Pipe flange welding robot control method, system, equipment, and storage medium

Abstract

The present disclosure provides a pipe flange welding robot control method, system, device, and storage medium, which belongs to the field of robot control. The method includes: in response to the curvature of the weld satisfying the first curvature condition, using the first control model as the target control model; in response to the curvature of the weld not satisfying the first curvature condition, using the second control model as the target control model; inputting the weld characteristics into the target control model to obtain the target control strategy, and controlling the pipe flange welding robot to perform welding based on the target control strategy; wherein the weld is the gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different. The pipe flange welding robot control method, system, device, and storage medium provided by the present disclosure can improve the accuracy and reliability of the welding robot control.

Description

Technical Field

[0001] The present disclosure belongs to the field of robot control, and more specifically, to a pipe flange welding robot control method and system, equipment, and storage medium. Background Art

[0002] In modern engineering construction, pipe jacking construction technology is widely used in various underground pipeline laying projects. The welding quality of pipe jacking flange and rib plate plays a vital role in the stability and safety of the entire pipe jacking structure.

[0003] The existing pipe flange welding technology usually adopts a fixed welding control method, which cannot adjust the welding control method according to the weld, resulting in unstable welding quality of the welded top pipe and difficulty in ensuring the consistency and stability of the welding quality.

[0004] Therefore, an accurate and reliable pipe flange welding robot control method is urgently needed. Summary of the invention

[0005] The purpose of the present disclosure is to provide a pipe flange welding robot control method and system, equipment, and storage medium to improve the accuracy and reliability of welding robot control.

[0006] A first aspect of an embodiment of the present disclosure provides a pipe flange welding robot control method, comprising:

[0007] In response to the curvature of the weld satisfying a first curvature condition, using the first control model as a target control model;

[0008] In response to the curvature of the weld not satisfying the first curvature condition, using the second control model as the target control model;

[0009] The weld characteristics are input into the target control model to obtain the target control strategy, and the pipe flange welding robot is controlled to perform welding based on the target control strategy;

[0010] The weld is a gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different.

[0011] A second aspect of the embodiments of the present disclosure provides a pipe flange welding robot control system, comprising:

[0012] A first model determination module, configured to use the first control model as a target control model in response to the weld satisfying a first curvature condition;

[0013] a second model determination module, configured to use the second control model as a target control model in response to the weld not satisfying the first curvature condition;

[0014] A control module is used to determine a target control strategy based on a target control model and weld characteristics, and control a pipe flange welding robot to perform welding based on the target control strategy; wherein the weld is a gap between a top pipe flange and a rib plate to be welded, and the control parameters of the first control model and the second control model are different.

[0015] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of the above-mentioned pipe flange welding robot control method are implemented.

[0016] In a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned pipe flange welding robot control method are implemented.

[0017] The pipe flange welding robot control method, system, device, and storage medium provided by the embodiments of the present disclosure have the following beneficial effects:

[0018] The present disclosure selects an applicable control model by judging whether the curvature of the weld meets a specific first curvature condition, so that the welding robot can make adaptive adjustments according to the complexity and shape of different welds, thereby improving the flexibility and adaptability of welding. The present disclosure can ensure that the welding process is more accurate and stable by selecting the control model that best matches the target weld characteristics, thereby improving the welding quality, avoiding the unstable welding quality caused by using a single control model, and improving the stability and reliability of welding control. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A schematic flow chart of a pipe flange welding robot control method provided in one embodiment of the present disclosure;

[0021] Figure 2 A structural block diagram of a pipe flange welding robot control system provided in one embodiment of the present disclosure;

[0022] Figure 3 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present disclosure. However, it should be clear to those skilled in the art that the present disclosure may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obstructing the description of the present disclosure with unnecessary details.

[0024] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0025] Please refer to Figure 1 , Figure 1 A flow chart of a pipe flange welding robot control method provided in an embodiment of the present disclosure, the method comprising:

[0026] S101: In response to the curvature of the weld satisfying a first curvature condition, taking a first control model as a target control model.

[0027] In this embodiment, the weld refers to the gap between the top pipe flange and the rib plate to be welded. The main function of the rib plate is to support the flange and assist in transmitting the top thrust. Welding is a key way to achieve a reliable connection between the rib plate and the flange and ensure effective force transmission. The flange and the rib plate jointly bear and transmit the top thrust.

[0028] The first curvature condition may be that the curvature of the weld is less than the first curvature. The curvature of the weld can be obtained through machine vision and curvature calculation, or through a neural network model trained with a large number of welds and their corresponding curvature data. The first curvature may be determined based on historical experience in welding welds. For example, in historical welding processes, when the curvature is less than a certain threshold, the welding strategy can be changed without changing a relatively perfect welding effect. However, when the curvature is greater than a certain threshold and the welding strategy remains unchanged, insufficient welding strength may occur. Then the secondary threshold may be used as the first curvature.

[0029] The first control model refers to a control model suitable for the weld that satisfies the first curvature condition, that is, when the weld has a relatively small curvature, the first control model can be used. The first control model can be a control model based on the PID algorithm. The weld characteristics can be input into the PID algorithm model, and the control strategy can be output based on the PID algorithm model.

[0030] In this embodiment, the matching degree between the input weld feature and the standard weld feature can be calculated, and the standard control strategy corresponding to the standard weld feature with the highest matching degree can be used as the control strategy. The standard weld feature and the standard control strategy corresponding thereto can be determined based on historical actual welding experience. The standard weld feature is the feature of the sample weld, which is a data set of known values, and the standard control strategy is the standard welding control strategy corresponding to the standard weld feature.

[0031] For example, in the actual welding process, it is found that the relationship between the depth of penetration and the voltage is relatively close, the relationship between the weld width and the current is relatively close, and the welding speed of the weld with a curvature less than the first curvature, the depth of the weld should be consistent with the depth of penetration during welding, the width of the weld should be consistent with the weld width during welding, and for welds with a curvature less than the first curvature, a constant speed welding method can be adopted, and the welding speed can be determined based on actual welding experience.

[0032] In this embodiment, the relationship between the weld depth and the welding voltage and the weld width and the welding current can be established respectively, and the welding control strategy corresponding to the standard weld characteristics can be determined according to the relationship. During the welding process, the weld tracking method can be used to control the welding path. For example, the weld position can be detected in real time by mechanical, photoelectric or laser tracking methods. When the weld position is detected to be offset, the control system will send a signal to automatically adjust the position of the welding gun so that it is always aligned with the center of the weld.

[0033] S102: In response to the curvature of the weld not satisfying the first curvature condition, taking the second control model as the target control model.

[0034] In this embodiment, when the curvature of the weld is greater than or equal to the first curvature, it means that the curvature of the weld is relatively large. The welding path of the weld with a large curvature is relatively complex, and the welding speed needs to be adjusted in real time according to the curvature and radius of the curve. The welding speed should be reduced in the area with a larger curvature to ensure that the penetration depth and width of the weld meet the welding requirements.

[0035] Considering welds with different curvatures, even if the weld width and depth are the same, the corresponding voltage, current and welding time are different. The situation is more complicated, so a more accurate algorithm can be used to control the welding. The second control model can be a particle swarm algorithm, and the weld characteristics can be used as the position of the particles in the particle swarm algorithm. For example, the weld characteristics include: weld depth, weld width and weld length. The position of the particle in the particle swarm algorithm can be defined as [welding current, welding voltage, welding speed], that is, the dimension of the particle is three-dimensional.

[0036] The first control model is simpler than the second control model. Due to the difference in algorithms, the corresponding control parameters are also different.

[0037] S103: inputting the weld characteristics into a target control model to obtain a target control strategy, and controlling the pipe flange welding robot to perform welding based on the target control strategy;

[0038] The weld is a gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different.

[0039] In this embodiment, the weld characteristics refer to some parameters that can reflect the characteristics of the weld between the top pipe and the rib plate. The weld characteristics can be weld width, weld depth and weld length. These characteristics can help determine the control strategy during the welding process.

[0040] The target control model is the model that best suits the curvature of the current weld. The control parameters of the first control model and the second control model are different so as to better fit the welding of the current weld.

[0041] The target control strategy is the most suitable welding control strategy for the current weld, which can include welding voltage, welding current and welding speed. The weld characteristics are input into the determined target control model, and the control strategy is output after calculation by the target control model. The welding is controlled according to the control strategy to achieve the purpose of accurate control.

[0042] It can be concluded from the above that the present disclosure selects an applicable control model by judging whether the curvature of the weld meets a specific first curvature condition, so that the welding robot can make adaptive adjustments according to the complexity and shape of different welds, thereby improving the flexibility and adaptability of welding. The present disclosure can ensure that the welding process is more accurate and stable by selecting the control model that best matches the target weld characteristics, thereby improving the welding quality, avoiding the unstable welding quality caused by using a single control model, and improving the stability and reliability of welding control.

[0043] In one embodiment of the present disclosure, the weld characteristics are input into the target control model to obtain the target control strategy, including:

[0044] Determine the number of particles of the particle swarm algorithm and the learning factor of the particle swarm algorithm based on the weld characteristics;

[0045] Iterate the calculation based on the number of particles, the inertia weight reference value and the learning factor until the number of iterations reaches the target number of iterations or the difference in the fitness function of the consecutive target number is less than the target threshold, and the global optimal position is obtained;

[0046] The control strategy corresponding to the global optimal position is used as the target control strategy corresponding to the weld characteristics.

[0047] In one embodiment of the present disclosure, the weld characteristics include: weld width and weld depth;

[0048] The number of particles in the particle swarm algorithm is determined based on the weld characteristics, including:

[0049] In response to the weld width being greater than the first width and the weld depth being less than or equal to the first depth, increasing a reference value of the particle number based on the first particle number to obtain a particle number of the particle swarm algorithm;

[0050] In response to the weld depth being greater than the first depth and the weld width being less than or equal to the first width, increasing a reference value of the particle number based on the second particle number to obtain a particle number of the particle swarm algorithm;

[0051] In response to the weld width being greater than the first width and the weld depth being greater than the first depth, a reference value of the particle number is increased based on the third particle number to obtain the particle number of the particle swarm algorithm.

[0052] In this embodiment, the purpose is to determine the various parameters of the particle swarm algorithm. First, the number of particles of the particle swarm algorithm can be determined according to the characteristics of the weld. The number of particles in the particle swarm algorithm, that is, the number of particles, will affect the search ability and computational efficiency of the algorithm. Considering that when the width of the weld is larger, that is, greater than the first width, it means that the weld needs more computing resources and more precise control in the width direction. Similarly, when the depth of the weld is larger, that is, greater than the first depth, it means that the weld needs more computing resources in the depth direction. At this time, the number of particles should be appropriately adjusted. The first width and the first depth can be determined according to actual conditions and experience.

[0053] Considering that when the depth and width of the weld are both greater than their corresponding thresholds (i.e., the weld depth is greater than the first depth and the weld width is greater than the first width), the complexity of the weld characteristics at this time increases, and the number of particles should be adjusted again. The above adjustments are all based on the reference value of the number of particles. The reference value of the number of particles can be determined during the experiment. For example, when the weld depth and the weld width are both less than or equal to their corresponding thresholds, the number of particles is 50, and a relatively fast and accurate control strategy can be completed. The reference value of the number of particles can be set to 50.

[0054] The number of first particles, the number of second particles and the number of third particles can all be determined according to actual conditions. The number of first particles and the number of second particles can be equal. Considering that when the weld depth and the weld width exceed their corresponding thresholds at the same time, the weld characteristics are more complex and not simply superimposed, the number of third particles can be greater than the sum of the number of first particles and the number of second particles.

[0055] Secondly, there is the learning factor, which is also called the acceleration constant. It is used to adjust the speed at which particles learn from their own historical optimal position and the group's historical optimal position. It includes group learning factors and individual learning factors, which can guide particles to search for the optimal solution more effectively. The learning factor can be determined by inputting the weld features into the support vector machine model and outputting it from the support vector machine. The support vector machine model is trained with a large number of standard weld features and the corresponding standard learning factors.

[0056] Secondly, the inertia weight reference value can be determined based on experimental debugging, such as through grid search, in a larger range of inertia weight values, and grid search is performed at a certain step size. For example, in the range of 0.1 to 0.9, with a step size of 0.1, 0.1, 0.2, 0.3, etc. are taken as inertia weight reference values, and the particle swarm algorithm is run multiple times for each value, and the performance indicators of the algorithm, such as convergence speed, optimal solution quality, etc., are recorded, and then the inertia weight value that makes the algorithm perform best is selected as the reference value.

[0057] After determining the number of particles, the inertia weight reference value, and the learning factor, iterative calculations can be performed, and other parameters can be set according to the reference parameters of the particle swarm algorithm. For example, the velocity update formula of a particle can be: ,in Represents particles exist The speed at the iteration, represents the inertia weight reference value, represents the learning factor, is a random number between 0 and 1. Represents particles The best historical position, is the historical optimal position of the group, Represents particles In the The position at the iteration.

[0058] The iterative calculation can be iterating to a preset target number of iterations or the difference of the fitness function of the continuous target number is less than the preset target threshold. The target number of iterations can be an iteration based on the comprehensive evaluation of the data in the experimental process. Considering that when the number of particles is small, due to the limited number of particles, the coverage of the search space is relatively small, and more iterations are required to allow particles to explore as many areas of the search space as possible to make up for the incomplete search problem caused by the insufficient number of particles. At the same time, the amount of calculation for each iteration is relatively small, and the computational complexity is low. In this case, the target number of iterations can be appropriately increased to give particles more opportunities to explore in the search space to increase the possibility of finding the global optimal solution. When the number of particles is large, many particles can cover the search space more comprehensively, and each particle can search from a different initial position, making it easier to find the area where the global optimal solution is located. At this time, it does not require too many iterations to allow particles to explore the search space to a greater extent and find a better solution. At this time, the target number of iterations can be appropriately reduced.

[0059] For example, in response to the number of particles being less than the first iteration number, increasing the target number of iterations by the first particle adjustment step size;

[0060] In response to the particle number being greater than the second iteration number, the target iteration number is reduced by a second particle adjustment step size.

[0061] The first iteration number, the second iteration number, the first particle adjustment step size, and the second particle adjustment step size may be determined empirically.

[0062] The fitness function of the continuous target number refers to the fitness function corresponding to multiple consecutive iterations. The target number can be determined according to the actual situation. The expression of the fitness function can be: ,in is the weight coefficient, which can be determined according to user preference. For example, if the weld width is considered to be more important, it can be appropriately increased. , Indicates the relative deviation between the actual melting depth and the target melting depth. Indicates the relative deviation between the actual weld width and the target weld width.

[0063] ,in Indicates the actual penetration depth. Indicates target penetration. ,in Indicates the actual weld width. Indicates the target weld width.

[0064] , ,in represents the empirical coefficient, which can be determined based on historical experiments. represents the heat source efficiency, an inherent property of the welding gun, is the density of welding material, is the welding current, is the welding voltage, represents the specific heat capacity of the welding material, Indicates welding speed. The above parameters are all numerical values, i.e. dimensionless parameters.

[0065] After the iteration is completed, the particle position corresponding to the global optimal position is the optimal position, that is, the position of the particle with the largest fitness function is the optimal position. At this time, the position of the particle is the target control strategy, which can include welding voltage, welding current and welding speed.

[0066] It can be concluded from the above that the present disclosure introduces a particle swarm algorithm to determine the target control strategy, and can perform refined control strategy calculations for the specific characteristics of the weld, thereby improving the accuracy and consistency of welding and ensuring the quality of welding. The present embodiment dynamically adjusts the parameters of the particle swarm algorithm according to the complexity of the weld characteristics, which can more effectively search for the global optimal solution, improve the search efficiency and accuracy of the algorithm, and enhance the stability and reliability of welding control. The present embodiment sets a reasonable number of iterations and a fitness function difference threshold, which can reduce unnecessary iterative calculations while ensuring the comprehensiveness of the search and improve the calculation efficiency.

[0067] In one embodiment of the present disclosure, the pipe flange welding robot control method further includes:

[0068] In response to the number of iterations being less than or equal to the first number, increasing the particle swarm algorithm inertia weight reference value by a first inertia step size;

[0069] In response to the number of iterations being greater than the first number, the inertia weight reference value of the particle swarm algorithm is reduced by a second inertia step size.

[0070] In one embodiment of the present disclosure, the weld characteristics further include: weld length;

[0071] The pipe flange welding robot control method further includes:

[0072] In response to the weld length being greater than or equal to the first length, increasing the first amount by a first compensation amount;

[0073] In response to the weld length being less than the first length, the first amount is reduced by a second compensating amount.

[0074] In this embodiment, considering that in the early stage of the particle swarm algorithm iteration, increasing the inertia weight reference value can make the particles have greater inertia, and can perform a large-scale search in the search space with a larger step size. This helps the particles to explore the entire search space more fully in the initial stage, avoid falling into the local optimal solution too early, and increase the possibility of finding the global optimal solution.

[0075] When the number of iterations exceeds a certain number, the particle swarm algorithm has a certain understanding of the search space. At this time, reducing the inertia weight reference value can make the search step of the particles smaller, and pay more attention to fine search in the current local area. Because as the iteration proceeds, the particles are closer to the optimal solution, and it is necessary to find the optimal solution more accurately in the local range. The smaller inertia weight allows the particles to conduct more detailed exploration near the current optimal position, improving the convergence accuracy of the algorithm.

[0076] Therefore, a threshold can be set. When the number of iterations exceeds the threshold, the inertia weight reference value is appropriately reduced. When the number of iterations is less than or equal to the threshold, the inertia weight reference value is appropriately increased. The threshold is the first number. It should be noted that the first number should be set less than the target number of iterations. The first number can be determined according to a certain ratio of the target number of iterations, for example, it can be one-third to one-half of the target number of iterations.

[0077] The first inertia step length and the second inertia step length are step lengths for adjusting the inertia weight reference value, and can be set according to experimental process data.

[0078] When the weld is long, it means that the welding process is more complicated, and more iterations are needed to fully search for the optimal control strategy. Increasing the first number prolongs the stage of global search by increasing the inertia weight, so that particles have more time and opportunities to find potential optimal solutions in a larger search space. For shorter welds, the welding task is relatively simple, and there is no need for too many iterations to perform global search. Reducing the first number can enter the stage of local fine search by reducing the inertia weight more quickly, accelerate the convergence speed of the algorithm, improve the efficiency of the algorithm, and find the optimal control strategy suitable for short welds more quickly.

[0079] Therefore, another threshold can be set. When the weld length is greater than or equal to the threshold, the first quantity is appropriately increased; when the weld length is less than the threshold, the first quantity is appropriately reduced. The threshold is the first length, and the first compensation quantity and the second compensation quantity are the degree of adjustment of the first quantity. The first length, the first compensation quantity and the second compensation quantity can all be set according to actual conditions.

[0080] It can be concluded from the above that in the early stage of iteration, by increasing the inertia weight reference value, the particles can conduct extensive exploration in the search space with a larger step size, which helps to avoid the algorithm from falling into the local optimal solution too early, thereby increasing the chance of finding the global optimal solution. As the number of iterations increases, reducing the inertia weight reference value can enable the particles to conduct a more detailed search near the current optimal position, thereby improving the convergence accuracy and stability of the algorithm. This embodiment adjusts the number of iterations according to the length of the weld, so that the algorithm can be flexibly adjusted according to the actual conditions of different welds, further improving the adaptability and versatility of the algorithm. By dynamically adjusting the inertia weight reference value and the number of iterations, the algorithm can maintain stable performance under different conditions, thereby improving the stability and reliability of welding control.

[0081] In one embodiment of the present disclosure, a learning factor of a particle swarm algorithm is determined based on a trained support vector machine model, and the learning factor includes: a group learning factor and an individual learning factor;

[0082] The training process of the support vector machine model includes:

[0083] In response to the standard weld length being greater than the second length, the standard weld width being less than or equal to the second width, and the standard weld depth being less than or equal to the second depth, increasing the standard group learning factor by a first group step to obtain a training group learning factor, and decreasing the standard individual learning factor by a first individual step to obtain a training individual learning factor; the standard weld features comprising: a standard weld length, a standard weld width, and a standard weld depth;

[0084] In response to the standard weld width being greater than the second width and / or the standard weld depth being greater than the second depth, reducing the standard group learning factor with a second group step size to obtain a training group learning factor, and reducing the standard individual learning factor with a second individual step size to obtain a training individual learning factor; the standard weld length, the standard weld width and the standard weld depth are obtained based on the sample data;

[0085] The support vector machine is trained based on standard weld features, training group learning factors and training individual learning factors.

[0086] In this embodiment, the purpose is to train the support vector machine model. Before training the support vector machine model, the standard weld characteristics and the corresponding standard learning factors must first be determined. The standard learning factors include training group learning factors and training individual learning factors, that is, sample data, which is one of the data for training the support vector machine model.

[0087] The standard weld length, standard weld depth and standard weld length are all known, namely, sample data, which is one of the data for training the support vector machine model. The standard learning factor refers to a learning factor with a wide range of applicability, which can be set according to the reference parameters of the particle swarm algorithm.

[0088] Characteristics such as the length, width and depth of the weld will affect the complexity of the welding process and the requirements for the control strategy. The group learning factor and individual learning factor in the particle swarm algorithm guide the particles to move to the group optimal position and the individual historical optimal position respectively. By adjusting these two learning factors according to the weld characteristics, the particles can more effectively find the optimal solution in the search space, that is, the best control strategy suitable for specific weld characteristics.

[0089] Considering that the weld length is long (i.e., the standard weld length is greater than the second length), but the width and depth are relatively small (i.e., the standard weld width is less than or equal to the second width, and the standard weld depth is less than or equal to the second depth), it means that the welding process needs to maintain a certain consistency over a longer distance, and it relies more on the experience of the group to find a suitable control strategy. Increasing the group learning factor makes the particles more inclined to move closer to the group's optimal position during the search process, which helps to achieve overall coordination and optimization on longer welds and enhance the algorithm's global search ability to meet the requirements of longer welds. A relatively small individual learning factor can reduce the particle's dependence on its own historical optimal position, avoid excessive attention to local areas, and prevent particles from falling into local optimal solutions during the welding process of long welds, thereby focusing more on obtaining the global optimization direction from the group experience.

[0090] Therefore, when determining the training group learning factor and the training individual learning factor, they can be determined and adjusted according to the standard weld characteristics, and the sample data is the standard weld characteristic data.

[0091] The second length, the second width and the second depth may be determined according to actual conditions during the experiment, and the first group step length and the first individual step length may both be determined based on experience and adjustment methods for other similar problems.

[0092] Considering that the weld width or depth is large (i.e., the standard weld width is less than or equal to the second width, and the standard weld depth is less than or equal to the second depth), it indicates that the complexity of the welding process in the local area increases. Each local area may have its own unique optimal solution. Reducing the group learning factor can reduce the dependence of particles on the group's optimal position and give particles more opportunities to explore local areas autonomously to adapt to complex local welding requirements. Reducing the individual learning factor is to prevent particles from being overly obsessed with their own historical optimal positions, because in such complex local situations, the previous historical optimal positions may no longer be applicable to the current local area. By reducing the individual learning factor, particles can more flexibly adjust the search direction and find more suitable solutions in complex local areas.

[0093] Therefore, when the weld width or weld depth is large (i.e., the standard weld width is greater than the second depth or the standard weld depth is greater than the second depth), the group learning factor and the individual learning factor can be appropriately reduced. The second group step size and the second individual step size can both be determined based on experiments.

[0094] Through the above adjustment of the standard learning factor, a more accurate learning factor can be obtained. Considering that the sample data, that is, the data of standard weld characteristics, cannot cover all weld characteristics, the support vector machine model can be trained to learn the intrinsic relationship between weld characteristics and learning factors, so as to achieve the purpose of accurately outputting the learning factor.

[0095] The support vector machine model can be trained based on standard weld features, training group learning factors and training individual learning factors; two support vector machine models can be trained separately to process group learning factors and individual learning factors respectively, or the training group learning factors and training individual learning factors can be used as a data set to train the support vector machine model, and the output is also a set of group learning factors and individual learning factors, or a multi-output support vector machine model can be used to output group learning factors and individual learning factors.

[0096] It can be concluded from the above that the present disclosure can significantly improve the performance of the algorithm in specific welding tasks by dynamically adjusting the group learning factor and individual learning factor in the particle swarm algorithm according to the weld characteristics, so that the algorithm can more flexibly cope with the complexity of different weld characteristics, thereby more effectively finding the optimal solution in the search space. This embodiment uses the support vector machine model to train the intrinsic relationship between the learning factor and the weld characteristics, so that the algorithm can still make reasonable predictions and adjustments when faced with unseen weld characteristics. This embodiment can more effectively control the welding process by accurately adjusting the learning factor, thereby improving welding quality and efficiency, and enhancing the stability and reliability of welding control.

[0097] In one embodiment of the present disclosure, the training process of the support vector machine model further includes:

[0098] In response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a first distribution condition, reducing a bandwidth parameter reference value of the support vector machine model with a first bandwidth;

[0099] In response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a second distribution condition, a bandwidth parameter reference value of the support vector machine model is increased with a second bandwidth.

[0100] In one embodiment of the present disclosure, the training process of the support vector machine model further includes:

[0101] Based on the clustering algorithm, the standard weld features are clustered to obtain multiple target clusters;

[0102] In response to the standard weld feature belonging to a first target cluster, a penalty parameter of the support vector machine is reduced based on a first penalty step size; the first target cluster is a target cluster containing a number of standard weld features less than a second number.

[0103] In this embodiment, a Gaussian kernel function may be used as the kernel function of the support vector machine model. The formula of the Gaussian kernel function is: ,in Indicates weld feature vectors, Indicates weld feature vectors, is the bandwidth parameter of the Gaussian kernel function.

[0104] The first distribution condition may be that the distribution of the training group learning factor and the training individual learning factor in the data space is concentrated, and the second distribution condition may be that the distribution of the training group learning factor and the training individual learning factor in the data space is dispersed. Whether the distribution of the training group learning factor and the training individual learning factor in the data space is concentrated or dispersed can be determined by calculating the variance of the training group learning factor and the training individual learning factor. In this embodiment, the training group learning factor and the training individual learning factor are collectively referred to as training learning factors. For example, in response to the variance of the training learning factor being less than the first variance, the first distribution condition is satisfied; in response to the variance of the training learning factor being greater than the second variance, the second distribution condition is satisfied. The first variance and the second variance may be set in advance.

[0105] Alternatively, the nearest neighbor distance may be calculated to determine whether the distribution is concentrated or dispersed. For example, the distance from each data point to its nearest neighbor is calculated. If the average nearest neighbor distance is less than the first distance, the first distribution condition is met; if the average nearest neighbor distance is greater than the second distance, the second distribution condition is met. The first distance and the second distance may be set in advance.

[0106] When the distribution of the training group learning factor and the training individual learning factor in the data space meets the first distribution condition, it means that the distribution of the learning factor is relatively concentrated. At this time, lowering the bandwidth parameter reference value of the support vector machine model can allow the model to focus more on these local features, better capture the details of the learning factor distribution, and improve the model's fitting ability for specific data areas, so that the model can be more accurately adjusted and optimized according to the current distribution of learning factors to adapt to possible local laws or characteristics.

[0107] When the distribution of learning factors meets the second distribution condition, it means that the distribution of learning factors is relatively dispersed. Increasing the bandwidth parameter reference value can enable the model to consider the relationship between data points in a larger range, enhance the global fitting ability of the model, avoid the model focusing too much on local fluctuations and ignoring the overall trends and laws, enable the model to better adapt to the diversity and extensiveness of the distribution of learning factors, and improve the generalization ability of the model. The first bandwidth, the second bandwidth and the bandwidth parameter can be determined based on experience.

[0108] In the support vector machine model, the penalty parameter is used to control the model's tolerance for misclassification or fitting errors. A larger penalty parameter will make the model more inclined to reduce classification errors or fitting errors in the training data, which may make the model too complex and prone to overfitting; a smaller penalty parameter allows the model to have a certain degree of error tolerance, making the model simpler and potentially having better generalization ability.

[0109] After clustering the standard weld features based on the clustering algorithm, the first target cluster is a target cluster that contains less standard weld features than the second number, and is a relatively small cluster. The standard weld features in this cluster may have some unique and relatively rare characteristics. For these features, if the penalty parameter is too high, the model may overfit these few samples, resulting in a decrease in the generalization ability for most other data. Therefore, reducing the penalty parameter of the support vector machine based on the first penalty step can relax the fitting requirements for the weld features in this small cluster to a certain extent, so that the model will not pay too much attention to the errors of these few samples, thereby avoiding overfitting, improving the generalization ability of the model for the overall data, and better balancing the performance of the model on different weld feature clusters.

[0110] The clustering algorithm can be a K-Means clustering algorithm or a hierarchical clustering algorithm. When using the K-Means clustering algorithm, the K value, that is, the number of target clusters, can be determined based on the elbow rule. The maximum number of iterations can be determined based on experiments, and the distance measurement method can be Euclidean distance or Manhattan distance.

[0111] The second number may be determined according to the number of standard weld features, for example, a certain proportion of standard weld features, specifically 5% or 8%, etc. The proportion may be determined based on experience.

[0112] It can be concluded from the above that the present disclosure adjusts different bandwidth parameters according to the distribution of learning factors, which helps the model to better adapt to the characteristics of the distribution of learning factors and improve the generalization ability of the support vector machine model. This embodiment clusters the standard weld features through a clustering algorithm, and can identify target clusters containing fewer features. For these unique and rarely occurring features, lowering the penalty parameter can relax the fitting requirements for these features to a certain extent, avoid overfitting the model to these few samples, and help improve the generalization ability of the support vector machine model for the overall data. At the same time, it can also reduce the complexity and overfitting risk of the model due to overfitting a few samples, improve the reliability and accuracy of the output learning factors, and thus improve the stability and reliability of welding control.

[0113] In one embodiment of the present disclosure, the target control strategy includes: welding voltage, welding current and welding speed;

[0114] The pipe flange welding robot control method further includes:

[0115] Determine the heat concentration based on welding voltage, welding current and welding time;

[0116] In response to the heat concentration being less than or equal to the first concentration, the rib plate is not skipped and welding is continued;

[0117] In response to the heat integration level being greater than the first concentration level, skipping a third number of ribs and continuing welding;

[0118] The welding time is the time to weld the gap between the top pipe flange and one rib plate.

[0119] In this embodiment, considering that the welding heat source will continuously input heat to the weldment during the welding process, if the normal welding method is followed, that is, the next rib plate is welded immediately after one rib plate is welded, then during the continuous welding process, heat will continue to accumulate on the pipe wall. Since the heat is highly concentrated in the welding area during welding, and the concrete layer on the inner side of the pipe wall has limited heat dissipation and bearing capacity, when the accumulated heat reaches a certain level, it will cause a large thermal stress inside the concrete layer. When this thermal stress exceeds the bearing limit of the concrete layer, it will cause the concrete layer to burst, thereby affecting the welding quality and the overall performance of the weldment.

[0120] In order to solve this problem, the method of rotating multiple ribs at one time is adopted. In this way, after welding one rib, instead of welding the next adjacent rib immediately, the pipe is rotated so that multiple ribs enter the welding position in sequence before welding. The advantage of this is that during the rotation of the pipe, the welded ribs have a certain cooling time, and the welding heat can be distributed in a larger area, avoiding excessive concentration of heat in a certain local area, thereby effectively reducing the thermal stress on the concrete layer inside the pipe wall, reducing the possibility of the concrete layer bursting due to excessive thermal stress, and ensuring the welding quality and the integrity of the weldment.

[0121] The welding time can be determined according to the time between the start and the end of welding the rib plate, and the heat concentration can be determined according to the first formula, which can be: ,in Indicates welding voltage, Indicates welding current, Indicates welding time. Indicates the thermal diffusion coefficient. Different welding materials have different thermal diffusion coefficients, which reflects the material's ability to conduct heat. It indicates the radius of the jacking pipe and reflects the range of action of the welding heat. The smaller the radius, the more concentrated the heat. It represents the environmental heat dissipation coefficient, which can be set according to the outside temperature and ventilation environment, or determined based on multiple experiments.

[0122] In the denominator Indicates at time The amount of heat that is dissipated from the weld area by thermal diffusion, Indicates at time The amount of heat dissipated from the soldering area by ambient heat dissipation, The larger the value, the higher the heat concentration.

[0123] When the heat concentration is less than or equal to the first concentration, it is considered that the heat concentration is within an acceptable range, and there is no need to skip the rib plate, and each rib plate is welded normally in sequence. When the heat is greater than the first concentration, it is considered that the rib plate needs to be skipped. The first concentration can be determined by the welding quality and effect during the actual welding operation.

[0124] In this embodiment, in response to the heat concentration level being greater than the first concentration level and less than or equal to the second concentration level, the third amount is determined based on the second formula;

[0125] In response to the heat concentration level being greater than the second concentration level, the third amount is set to a fixed value.

[0126] The second formula can be , .in Indicates rounding up. Indicates the number of skips, Indicates the first concentration level, Indicates the second concentration level, Indicates a fixed value, that is, the maximum number of ribs that can be skipped. The fixed value can be determined based on experience. For example, in general, skipping 3 ribs can effectively reduce the thermal stress on the concrete layer inside the pipe wall, so the fixed value can be determined as 3.

[0127] It should be noted that when the cycle continues like this and no weld is encountered, that is, the rib plate has been welded, the process will automatically be postponed until the fourth number of rib plates have no welds, and the fourth number is equal to the number of rib plates that need to be welded.

[0128] It can be concluded from the above that by introducing the calculation of heat concentration, the control method can accurately evaluate the accumulation of heat during welding, thereby effectively avoiding the problem of concrete layer cracking caused by excessive heat concentration. This significantly improves the welding quality and the overall performance of the weldment. This embodiment calculates the heat concentration based on parameters such as welding voltage, welding current, and welding time, and adjusts the number of ribs skipped according to actual conditions, so that the welding control method can adapt to different welding conditions and requirements, and improves the stability and reliability of welding control.

[0129] Corresponding to the pipe flange welding robot control method of the above embodiment, Figure 2 This is a structural block diagram of a pipe flange welding robot control system provided by an embodiment of the present disclosure. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 2 The pipe flange welding robot control system 20 includes: a first model determination module 21, a second model determination module 22 and a control module 23.

[0130] The first model determination module 21 is used to use the first control model as the target control model in response to the weld satisfying the first curvature condition;

[0131] A second model determination module 22, for using the second control model as a target control model in response to the weld not satisfying the first curvature condition;

[0132] The control module 23 is used to determine the target control strategy based on the target control model and the weld characteristics, and control the pipe flange welding robot to perform welding based on the target control strategy; wherein the weld is the gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different.

[0133] In one embodiment of the present disclosure, the control module 23 is specifically used to determine the number of particles of the particle swarm algorithm and the learning factor of the particle swarm algorithm based on the weld characteristics;

[0134] Iterate the calculation based on the number of particles, the inertia weight reference value and the learning factor until the number of iterations reaches the target number of iterations or the difference in the fitness function of the consecutive target number is less than the target threshold, and the global optimal position is obtained;

[0135] The control strategy corresponding to the global optimal position is used as the target control strategy corresponding to the weld characteristics.

[0136] In one embodiment of the present disclosure, the weld characteristics include: weld width and weld depth;

[0137] The control module 23 is further configured to increase the reference value of the number of particles based on the first number of particles to obtain the number of particles of the particle swarm algorithm in response to the weld width being greater than the first width and the weld depth being less than or equal to the first depth;

[0138] In response to the weld depth being greater than the first depth and the weld width being less than or equal to the first width, increasing a reference value of the particle number based on the second particle number to obtain a particle number of the particle swarm algorithm;

[0139] In response to the weld width being greater than the first width and the weld depth being greater than the first depth, a reference value of the particle number is increased based on the third particle number to obtain the particle number of the particle swarm algorithm.

[0140] In one embodiment of the present disclosure, the pipe flange welding robot control system 20 further includes: an inertia weight adjustment module;

[0141] an inertia weight adjustment module, for increasing the inertia weight reference value of the particle swarm algorithm by a first inertia step length in response to the number of iterations being less than or equal to a first number;

[0142] In response to the number of iterations being greater than the first number, the inertia weight reference value of the particle swarm algorithm is reduced by a second inertia step size.

[0143] In one embodiment of the present disclosure, a learning factor of a particle swarm algorithm is determined based on a trained support vector machine model, and the learning factor includes: a group learning factor and an individual learning factor; standard weld features include: standard weld length, standard weld width, and standard weld depth;

[0144] The pipe flange welding robot control system 20 also includes: a model training module;

[0145] a model training module, for, in response to the standard weld length being greater than the second length, the standard weld width being less than or equal to the second width, and the standard weld depth being less than or equal to the second depth, increasing the standard group learning factor by a first group step length to obtain a training group learning factor, and decreasing the standard individual learning factor by a first individual step length to obtain a training individual learning factor;

[0146] In response to the standard weld width being greater than the second width and / or the standard weld depth being greater than the second depth, reducing the standard group learning factor with a second group step size to obtain a training group learning factor, and reducing the standard individual learning factor with a second individual step size to obtain a training individual learning factor; the standard weld length, the standard weld width and the standard weld depth are obtained based on the sample data;

[0147] The support vector machine is trained based on standard weld features, training group learning factors and training individual learning factors.

[0148] In one embodiment of the present disclosure, the model training module is specifically used to reduce the bandwidth parameter reference value of the support vector machine model with a first bandwidth in response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a first distribution condition;

[0149] In response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a second distribution condition, a bandwidth parameter reference value of the support vector machine model is increased with a second bandwidth.

[0150] In one embodiment of the present disclosure, the model training module is further used to perform clustering processing on standard weld features based on a clustering algorithm to obtain multiple target clusters;

[0151] In response to the standard weld feature belonging to a first target cluster, a penalty parameter of the support vector machine is reduced based on a first penalty step size; the first target cluster is a target cluster containing a number of standard weld features less than a second number.

[0152] See also Figure 3 , Figure 3 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. Figure 3The electronic device 300 in the embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303 and one or more memories 304. The processors 301, input devices 302, output devices 303 and memories 304 communicate with each other via a communication bus 305. The memory 304 is used to store computer programs, which include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. The processor 301 is configured to call the program instructions to execute the functions of each module / unit in the above-mentioned device embodiments, such as Figure 2 The functions of modules 21 to 23 are shown.

[0153] It should be understood that in the embodiment of the present disclosure, the processor 301 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0154] The input device 302 may include a touch panel, a fingerprint collection sensor (for collecting the user's fingerprint information and fingerprint direction information), a microphone, etc., and the output device 303 may include a display (LCD, etc.), a speaker, etc.

[0155] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0156] In a specific implementation, the processor 301, input device 302, and output device 303 described in the embodiments of the present disclosure can execute the implementation methods described in the first and second embodiments of the pipe flange welding robot control method provided in the embodiments of the present disclosure, and can also execute the implementation methods of the electronic device described in the embodiments of the present disclosure, which will not be repeated here.

[0157] In another embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by the processor, all or part of the processes in the above-mentioned embodiment method are implemented, and the computer program can also be completed by instructing the relevant hardware through the computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0158] The computer-readable storage medium may be an internal storage unit of the electronic device of any of the aforementioned embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0159] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0160] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0161] In the several embodiments provided in the present application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces or units, or it can be an electrical, mechanical or other form of connection.

[0162] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present disclosure.

[0163] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0164] The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A pipe flange welding robot control method, characterized in that: include: In response to the curvature of the weld satisfying a first curvature condition, using the first control model as a target control model; In response to the curvature of the weld not satisfying the first curvature condition, using the second control model as the target control model; The first curvature condition is that the curvature of the weld is less than the first curvature, the first control model is a PID algorithm control model; the second control model is a particle swarm algorithm model; Inputting the weld characteristics into the target control model to obtain a target control strategy, and controlling the pipe flange welding robot to perform welding based on the target control strategy; Wherein, the weld is a gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different; Inputting the weld characteristics into the second control model to obtain a target control strategy includes: Determine the number of particles of the particle swarm algorithm and the learning factor of the particle swarm algorithm based on the weld characteristics; Iterative calculation is performed based on the number of particles, the inertia weight reference value and the learning factor until the number of iterations reaches the target number of iterations or the difference of the fitness function of the consecutive target number is less than the target threshold, thereby obtaining a global optimal position; The control strategy corresponding to the global optimal position is used as the target control strategy corresponding to the weld feature.

2. The pipe flange welding robot control method according to claim 1, characterized in that: The weld characteristics include: weld width and weld depth; The method of determining the number of particles of the particle swarm algorithm based on the weld characteristics comprises: In response to the weld width being greater than a first width and the weld depth being less than or equal to a first depth, increasing a reference value of a particle number based on a first particle number to obtain a particle number of the particle swarm algorithm; In response to the weld depth being greater than the first depth and the weld width being less than or equal to the first width, increasing a reference value of the particle number based on the second particle number to obtain the particle number of the particle swarm algorithm; In response to the weld width being greater than the first width and the weld depth being greater than the first depth, a reference value of the particle number is increased based on the third particle number to obtain the particle number of the particle swarm algorithm.

3. The pipe flange welding robot control method according to claim 2, characterized in that: Also includes: In response to the number of iterations being less than or equal to a first number, increasing the inertia weight reference value of the particle swarm algorithm by a first inertia step size; In response to the number of iterations being greater than the first number, the inertia weight reference value of the particle swarm algorithm is reduced by a second inertia step size.

4. The pipe flange welding robot control method according to claim 1, characterized in that: Determine the learning factor of the particle swarm algorithm based on the trained support vector machine model, wherein the learning factor includes: a group learning factor and an individual learning factor; the standard weld features include: a standard weld length, a standard weld width and a standard weld depth; The training process of the support vector machine model includes: In response to the standard weld length being greater than the second length, the standard weld width being less than or equal to the second width, and the standard weld depth being less than or equal to the second depth, increasing the standard group learning factor by a first group step to obtain a training group learning factor, and decreasing the standard individual learning factor by a first individual step to obtain a training individual learning factor; In response to the standard weld width being greater than the second width and / or the standard weld depth being greater than the second depth, reducing the standard group learning factor with a second group step size to obtain a training group learning factor, and reducing the standard individual learning factor with a second individual step size to obtain a training individual learning factor; the standard weld length, standard weld width and standard weld depth are obtained based on sample data; The support vector machine is trained based on the standard weld features, the training group learning factor and the training individual learning factor.

5. The pipe flange welding robot control method according to claim 4, characterized in that: The training process of the support vector machine model also includes: In response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a first distribution condition, reducing a bandwidth parameter reference value of a support vector machine model with a first bandwidth; In response to the distribution of the training group learning factor and the training individual learning factor in the data space satisfying a second distribution condition, a bandwidth parameter reference value of the support vector machine model is increased with a second bandwidth.

6. The pipe flange welding robot control method according to claim 4, characterized in that: The training process of the support vector machine model also includes: Clustering the standard weld features based on a clustering algorithm to obtain multiple target clusters; In response to the standard weld feature belonging to a first target cluster, a penalty parameter of the support vector machine is reduced based on a first penalty step size; the first target cluster is a target cluster containing a number of standard weld features less than a second number.

7. A pipe flange welding robot control system, characterized in that: include: A first model determination module, configured to use the first control model as a target control model in response to the weld satisfying a first curvature condition; a second model determination module, configured to use the second control model as a target control model in response to the weld not satisfying the first curvature condition; The first curvature condition is that the curvature of the weld is less than the first curvature, the first control model is a PID algorithm control model; the second control model is a particle swarm algorithm model; A control module, used for determining a target control strategy based on the target control model and weld characteristics, and controlling the pipe flange welding robot to perform welding based on the target control strategy; wherein the weld is a gap between the top pipe flange and the rib plate to be welded, and the control parameters of the first control model and the second control model are different; The control module is specifically used to determine the number of particles of the particle swarm algorithm and the learning factor of the particle swarm algorithm based on the weld characteristics; Iterative calculation is performed based on the number of particles, the inertia weight reference value and the learning factor until the number of iterations reaches the target number of iterations or the difference of the fitness function of the consecutive target number is less than the target threshold, thereby obtaining a global optimal position; The control strategy corresponding to the global optimal position is used as the target control strategy corresponding to the weld feature.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images