for obtaining the welding data after recursive division includes:

[0157] S300: Record the octree recursion number as R, and the initial value of r is 1, and the value range of r is 1 to R;

[0158] S301: If r is equal to 1, create a recursive result set, evenly divide the welding data bounding box into eight welding data sub-bounding boxes, and add the divided welding data sub-bounding boxes to the recursive result set; if r is greater than 1, each welding data sub-bounding box in the recursive result set is evenly octree divided, that is, each welding data sub-bounding box is again evenly divided into eight new welding data sub-bounding boxes, and the divided new welding data sub-bounding boxes are covered to the recursive result set;

[0159] S302: Let r=r+1, if r is less than or equal to R, continue to execute S301; if r is greater than R, the recursive result set is the welding data after recursive division, and the current process is ended.

[0160] Embodiment 2

[0161] Please refer to Figure 3 As shown in the figure, the welding data automatic processing system provided in the embodiment further includes:

[0162] The number setting module is configured to input the network fluctuation diagnosis result, the interaction feature data of the N welding robots, and the network feature data into the number setting model to obtain the octree recursion number.

[0163] The training method of the number setting model includes:

[0164] A pre-constructed device number setting data set is set, the device number setting data set includes Q sets of device number setting data and the octree recursion number corresponding to the Q sets of device number setting data, Q is a positive integer greater than 0, the device number setting data includes network fluctuation diagnosis results, interactive feature data of N welding robots and network feature data; the device number setting data set is divided into a device number setting data training set and a device number setting data verification set, wherein the device number setting data training set is used for parameter learning of the number setting model, and the device number setting data verification set is used for real-time evaluation of the generalization ability of the number setting model;

[0165] In the number setting model training process, a deep neural network structure based on a multilayer perceptron is used, the device number setting data is converted into a feature vector as input, the nonlinear features in the data are extracted through the hidden layer, and finally the probability distribution of the octree recursion number is generated by using a softmax activation function in the output layer, and the octree recursion number corresponding to the maximum probability is output as the final prediction result; the training process aims to minimize the cross-entropy loss function, and an early stopping strategy is introduced to monitor the performance of the device number setting data verification set. When the prediction accuracy on the device number setting data verification set reaches a preset prediction accuracy threshold, it is considered that the number setting model has converged, and the training is stopped. For example, in the technical solution of the present application, the prediction accuracy threshold can be set to 95%.

[0166] It should be noted that by combining the network fluctuation diagnosis results, the interactive feature data of N welding robots and the network feature data with the machine learning model, the dynamic adjustment of the octree recursion number can be realized, which is more close to the real application demand, thereby improving the accuracy of the octree recursion division, and then the unnecessary calculation burden can be reduced while ensuring the data transmission and processing efficiency, and the data processing process is optimized. Ultimately, the welding robot can adaptively adjust the processing accuracy of the welding data under different network conditions, ensure efficient use of network resources, and improve the stability and accuracy of the welding task, thereby meeting the strict requirements of welding quality and efficiency in industrial environments.

[0167] Embodiment 3

[0168] Please refer to Figure 2 The embodiment provides a welding data automatic processing method, which comprises the following steps:

[0169] A semaphore mechanism is set for N welding robots in a unit of time and is interacted to obtain a semaphore mechanism interaction result set;

[0170] An interaction failure matrix is constructed based on the semaphore mechanism interaction result set;

[0171] Obtain interaction feature data of the N welding robots based on the interaction failure matrix;

[0172] Collect network running data in the welding robot working environment in a unit time;

[0173] Process the network running data to obtain network feature data;

[0174] Input the interaction feature data of the N welding robots and the network feature data into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result;

[0175] Intelligently and adaptively process the welding data based on the network fluctuation diagnosis result.

[0176] The above merely describes specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any modification or substitution within the technical range disclosed by the present application can be easily thought of by those skilled in the art, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0177] Finally: The above merely describes preferred embodiments of the present application, and is not intended to limit the present application, and any modification, equivalent substitution, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for automatic processing of welding data, characterized in that The method comprises the following steps: interacting with the semaphore mechanism for N welding robots in a unit time to obtain a semaphore mechanism interaction result set; constructing an interaction failure matrix based on the semaphore mechanism interaction result set; obtaining interaction feature data of the N welding robots based on the interaction failure matrix; collecting network operation data in a welding robot working environment in a unit time; processing the network operation data to obtain network feature data; inputting the interaction feature data of the N welding robots and the network feature data into a fluctuation diagnosis model to obtain a network fluctuation diagnosis result; constructing welding data into a welding data bounding box; and obtaining the octree recursion times corresponding to the network fluctuation diagnosis result from a pre-constructed network fluctuation diagnosis result-octree recursion times matching table. The network fluctuation diagnosis result-octree recursion times matching table comprises network fluctuation diagnosis results and corresponding octree recursion times; and the welding data bounding box is octree recursively divided according to the octree recursion times to obtain recursively divided welding data.

2. The method of claim 1, wherein, The method for constructing the welding data into the welding data bounding box comprises the following steps: obtaining three-dimensional coordinate data of the welding data, and obtaining an X-axis minimum coordinate, an X-axis maximum coordinate, a Y-axis minimum coordinate, a Y-axis maximum coordinate, a Z-axis minimum coordinate and a Z-axis maximum coordinate from the three-dimensional coordinate data; obtaining the length of the welding data bounding box by subtracting the X-axis minimum coordinate from the X-axis maximum coordinate, obtaining the width of the welding data bounding box by subtracting the Y-axis minimum coordinate from the Y-axis maximum coordinate, and obtaining the height of the welding data bounding box by subtracting the Z-axis minimum coordinate from the Z-axis maximum coordinate; and constructing the center coordinate of the welding data bounding box by taking the average of the X-axis maximum coordinate and the X-axis minimum coordinate, the average of the Y-axis maximum coordinate and the Y-axis minimum coordinate, and the average of the Z-axis maximum coordinate and the Z-axis minimum coordinate; constructing the welding data bounding box according to the center coordinate, the length, the width and the height of the welding data bounding box.

3. The method of claim 1, wherein, The method for obtaining the recursively divided welding data comprises the following steps: S300: taking the octree recursion times as R, and presetting the initial value of r as 1, wherein the value range of r is 1 to R; S301: if r is equal to 1, creating a recursive result set, evenly dividing the welding data bounding box into eight welding data sub-bounding boxes, and adding the divided welding data sub-bounding boxes to the recursive result set; if r is greater than 1, evenly octree-dividing each welding data sub-bounding box in the recursive result set, i.e., evenly dividing each welding data sub-bounding box into eight new welding data sub-bounding boxes again, and covering the divided new welding data sub-bounding boxes to the recursive result set; S302: setting r = r + 1, if r is less than or equal to R, continuing to execute S301; if r is greater than R, taking the recursive result set as the recursively divided welding data, and ending the current process.

4. The method of claim 1, wherein, The method for obtaining the semaphore mechanism interaction result set comprises the following steps: Step one: presetting a communication interval time, and interacting with the semaphore mechanism according to the communication interval time in a unit time, wherein the number of times of interaction is recorded as CL; and setting the initial value of cl as 1, wherein the value range of cl is 1 to CL. Step two: for N welding robots, create a corresponding semaphore set for each robot; initialize the semaphore set; the semaphore set contains an interaction request semaphore, an interaction response semaphore, and an interaction failure semaphore; Step three: build N welding robots into a set of robots to be interacted; based on the set of robots to be interacted, the semaphore mechanism interaction is carried out to obtain the semaphore mechanism interaction result of the clth time semaphore mechanism interaction; Step four: let cl=cl+1, if cl is less than or equal to CL, continue to execute step two to step three; if cl is greater than CL, build CL semaphore mechanism interaction results into a semaphore mechanism interaction result set, and end the current process.

5. The method of claim 4, wherein, The method for obtaining the semaphore mechanism interaction result of the clth time semaphore mechanism interaction includes: Let the initial value of i be 1, and the value range of i is 1 to ; represents the floor function. Randomly select a welding robot from the set of robots to be interacted as the ith interaction request robot, and remove the ith interaction request robot from the set of robots to be interacted; randomly select a welding robot from the set of robots to be interacted as the semaphore interaction object of the ith interaction request robot, and remove the semaphore interaction object from the set of robots to be interacted; Build the welding robot number of the ith interaction request robot, the semaphore initiation timestamp, and the randomly generated semaphore verification code into the semaphore communication content; send the semaphore communication content to the semaphore interaction object, and upload the semaphore verification code and the robot number of the semaphore interaction object to the sentinel server; set the semaphore value of the interaction request semaphore of the ith interaction request robot to 1; When the semaphore interaction object of the ith interaction request robot receives the semaphore communication content of the ith interaction request robot, record the interaction completion time, and set the semaphore value of the interaction response semaphore of the semaphore interaction object to 1; Based on the interaction completion time and the semaphore verification code, update the corresponding semaphore set, and build the updated semaphore set of the ith interaction request robot and the semaphore interaction object into an interaction semaphore set; Let i = i + 1, repeat the above process, when i is greater than , construct the set of semaphore interactions into the semaphore interaction result of the clth time semaphore mechanism interaction, end the current process.

6. The method of claim 5, wherein, The method for updating the corresponding semaphore set based on the interaction completion time and the semaphore verification code includes: If the interaction completion time of the ith interaction request robot and the semaphore interaction object is greater than or equal to the preset time interval threshold, it is determined that the semaphore transmission fails, the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 2; If the interaction completion time of the ith interaction request robot and the semaphore interaction object is less than the preset time interval threshold, the semaphore interaction object obtains the corresponding semaphore verification code from the sentinel server, and compares the semaphore verification code with the semaphore verification code in the semaphore communication content; if the comparison result is consistent, it is determined that the semaphore transmission is successful, the semaphore value of the interaction request semaphore of the corresponding semaphore interaction object is set to 2, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 2, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 1; if the comparison result is inconsistent, it is determined that the semaphore transmission fails, the semaphore value of the interaction request semaphore of the ith interaction request robot is set to 3, the semaphore value of the interaction response semaphore of the semaphore interaction object is set to 3, and the semaphore value of the interaction failure semaphore of the ith interaction request robot and the semaphore interaction object is set to 2.

7. The method of claim 1, wherein, The method for constructing the interaction failure matrix comprises: S100: set the initial value of cl as 1, the value range of cl is 1 to CL, CL is the number of semaphore mechanism interactions; pre-construct an interaction failure matrix; the interaction failure matrix has N rows and CL columns, the row is the number of the welding robot, and the column is the number of semaphore mechanism interactions; initialize all values in the interaction failure matrix to 0; S101: obtain the clth semaphore mechanism interaction result from the semaphore mechanism interaction result set; the number of interaction semaphore sets in the clth semaphore mechanism interaction result is denoted as SL; set the initial value of sl as 1, and the value range of sl is 1 to SL; S102: obtain the slth interaction semaphore set from the semaphore mechanism interaction result, extract the number of the welding robot and the corresponding interaction failure semaphore from the slth interaction semaphore set, and update the matrix element in the interaction failure matrix at the position corresponding to the number of the welding robot in the row and cl in the column to the semaphore value of the interaction failure semaphore; S103: set sl = sl + 1, if sl is less than or equal to SL, continue to execute S102; if sl is greater than SL, set cl = cl + 1, if cl is less than or equal to CL, continue S101 to S102, if cl is greater than CL, obtain the interaction failure matrix, and end the current process.

8. The method of claim 1, wherein, The method for obtaining the interaction feature data comprises: S200: set the initial value of n as 1, and the value range of n is 1 to N; S201: obtain the nth row of matrix data from the interaction failure matrix, count the number of interaction failure times of the semaphore value of 3 in the nth row of matrix data, and denote it as the interaction failure times; calculate the interaction failure rate, the interaction failure rate dispersion degree and the time sequence interaction fluctuation index of the nth welding robot according to the interaction failure times and the number of semaphore mechanism interactions; S202: construct the interaction feature data of the nth welding robot by using the interaction failure rate, the interaction failure rate dispersion degree and the time sequence interaction fluctuation index of the nth welding robot. S203: Let n = n + 1. If n is less than or equal to N, continue executing S201 to S202; if n is greater than N, obtain the interaction feature data of N welding robots and end the current process.

9. The method of claim 1, wherein, The method for obtaining the network feature data includes: The network interval time is preset, and the unit time is divided into JG network operation data sampling points according to the network interval time; The bandwidth utilization fluctuation is calculated based on the bandwidth utilization of JG network operation data sampling points; the average network latency and network latency standard deviation are calculated based on the network latency of JG network operation data sampling points; and the link stability index is calculated based on the TCP retransmission rate and data transmission rate of JG network operation data sampling points. Network characteristic data is constructed by measuring bandwidth utilization fluctuations, average network latency, network latency standard deviation, and link stability index.

10. A welding data automatic processing system for implementing the welding data automatic processing method according to any one of claims 1 to 9, characterized by include: The signal interaction module sets up semaphore mechanisms for N welding robots and enables them to interact within a unit of time, obtaining a set of semaphore mechanism interaction results. The matrix construction module constructs an interaction failure matrix based on the set of interaction results from the semaphore mechanism. The feature extraction module obtains the interaction feature data of N welding robots based on the interaction failure matrix; The data acquisition module is used to collect network operation data in the working environment of the welding robot per unit time. The data processing module is used to process network operation data to obtain network characteristic data; The fluctuation diagnosis module is used to input the interaction feature data and network feature data of N welding robots into the fluctuation diagnosis model to obtain the network fluctuation diagnosis results. The intelligent adjustment module is used to construct welding data bounding boxes from welding data; and to obtain the octree recursion count corresponding to the network fluctuation diagnosis result from the pre-built network fluctuation diagnosis result-octree recursion count matching table. The network fluctuation diagnosis result-octree recursion number matching table includes the network fluctuation diagnosis result and the corresponding octree recursion number; according to the octree recursion number, the welding data bounding box is recursively partitioned into octrees to obtain the recursively partitioned welding data; the welding robot transmits data according to the recursively partitioned welding data.

Images