Weldability evaluation method, device and equipment based on welding sequence task

Abstract

The invention relates to the technical field of welding path optimization, and discloses a weldability evaluation method, device and equipment based on a welding sequence task, and the method comprises the following steps: obtaining a candidate welding sequence of a corner fitting box to be welded, and determining an attitude score of a target welding task by combining a welding gun feasible attitude number of the target welding task, determining the motion score of the target welding task in combination with the motion posture of the positioner; and on the basis of the posture score and the motion score of each welding task in each candidate welding sequence, searching an optimal welding sequence in the candidate welding sequences by utilizing a Monte Carlo tree search algorithm. According to the method, the candidate welding sequences are obtained, the posture score is determined by combining the feasible posture number of the welding gun, the motion score is determined by combining the moving posture of the positioner, the weldability of the welding task is accurately evaluated, the welding defect risk is reduced, the optimal sequence is screened through the Monte Carlo tree search algorithm, the welding process is optimized, equipment idling and adjustment are reduced, and the welding efficiency is improved. Operation efficiency is improved, cost is reduced, and a new technical path is provided for welding sequence planning.

Description

Technical Field

[0001] This invention relates to the field of welding path optimization technology, and specifically to a weldability assessment method, apparatus, and equipment based on welding sequence tasks. Background Technology

[0002] The welding quality of corner boxes depends on the coordination between the welding torch posture and the positioner movement. In traditional welding, the adjustment of these two aspects relies on the welder's experience and lacks quantitative evaluation standards, leading to the following problems: First, the welding torch is prone to collision with components in narrow spaces (such as the gap between the side plate and the top plate), resulting in welding defects; second, the positioner (A-axis, B-axis) movement angles are unreasonable, resulting in redundant welding paths and low efficiency; third, the feasibility of the welding sequence cannot be assessed in advance, often resulting in the situation that "the planned sequence cannot be actually executed", increasing rework costs.

[0003] In existing technologies, weldability assessment often relies on adjusting the welding torch posture and locator movement using a single index, such as whether the welding torch collides with the corner box. This does not comprehensively consider the feasibility of the posture and the movement cost of the locator, resulting in problems such as a lack of quantitative standards, unstable welding quality, easy collision of the welding torch in confined spaces, and inability to efficiently identify feasible postures. Furthermore, a standardized scoring system has not been formed, making it difficult to support path optimization for robotic automated welding. Summary of the Invention

[0004] In view of this, the present invention provides a weldability assessment method, apparatus and equipment based on welding sequence tasks to solve the problem that automated welding paths cannot be quantitatively assessed.

[0005] In a first aspect, the present invention provides a weldability assessment method based on a welding sequence task, which utilizes a welding torch and a positioner to complete the welding task of a corner piece box to be welded, the method comprising:

[0006] Obtain the candidate welding sequence of the corner box to be welded, and determine the attitude score of the target welding task by combining the number of feasible welding gun postures of the target welding task in each candidate welding sequence, and determine the motion score of the target welding task by combining the positioner motion posture of the target welding task.

[0007] Based on the attitude and motion scores of each welding task in each candidate welding sequence, the Monte Carlo tree search algorithm is used to search for the optimal welding sequence among the candidate welding sequences.

[0008] The weldability assessment method based on welding sequence tasks provided by this invention obtains candidate welding sequences, determines the attitude score by combining the number of feasible welding torch postures and the motion score by the positioner motion posture, accurately assesses the weldability of welding tasks, reduces the risk of welding defects, uses the Monte Carlo tree search algorithm to screen the optimal sequence, optimizes the welding process, reduces equipment idling and adjustment, improves work efficiency and reduces costs, and provides a new technical path for welding sequence planning.

[0009] In one optional implementation, the candidate welding sequence includes multiple welding tasks. The attitude score of the target welding task is determined by combining the number of feasible welding torch attitudes for the target welding task in each candidate welding sequence, including:

[0010] The number of feasible welding torch postures for the target welding task is determined by combining collision analysis, and the number of feasible welding torch postures is no more than 3.

[0011] The number of feasible welding torch postures is used as the posture score of the target welding task.

[0012] In one optional implementation, the positioner includes at least one rotating axis, and the motion component of the target welding task is determined in conjunction with the positioner's motion posture for the target welding task, including:

[0013] When performing the target welding task, obtain the rotation angle of each rotating axis in the positioner;

[0014] Based on the preset mapping relationship between the rotation axis angle and the score, the rotation score of each rotation axis is determined;

[0015] The sum of the rotation scores of each rotating axis is taken as the motion score of the target welding task.

[0016] The weldability assessment method based on welding sequence tasks provided by this invention limits the number of feasible welding torch postures through collision analysis and uses them as posture scores to accurately screen suitable postures, avoid welding interference, and ensure welding quality. It obtains the rotation angle based on at least one rotation axis, calculates the score according to a preset mapping relationship, and sums them as motion scores to refine the positioner motion assessment, accurately reflect motion costs, help optimize equipment scheduling, and improve the collaborative efficiency of positioner and welding torch in the welding process.

[0017] In one alternative implementation, the method further includes:

[0018] Obtain the attitude score and number of welding tasks for each welding task in the candidate welding sequence, and calculate the average attitude score of each welding task as the average attitude score of the candidate welding sequence.

[0019] Obtain the motion score of each welding task in the candidate welding sequence, and calculate the sum of the motion scores of each welding task as the motion score of the candidate welding sequence;

[0020] The optimal welding sequence is selected from the candidate sequences based on the average attitude score and motion score of each candidate welding sequence.

[0021] The weldability assessment method based on welding sequence tasks provided by this invention comprehensively considers the overall adaptability of welding torch postures in candidate welding sequences by calculating the average posture score, avoiding interference from single-task posture issues with global judgment; and statistically quantifies the positioner movement cost by calculating the motion score, clearly presenting the differences in equipment scheduling efficiency. Based on these two key indicators, the optimal sequence is selected, eliminating reliance on experience and establishing an objective and quantifiable evaluation standard for welding process planning. It can accurately select the solution with good welding torch posture coordination and efficient positioner movement from the candidate sequences, thereby improving the quality and efficiency of welding operations.

[0022] In one optional implementation, based on the attitude and motion scores of each welding task in each candidate welding sequence, a Monte Carlo tree search algorithm is used to search for the optimal welding sequence among the candidate welding sequences, including:

[0023] Based on each candidate welding sequence, a state-action-transition decision model for the search tree is constructed. The state is used to represent the prefix sequence of the completed welding task and is associated with the cumulative path score of the current welding task and the locator's posture in the previous welding task. The action is used to represent the feasible welding torch posture of the current welding task, and the posture score of the current welding task is used to represent the feasible welding torch posture.

[0024] Taking the welding task with an empty prefix sequence as the root node, the sequence search process is executed cyclically starting from the root node. The sequence search process includes: selecting child nodes downwards according to the upper confidence boundary strategy until a node that has not been fully expanded is reached; adding new actions to the node that has not been fully expanded to form a new node; simulating the welding points and feasible postures of the welding gun for the remaining path starting from the new node; determining the score of the simulated path and feeding it back to the previous node layer by layer to obtain the cumulative score of each node; calculating the score of each path based on the cumulative score of each node until the path with the highest cumulative score is determined as the optimal welding sequence.

[0025] In one optional implementation, the formula for calculating the cumulative path score of the current welding task in the state-action-transition decision model is:

[0026] G(s')=G(s)+pose_score(a)+motion_score(prev_pose→a)

[0027] Where G(s') represents the cumulative path score of the current welding task, G(s) represents the cumulative path score of the previous welding task, pose_score(a) represents the pose score of pose a, and motion_score(prev_pose→a) represents the motion score from the previous welding task to the current welding task.

[0028] The weldability assessment method based on welding sequence tasks provided by this invention constructs a state-action-transition decision model. Using the prefix sequence as the root node, it iteratively searches using an upper confidence boundary strategy. By simulating the remaining path and receiving feedback scores, it accurately quantifies the impact of each task's attitude and motion scores on the path. It uses a clear formula to calculate the cumulative score, which not only systematically sorts out the association of welding tasks (completed tasks, locator attitude, etc.), but also efficiently traverses candidates through intelligent search, scientifically finding the optimal sequence with the highest cumulative score. This improves the systematicness, accuracy, and efficiency of welding planning and helps to optimize the intelligent welding process.

[0029] In an alternative implementation, before cyclically performing the sequence search process, the method further includes:

[0030] Calculate the number of feasible welding gun postures for each welding task in each welding sequence. If there is a welding task with 0 feasible welding gun postures, mark the welding sequence as an infeasible sequence.

[0031] Before performing the sequence search process in a loop, infeasible sequences in the original search tree are filtered out, and the remaining sequences are used as the search tree.

[0032] The weldability assessment method based on welding sequence tasks provided by this invention calculates the number of feasible welding gun postures for a welding task, marks and filters out infeasible sequences containing welding tasks with 0 feasible postures, forms a search tree, and preemptively filters out sequences that cannot be welded at all, avoiding wasted effort in subsequent searches, significantly reducing the size of the search tree, reducing computational load, and improving sequence search efficiency. At the same time, it avoids invalid planning caused by infeasible sequences from the source, ensuring that subsequent searches focus on actually executable candidate sequences.

[0033] Secondly, the present invention provides a weldability assessment device based on a welding sequence task, which utilizes a welding torch and a positioner to complete the welding task of a corner piece box to be welded. The device includes:

[0034] The welding task score determination module is used to obtain candidate welding sequences of the corner box to be welded, and determine the attitude score of the target welding task by combining the number of feasible welding gun postures of the target welding task in each candidate welding sequence, and determine the motion score of the target welding task by combining the positioner motion posture of the target welding task.

[0035] The optimal sequence search module is used to search for the optimal welding sequence among the candidate welding sequences based on the attitude and motion scores of each welding task in each candidate welding sequence using the Monte Carlo tree search algorithm.

[0036] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the method described in the first aspect or any corresponding embodiment thereof.

[0037] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0038] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating the weldability assessment method based on welding sequence tasks according to an embodiment of the present invention.

[0040] Figure 2 This is a schematic diagram of four welding torch postures in the weldability assessment method based on welding sequence tasks according to an embodiment of the present invention.

[0041] Figure 3 This is a schematic diagram of a non-standard posture determined by collision avoidance in a highly confined space in the weldability evaluation method based on welding sequence tasks according to an embodiment of the present invention.

[0042] Figure 4 This is a flowchart illustrating another weldability assessment method based on a welding sequence task according to an embodiment of the present invention;

[0043] Figure 5 This is a flowchart illustrating another weldability assessment method based on welding sequence tasks according to an embodiment of the present invention.

[0044] Figure 6 This is a schematic diagram of attitude scoring in the weldability assessment method based on welding sequence tasks according to an embodiment of the present invention;

[0045] Figure 7 This is a schematic diagram of the rotating shaft of a dual-axis positioner in a weldability assessment method based on a welding sequence task according to an embodiment of the present invention.

[0046] Figure 8 This is a schematic diagram of eight discrete positioner postures of a dual-axis positioner in the weldability evaluation method based on welding sequence tasks according to an embodiment of the present invention.

[0047] Figure 9 This is a schematic diagram illustrating attitude scoring and motion scoring in the weldability assessment method based on welding sequence tasks according to an embodiment of the present invention.

[0048] Figure 10 This is a structural block diagram of a weldability evaluation device based on a welding sequence task according to an embodiment of the present invention;

[0049] Figure 11 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] This invention provides a weldability assessment method based on welding sequence tasks. By calculating the attitude and motion scores of each welding task, the weldability of the welding task is accurately assessed, thereby achieving the effect of quantitatively assessing automated welding paths.

[0052] According to an embodiment of the present invention, a solderability assessment method based on a welding sequence task is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0053] This embodiment provides a weldability evaluation method based on welding sequence tasks, which can be used in the aforementioned computer system. Figure 1 This is a flowchart of a weldability assessment method based on a welding sequence task according to an embodiment of the present invention. The welding task of the corner component box to be welded is completed using a welding torch and a positioner. Figure 1 As shown, the process includes the following steps:

[0054] Step S101: Obtain the candidate welding sequence of the corner box to be welded, and determine the attitude score of the target welding task by combining the number of feasible welding gun postures of the target welding task in each candidate welding sequence, and determine the motion score of the target welding task by combining the positioner motion posture of the target welding task.

[0055] Specifically, the basic information and local path topology of the weld in the corner box to be welded are obtained, and the local path topology of the welding operation is defined by the weld points forming lines or discretizing them into curve shapes. In the CAD model of the corner box to be welded, each weld point carries the direction information of the nearest wall (the local wall at the weld). The direction information can be used to determine the initial approach vector of the welding gun, so that the sampling-based method can identify feasible robotic welding operations. If the weld surface is not manually defined in the CAD model, the local wall direction information of the weld must be extracted to determine the actual weld surface. The extraction process of the local path topology includes: (1) collecting the intersection lines between the solid elements in the CAD model, identifying the welds between the corner box components to be welded, and for each plate, at least two different edge weld points are needed to fix the plate; (2) for long welds, select two points that trisect the weld as weld points, and for short welds, use the midpoint as the weld point; (3) calculate the local wall vector, bisect the dihedral angle between the two local surfaces to obtain the angle bisector vector, which defines the welding direction.

[0056] The candidate welding sequence of the corner box to be welded is the welding order of multiple welding tasks selected under the premise of the optimal assembly sequence of the corner box to be welded. The welding tasks include weld points and weld seams. Weld seams are equivalent to multiple consecutive weld points, so weld points can be used as welding tasks.

[0057] For the same welding task, the welding torch posture includes, for example: Figure 2 The four scenarios shown illustrate that, in actual welding processes, to avoid collisions between the welding torch and the corner piece box being welded, the torch posture needs to be adjusted. This is especially true in confined spaces where the likelihood of collisions is higher, making torch posture adjustment more difficult. To simplify calculations, this embodiment simplifies the welding torch into a "three-point, two-line" model. In such cases... Figure 2 Based on the four feasible welding torch postures shown, by converting the four fixed angle values ​​of the welding torch into an "adjustable angle range," the limitation of a single angle is broken, allowing the welding torch posture to flexibly adapt to the size requirements of the confined space within a reasonable range. Secondly, by combining collision detection methods, collision simulation analysis is performed on all potential welding postures within the adjustable angle range one by one, screening out postures that will not interfere with the surrounding structures such as the corner fitting box side plate and top plate, i.e., welding postures that satisfy tool collision constraints. For example... Figure 3 The diagram shows a non-standard posture determined by collision avoidance in a highly confined space. These special feasible postures are marked and recorded so that they can be directly called upon in subsequent welding tasks in similar confined spaces, reducing redundant calculations.

[0058] During the welding process, the welding torch posture and the positioner's motion posture directly affect the welding quality. In order to evaluate the feasibility of each welding task and optimize the welding path, the feasibility of the welding torch posture and the suitability of the positioner's motion are quantified.

[0059] Step S102: Based on the attitude and motion scores of each welding task in each candidate welding sequence, the Monte Carlo tree search algorithm is used to search for the optimal welding sequence among the candidate welding sequences.

[0060] Specifically, a search tree is constructed based on each candidate welding sequence. This search tree uses a Monte Carlo Tree Search (MCTS) framework. Given the existing assembly sequence and intermediate states, the optimal welding posture is selected for each welding task to maximize the total score of the entire sequence. A "state-action-transition" decision model is established based on the search tree. "State" represents the progress record of completed welding tasks, associated with the currently accumulated score and the positioner's posture in the previous welding round. "Action" corresponds to the feasible postures the welding torch can adopt when performing the current welding task, and the posture score reflects the value of that posture.

[0061] Starting with the state before welding begins, the search originates from the assembly sequence and candidate welding sequences, forming the root node. The search continues iteratively, selecting child nodes using an "upper confidence boundary strategy" until all unexplored nodes remain. For these nodes, new feasible welding torch postures are added to form new nodes. Starting from these new nodes, the welding points and torch postures for the remaining welding tasks are simulated, and the score for this simulated path is calculated. This score is then fed back to the previous nodes to update the cumulative score. This process is repeated until the path with the highest cumulative score is identified as the optimal welding sequence. Welding based on this optimal sequence is both efficient and reduces the probability of problems occurring.

[0062] The weldability assessment method based on welding sequence tasks provided in this embodiment obtains candidate welding sequences, determines the attitude score by combining the number of feasible welding torch postures and the motion score by determining the positioner's motion posture, accurately assesses the weldability of welding tasks, reduces the risk of welding defects, uses the Monte Carlo tree search algorithm to screen the optimal sequence, optimizes the welding process, reduces equipment idling and adjustments, improves operational efficiency, reduces costs, and provides a new technological path for welding sequence planning.

[0063] In some alternative implementations, such as Figure 4 As shown, the method also includes:

[0064] Step S103: Obtain the attitude score and number of welding tasks for each welding task in the candidate welding sequence, and calculate the average attitude score of each welding task as the average attitude score of the candidate welding sequence.

[0065] Specifically, when evaluating the entire welding task sequence, two key metrics are considered: average attitude score and motion score. The average number of feasible attitudes for each welding sequence is calculated by counting the average number of feasible welding torch attitudes across all welding tasks. The higher the average value, the greater the flexibility and adaptability of the welding path.

[0066] Step S104: Obtain the motion score of each welding task in the candidate welding sequence, and calculate the sum of the motion scores of each welding task as the motion score of the candidate welding sequence.

[0067] Specifically, the motion score of the locator in the welding sequence directly affects the welding efficiency and robot motion complexity. By assigning separate scores to the angle changes of each rotation axis and adding the scores of each rotation axis together, the total score is obtained as the motion score of the candidate welding sequence.

[0068] Step S105: Select the optimal welding sequence from the candidate sequences based on the average attitude score and motion score of each candidate welding sequence.

[0069] Specifically, the average attitude score reflects the flexibility of each candidate welding sequence. The higher the average attitude score, the higher the flexibility, and the easier it is to weld the corresponding candidate welding sequence. The motion score reflects the efficiency of each candidate welding sequence. The more the positioner rotates and the larger the angle, the more time is required and the lower the efficiency. By combining the calculation of the average attitude score and the motion score, the optimal welding sequence that is efficient and flexible can be selected.

[0070] The weldability assessment method based on welding sequence tasks provided in this embodiment calculates the average attitude score to comprehensively consider the overall adaptability of the welding torch attitude in the candidate welding sequence, avoiding interference from single-task attitude issues with the global judgment; it also calculates the total motion score to accurately quantify the positioner motion cost and clearly present the differences in equipment scheduling efficiency. Based on these two key indicators, the optimal sequence is selected, eliminating reliance on experience and establishing an objective and quantifiable evaluation standard for welding process planning. This method can accurately select the solution with good welding torch attitude coordination and efficient positioner motion from the candidate sequences, thereby improving the quality and efficiency of welding operations.

[0071] This embodiment provides a weldability evaluation method based on welding sequence tasks, which can be used in the aforementioned computer system. Figure 5 This is a flowchart of a weldability assessment method based on a welding sequence task according to an embodiment of the present invention. The welding task of the corner component box to be welded is completed using a welding torch and a positioner. Figure 5 As shown, the process includes the following steps:

[0072] Step S201: Obtain the candidate welding sequence of the corner box to be welded, and determine the attitude score of the target welding task by combining the number of feasible welding gun postures of the target welding task in each candidate welding sequence, and determine the motion score of the target welding task by combining the positioner motion posture of the target welding task.

[0073] Specifically, step S201 includes:

[0074] Step S2011: Combine collision analysis to determine the number of feasible welding torch postures for the target welding task. The number of feasible welding torch postures is no more than 3.

[0075] Step S2012: The number of feasible welding torch postures is used as the posture score of the target welding task.

[0076] Specifically, for each weld point in a given candidate welding sequence, the following is defined: Figure 2 The four possible welding machine postures shown are excluded because the posture with the welding torch head pointing downwards (A) is usually not feasible in practice. Collision checks are performed on the remaining three postures. Any posture that does not interfere with the surrounding structure is considered a feasible posture for the welding torch, and the number of feasible postures for the welding torch at each weld point is taken as the corresponding posture score.

[0077] like Figure 6 The figure shows the attitude scoring process of a corner box consisting of three components. The welding between the side panel and the back panel includes two welding tasks, each with an attitude score of 3. The welding between the side panel, the back panel, and the top panel includes four welding tasks. The attitude scores for each welding task are marked in the figure. Based on the attitude scores of each welding task, the average attitude score of the corner box can be calculated to determine that the welding attitude score is 2.67.

[0078] Step S2013: Obtain the rotation angle of each rotating axis in the positioner when the target welding task is being performed.

[0079] Specifically, the positioner can be a single-axis positioner or a multi-axis positioner, depending on the actual situation, obtaining the rotation angle of each rotating axis in the positioner during the target welding task. Taking a dual-axis positioner as an example, it includes a first rotating axis and a second rotating axis, such as... Figure 7 As shown, the values ​​of the first rotation axis include 0°, 90°, 180°, and 270°, and the values ​​of the second rotation axis include 0° and 90°. By combining the values ​​of the first and second rotation axes, the following is obtained: Figure 8 The eight discrete positioner postures shown form the basis for evaluating the positioner motion, and provide multiple operating directions to ensure that each welding task can be performed smoothly from different spatial positions and angles.

[0080] Step S2014: Determine the rotation score of each rotation axis according to the preset mapping relationship between the rotation axis angle and the score.

[0081] Specifically, when designing the scoring mechanism for the welding task sequence, the movement of the positioner in each step is considered to evaluate its impact on welding efficiency and path complexity. For a dual-axis positioner, its posture is determined by the changing angles of the two axes. In a given candidate welding sequence, the movement of the positioner in the target welding task includes: changes in the angle of the first rotation axis (the angle of the first rotation axis changes from one value to another), changes in the angle of the second rotation axis (the angle of the second rotation axis changes from one value to another), and combined changes (the angles of both rotation axes change). To quantify the movement of the positioner in each welding task, scores are assigned to each rotation axis. Taking a dual-axis positioner as an example, the preset mapping relationship between the rotation axis angle and the score is shown in Table 1.

[0082] Table 1

[0083]

[0084]

[0085] The rotation score for each rotating axis is determined based on the actual rotation angle of the two rotating axes during the target welding task.

[0086] Step S2015: The sum of the rotation scores of each rotating axis is taken as the motion score of the target welding task.

[0087] Specifically, the sum of the rotation scores of each rotating axis in the positioner is taken as the motion score of the target welding task. For example, if the first rotating axis changes from 0° to 90° in the welding step of a welding task, and the rotation score of the first rotating axis is -1, and the second rotating axis changes from 0° to 90°, and the rotation score of the second rotating axis is -1, then the motion score of the welding task is: (-1)(rotation score of the first rotating axis) + (-1)(rotation score of the second rotating axis) = -2 points. This is just an example, but it is not limited to this.

[0088] like Figure 9 As shown, the yellow box corresponds to the posture score of the welding task when each component is welded, and the blue box corresponds to the motion score of the welding task when each component is welded. The average posture score of the corner box is calculated to be 2.82 and the motion score is -0.18. This is only an example and is not limited to this.

[0089] The weldability assessment method based on welding sequence tasks provided in this embodiment limits the number of feasible welding torch postures through collision analysis and uses them as posture scores to accurately screen suitable postures, avoid welding interference, and ensure welding quality. It obtains the rotation angle based on at least one rotation axis, calculates the score according to a preset mapping relationship, and sums them as motion scores to refine the positioner motion assessment, accurately reflect motion costs, help optimize equipment scheduling, and improve the collaborative efficiency of the positioner and welding torch in the welding process.

[0090] Step S202: Based on the attitude and motion scores of each welding task in each candidate welding sequence, the Monte Carlo tree search algorithm is used to search for the optimal welding sequence among the candidate welding sequences.

[0091] Specifically, step S202 includes:

[0092] Step S2021: Based on each candidate welding sequence, construct a state-action-transition decision model for the search tree. The state is used to represent the prefix sequence of the completed welding task, and the path cumulative score of the current welding task and the locator's posture in the previous welding task are associated and stored. The action is used to represent the feasible welding torch posture of the current welding task, and the posture score of the current welding task is used to represent the feasible welding torch posture.

[0093] Specifically, the welding sequence optimization problem is transformed into a state-action-transition decision model, where each state s corresponds to the prefix sequence of the first k completed welding tasks, along with the cumulative score G(s) and the position of the locator in the previous welding task, which helps to evaluate the motion cost of subsequent actions. The search process considers possible variations in the sequence and feasible welding postures for each task. The action set starts with the number of selected welding sequences and then expands to all possible postures of the welding torch. For example, for the (k+1)th welding point, the action set A(s) includes all feasible welding torch postures—these postures are derived from a subset of the four standard postures, or, in high-difficulty scenarios, additional candidate postures generated by sampling within a predefined angular range.

[0094] In some optional implementations, after performing action a∈A(s), the formula for calculating the cumulative score of the path for the current welding task when the system transitions to the new state s′ is:

[0095] G(s')=G(s)+pose_score(a)+motion_score(prev_pose→a)

[0096] Where G(s') represents the cumulative path score of the current welding task, G(s) represents the cumulative path score of the previous welding task, pose_score(a) represents the pose score of pose a, and motion_score(prev_pose→a) represents the motion score from the previous welding task to the current welding task. If k+1 = N (i.e., the end of the sequence, where N represents the total number of welding tasks), then s′ is the termination state, and G(s′) is the total score of the complete welding sequence.

[0097] Step S2022: Taking the welding task with an empty prefix sequence as the root node, the sequence search process is executed cyclically starting from the root node. The sequence search process includes: selecting child nodes downwards according to the upper confidence boundary strategy until a node that has not been fully expanded is reached; adding new actions to the node that has not been fully expanded to form a new node; simulating the welding points and feasible postures of the welding gun for the remaining path starting from the new node; determining the simulation path score and feeding it back to the previous node layer by layer to obtain the cumulative score of each node; calculating the score of each path based on the cumulative score of each node until the path with the highest cumulative score is determined as the optimal welding sequence.

[0098] Specifically, to effectively search for the optimal pose assignment strategy among all candidate sequences, the MCTS algorithm is adopted, and the specific process includes:

[0099] (1) Starting from the root node (with an empty prefix), select child nodes downwards according to the Upper Confidence Bound (UCB) strategy of the tree until a node that has not been fully expanded (i.e. a node that has not yet tried all actions) is reached.

[0100]

[0101] Where Q(s,a) is the total cumulative reward obtained from performing action a from state s; N(s,a) and N(s) represent the number of visits to the edge (action) and node (state), respectively, and c represents the exploration coefficient.

[0102] During this process, priority is given to "actions" that have been tried before and scored highly. For example, the action "weld side plate 1 + bottom plate first" has a high total score, so it is given more priority. Occasionally, some "actions" that have not been tried much are also selected to avoid missing potential good solutions. For example, "weld top plate + side plate 3 first" has never been tried before, but it is tried occasionally. This continues until an "unexplored node" is reached, such as a state where there are still untried combinations of "weld points + postures".

[0103] (2) At the selected leaf node sL, if it is not a terminated state, randomly select an untried action a′ from the untried actions, perform a transition from the action set to obtain a new state s, and then add it to the search tree.

[0104] If the "unexplored node" is not the "end point" (meaning that not all solder joints have been soldered), then randomly select one of the "actions" that have not been tried in this node, such as "the next step is to solder the back plate and side plate, using the second feasible posture", and then turn this new action into a "new node" and add it to the "decision tree".

[0105] (3) Starting from s′, perform a “fast random” simulation: For the remaining Nk-1 solder joints, randomly (or heuristically) select feasible postures, accumulate scores until the termination state is reached, and obtain the final simulation score G. sim .

[0106] With a new node (new action), start from this node and quickly "simulate" all the remaining welding steps: for example, if there are 5 welding points left, randomly select the remaining welding points and feasible postures (don't be too precise, just go through them quickly) and weld until the last welding point, and calculate the final total score of this "simulated path".

[0107] (4) Backpropagate the simulated score Gsim to s′ and all its ancestor nodes, and update their Q and N values.

[0108] N(s)←N(s)+1,Q(s)←Q(s)+G sim

[0109] Similarly, for each edge along the path, update:

[0110] N(s,a), Q(s,a)

[0111] The simulated final score is transmitted back along the previous node of the new node, and the information of these nodes is updated: record "how many times this node has been visited" (for example, the root node has one more visit record); add the simulated score to the "cumulative total score" of these nodes.

[0112] The next time a node is selected, the system will know "which path previously scored high and is worth trying more, and which path has not been tried and needs to be explored occasionally".

[0113] The above four steps will be iterated many times. Each iteration will improve the "decision tree" and make the score judgment of each "action" and "sequence" more accurate. Finally, the path with the "highest cumulative total score" is selected from the "decision tree". The welding sequence corresponding to this path is the "optimal sequence" - it has the most reasonable welding torch posture, the lowest positioner movement cost, and the best welding efficiency and quality.

[0114] The weldability assessment method based on welding sequence tasks provided in this embodiment constructs a state-action-transition decision model. Using the prefix sequence as the root node, it iteratively searches using an upper confidence boundary strategy. By simulating the remaining path and receiving feedback scores, it accurately quantifies the impact of each task's attitude and motion scores on the path. It uses a formula to clearly calculate the cumulative score, which not only systematically sorts out the association of welding tasks (completed tasks, locator attitude, etc.), but also efficiently traverses candidates through intelligent search, scientifically finding the optimal sequence with the highest cumulative score. This improves the systematicness, accuracy, and efficiency of welding planning and helps to optimize the intelligent welding process.

[0115] In some optional implementations, the method further includes, before performing the sequence search process cyclically:

[0116] Calculate the number of feasible welding gun postures for each welding task in each welding sequence. If there is a welding task with 0 feasible welding gun postures, then mark the welding sequence as an infeasible sequence.

[0117] Before performing the sequence search process in a loop, infeasible sequences in the original search tree are filtered out, and the remaining sequences are used as the search tree.

[0118] Specifically, in actual welding tasks, there may be situations where no feasible welding posture exists for any welding point in certain candidate sequences. For infeasible sequences, they are first recorded and marked, including the sequence ID, the welding points involved, and the reasons for infeasibility. This information is stored in a dedicated database for subsequent analysis and processing.

[0119] In future welding tasks, when encountering similar infeasible sequences, the system can quickly determine whether the sequence is a known infeasible sequence by querying this database. If so, further exploration of the sequence can be skipped, saving computation time and resources. Furthermore, by analyzing these infeasible sequences, the welding path planning algorithm can be further optimized to avoid generating similar infeasible sequences in subsequent planning, thereby improving the overall efficiency and reliability of the welding path planning system.

[0120] The weldability assessment method based on welding sequence tasks provided in this embodiment calculates the number of feasible welding gun postures for welding tasks, marks and filters out infeasible sequences containing welding tasks with 0 feasible postures, forms a search tree, and preemptively filters out sequences that cannot be welded at all, avoiding wasted effort in subsequent searches, significantly reducing the size of the search tree, reducing computational load, and improving sequence search efficiency. At the same time, it avoids invalid planning caused by infeasible sequences from the source, ensuring that subsequent searches focus on actually executable candidate sequences.

[0121] This embodiment also provides a solderability evaluation device based on a welding sequence task. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0122] This embodiment provides a weldability assessment device based on a welding sequence task, which uses a welding torch and a positioner to complete the welding task of the corner component box to be welded, such as... Figure 10 As shown, it includes:

[0123] The welding task score determination module 1001 is used to obtain the candidate welding sequence of the corner box to be welded, and determine the attitude score of the target welding task by combining the number of feasible welding gun postures of the target welding task in each candidate welding sequence, and determine the motion score of the target welding task by combining the positioner motion posture of the target welding task.

[0124] The optimal sequence search module 1002 is used to search for the optimal welding sequence among the candidate welding sequences based on the attitude score and motion score of each welding task in each candidate welding sequence using the Monte Carlo tree search algorithm.

[0125] In some alternative implementations, the welding task score determination module 1001 includes:

[0126] The feasible attitude number determination unit for welding torch is used to determine the feasible attitude number of welding torch for the target welding task by combining collision analysis. The feasible attitude number of welding torch is no more than 3.

[0127] The attitude determination unit is used to determine the number of feasible attitudes of the welding torch as the attitude score of the target welding task.

[0128] The rotation angle determination unit is used to obtain the rotation angle of each rotating axis in the positioner when the target welding task is being performed.

[0129] The rotation score determination unit is used to determine the rotation score of each rotation axis based on the preset mapping relationship between the rotation axis angle and the score.

[0130] The motion score calculation unit is used to sum the rotation scores of each rotating axis as the motion score of the target welding task.

[0131] In some optional implementations, the optimal sequence search module 1002 includes:

[0132] The model building unit is used to construct a state-action-transition decision model of the search tree based on each candidate welding sequence. The state is used to represent the prefix sequence of the completed welding task and to store the cumulative path score of the current welding task and the locator's posture in the previous welding task. The action is used to represent the feasible welding torch posture of the current welding task, and the posture score of the current welding task is used to represent the feasible welding torch posture.

[0133] The cyclic search unit is used to take the welding task with an empty prefix sequence as the root node. Starting from the root node, it cyclically executes the sequence search process, which includes: selecting child nodes downwards according to the upper confidence boundary strategy until a node that has not been fully expanded is reached; adding new actions to the node that has not been fully expanded to form a new node; simulating the welding points and feasible postures of the welding gun for the remaining path from the new node; determining the score of the simulated path and feeding it back to the previous node layer by layer to obtain the cumulative score of each node; calculating the score of each path based on the cumulative score of each node until the path with the highest cumulative score is determined as the optimal welding sequence.

[0134] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0135] In this embodiment, the solderability evaluation device based on the welding sequence task is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0136] This invention also provides a computer device having the above-described features. Figure 10 The apparatus shown is a weldability assessment device based on welding sequence tasks.

[0137] Please see Figure 11 , Figure 11 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 11As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 11 Take a processor 10 as an example.

[0138] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0139] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0140] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0141] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0142] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0143] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0144] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for weldability assessment based on a welding sequence task, characterized by, The method comprises the following steps of: The welding task of the angle piece box to be welded is completed by using a welding torch and a positioner, and the method comprises the following steps of: A candidate welding sequence of the angle piece box to be welded is acquired, and a pose score of a target welding task is determined in combination with a number of feasible poses of the welding torch of the target welding task in the candidate welding sequence; 2. The method of claim 1, wherein, An optimal welding sequence in the candidate welding sequence is searched by using a Monte Carlo tree search algorithm based on the pose score and the motion score of each welding task in each candidate welding sequence. The candidate welding sequence comprises a plurality of welding tasks, and the pose score of the target welding task is determined in combination with the number of feasible poses of the welding torch of the target welding task in each candidate welding sequence, and the pose score of the target welding task comprises the following steps of: The number of feasible poses of the welding torch of the target welding task is determined in combination with collision analysis, and the number of feasible poses of the welding torch is not greater than 3; 3. The method of claim 1, wherein, The number of feasible poses of the welding torch is taken as the pose score of the target welding task. The positioner comprises at least one rotating shaft, and the motion score of the target welding task is determined in combination with a motion pose of the positioner of the target welding task, and the motion score of the target welding task comprises the following steps of: The rotating angles of the rotating shafts in the positioner when the target welding task is welded are acquired; The rotating scores of the rotating shafts are determined according to a mapping relationship between a preset rotating shaft angle and a score; 4. The method of claim 1, wherein, The sum of the rotating scores of the rotating shafts is taken as the motion score of the target welding task. The method further comprises the following steps of: The pose scores of the welding tasks and the number of the welding tasks in the candidate welding sequence are acquired, and the average value of the pose scores of the welding tasks is calculated as an average pose score of the candidate welding sequence; The motion scores of the welding tasks in the candidate welding sequence are acquired, and the sum of the motion scores of the welding tasks is calculated as a motion score of the candidate welding sequence; 5. The method of claim 1, wherein, An optimal welding sequence is selected from the candidate sequence based on the average pose score and the motion score of each candidate welding sequence. An optimal welding sequence in the candidate welding sequence is searched by using a Monte Carlo tree search algorithm based on the pose score and the motion score of each welding task in each candidate welding sequence, and the optimal welding sequence comprises the following steps of: A state-action-transition decision model of a search tree is constructed based on each candidate welding sequence, wherein the state is used to represent a prefix sequence of the completed welding tasks, and is used to store the path cumulative score of the current welding task and the pose of the positioner in the last welding task; the action is used to represent a feasible welding torch pose of the current welding task, and the pose score of the current welding task is used to represent the feasible welding torch pose; 6. The method of claim 5, wherein, A welding task with an empty prefix sequence is taken as a root node, and a sequence search process is circularly executed from the root node, and the sequence search process comprises the following steps of: according to an upper confidence bound strategy, a child node is selected downwardly until a node that has not been completely expanded is reached, a new action is added to the node to form a new node, a welding point and a feasible welding torch pose of a remaining path are simulated from the new node, a simulation path score is determined and is fed back to a last node layer by layer to obtain a cumulative score of each node, a path score is calculated based on the cumulative score of each node, and a path with the highest cumulative score is determined as an optimal welding sequence. In the state-action-transition decision model, the calculation formula of the path cumulative score of the current welding task is: G(s') = G(s) + pose_score(a) + motion_score(prev_pose→a) wherein G(s') represents the path cumulative score of the current welding task, G(s) represents the path cumulative score of the previous welding task, pose_score(a) represents the pose score of the pose a, and motion_score(prev_pose→a) represents the motion score from the previous welding task to the current welding task.

7. The method of claim 5, wherein, Before the sequence search process is executed in a loop, the method further comprises: calculating the number of feasible poses of the welding torch for each welding task in each welding sequence, and if there is a welding task with a number of feasible poses of the welding torch being 0, marking the welding sequence as an infeasible sequence; before the sequence search process is executed in a loop, filtering out the infeasible sequences in the original search tree as the search tree.

8. A weldability evaluation device based on a welding sequence task, characterized by, The welding task is completed by using a welding torch and a positioner on the box of angle pieces to be welded, and the device comprises: a welding task score determination module configured to obtain candidate welding sequences of the box of angle pieces to be welded, and determine a pose score of a target welding task in combination with the number of feasible poses of the welding torch for the target welding task in each candidate welding sequence, and determine a motion score of the target welding task in combination with a motion pose of the positioner for the target welding task; an optimal sequence search module configured to search for an optimal welding sequence in the candidate welding sequences by using a Monte Carlo tree search algorithm based on the pose score and the motion score of each welding task in each candidate welding sequence.

9. A computer device, comprising: comprise: a memory and a processor, which are communicatively connected to each other, and the memory stores computer instructions, and the processor executes the computer instructions to perform the method in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, and the computer instructions are used to make a computer execute the method in any one of claims 1 to 7.

Images