A collaborative control method and system for an automobile rail production line

Abstract

The application discloses a kind of collaborative control method and system for automobile longitudinal beam production line, its method includes: to the longitudinal beam production line is prevented and is checked;Longitudinal beam single piece is pretreated to obtain longitudinal beam base piece;Longitudinal beam base piece is sent into first target station and executes welding procedure, obtains semi-finished product longitudinal beam, obtains first attitude data;Second target station corresponding operation data and second attitude data are obtained;First process relay compensation domain is determined based on first attitude data and second attitude data;Dynamic safety domain is generated based on operation data;Process rhythm parameter is determined according to first process relay compensation domain and operation data;The dynamic pre-reliance point of mechanical arm in third target station and the clamping compensation value on second clamp are determined according to process rhythm parameter;The action reference template of second clamp is determined;Second clamp and the mechanical arm in third target station are respectively planned to generate collaborative strategy.The application can effectively improve the collaborative efficiency of automobile longitudinal beam production line.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of collaborative control of automotive component production lines, and particularly to a collaborative control method and system for automotive longitudinal beam production lines. Background Art

[0002] With the rapid development of the automotive manufacturing industry towards flexibility and customization, more and more vehicle manufacturers adopt the multi-model mixed-line production method to improve production capacity utilization and market response speed. In this context, as an important load-bearing component of the vehicle body structure, the production collaboration efficiency of automotive longitudinal beams directly affects the vehicle assembly quality and production cost. Therefore, the collaborative control ability of longitudinal beam production lines has become particularly crucial.

[0003] Currently, automotive longitudinal beam production lines usually use programmable logic controllers in a fixed mode to control the production line. The programmable logic controller divides the production line into multiple independent workstations or areas, and each workstation independently completes its preset process tasks. The host computer is mainly responsible for simple fixed-sequence control, status monitoring, and data collection of each workstation, and the data interaction and linkage response capabilities between workstations are relatively limited. However, the above technologies often cause problems such as insufficient information interaction and lack of real-time collaboration ability between workstations. When the production line faces product switching, process adjustment, or sudden failure, due to the lack of a deeper global collaborative optimization mechanism, it is difficult for each independent control unit to quickly and effectively perform linkage responses, resulting in unsmooth material flow, difficult fault location, and further affecting the production efficiency, flexibility, and product quality stability of the entire line, and even causing waste of energy and materials.

[0004] Therefore, there is an urgent need for a collaborative control method and system for automotive longitudinal beam production lines to solve the above problems. Summary of the Invention

[0005] The embodiments of the present application provide a collaborative control method and system for automotive longitudinal beam production lines to improve the collaborative efficiency of automotive longitudinal beam production lines.

[0006] To achieve the above objective, the embodiments of the present application adopt the following technical solutions:

[0007] In the first aspect, a collaborative control method for automotive longitudinal beam production lines is provided, which is applied to automotive longitudinal beam production lines. The automotive longitudinal beam production line includes at least two fixtures, workstations, and robotic arms. The method includes:

[0008] Perform anti-fooling verification on the longitudinal beam production line;

[0009] When the longitudinal beam production line passes the anti-fooling verification, perform preprocessing on the longitudinal beam single-piece to obtain a longitudinal beam basic part;

[0010] The longitudinal beam base component is locked by the first clamp and sent into the first target station to perform the welding process, thereby obtaining a semi-finished longitudinal beam and acquiring the first posture data of the semi-finished longitudinal beam.

[0011] When the second target station receives the semi-finished longitudinal beam, it acquires the operation data corresponding to the second target station and the second posture data of the semi-finished longitudinal beam on the second fixture.

[0012] The first process relay compensation domain is determined based on the first attitude data and the second attitude data.

[0013] A dynamic safety domain is generated based on the operation data corresponding to the second target workstation.

[0014] Obtain the initial process rhythm baseline for the second target station;

[0015] Using the dynamic safety domain as a constraint, the initial process rhythm benchmark is corrected based on the operation data corresponding to the first process relay compensation domain and the second target station, and the process rhythm parameters are determined.

[0016] The dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture are determined based on the process rhythm parameters.

[0017] The motion reference template of the second fixture is determined by the clamping compensation value on the second fixture.

[0018] Based on the dynamic pre-positioning points and motion reference templates, collaborative strategies are planned and generated for the robotic arms in the second fixture and the third target workstation, respectively.

[0019] In a preferred embodiment, the error-proofing verification of the longitudinal beam production line includes:

[0020] Initialize the fixture control module and signal acquisition module in the production line;

[0021] Verify the effectiveness of signal acquisition by the signal acquisition module;

[0022] After the signal acquisition module passes the signal acquisition validity verification, the preset foolproof sample is placed in the target tooling fixture, and it is determined whether the positioning point of the preset foolproof sample is consistent with that of the target tooling fixture.

[0023] When the preset error-proof sample and the target tooling fixture are aligned at their positioning points, control the target tooling fixture to execute the clamping command.

[0024] The signal acquisition module is activated to acquire the clamping status signal and error-proof identification signal of the target tooling fixture;

[0025] Based on the clamping status signal and the error prevention identification signal, determine whether the target tooling fixture and the preset error prevention sample meet the preset error prevention logic conditions.

[0026] After the preset error-proofing logic conditions are met, the target tooling fixture is controlled to execute a reset command, thus completing the error-proofing verification.

[0027] In a preferred embodiment, the longitudinal beam component includes a first longitudinal beam component, a second longitudinal beam component, a third longitudinal beam component, and a fourth longitudinal beam component. The preprocessing of the longitudinal beam components to obtain the longitudinal beam base component includes:

[0028] The first longitudinal beam component is placed in the first pre-processing station using the first pre-processing fixture and welded to obtain the first spare component.

[0029] The second longitudinal beam component is placed in the second pretreatment station using the second pretreatment fixture and welded to obtain the second spare component.

[0030] The first spare part and the second spare part are placed in the third pretreatment station using the third pretreatment fixture and welded to obtain the first longitudinal beam assembly;

[0031] The third longitudinal beam component is placed in the fourth pre-processing station using the fourth pre-processing fixture and welded to obtain the third spare component.

[0032] The fourth longitudinal beam component is placed in the fifth pre-processing station using the fifth pre-processing fixture and welded to obtain the fourth spare component.

[0033] The third and fourth spare parts are placed in the sixth pretreatment station using the fifth pretreatment fixture and welded to obtain the second longitudinal beam assembly.

[0034] The longitudinal beam base component is obtained by using the first longitudinal beam assembly and the second longitudinal beam assembly.

[0035] In a preferred embodiment, determining the first process relay compensation domain based on the first attitude data and the second attitude data includes:

[0036] A first key point dataset is extracted based on the first posture data. The first key point dataset includes the first weld positioning point data, the first fixture positioning center point data, and the first reinforcing rib support point data.

[0037] The second key point dataset is extracted based on the second posture data. The second key point dataset includes the second weld positioning point data, the second fixture positioning center point data, and the second reinforcing rib support point data.

[0038] The first and second key point datasets are mapped to the reference coordinate system, and the deflection curve deviation, torsional attitude deviation and planar positioning deviation are calculated based on the first and second key point datasets.

[0039] The deviation type is determined by combining the deflection curve deviation, torsional attitude deviation, and planar positioning deviation. The deviation types include clamping positioning inaccuracy and transport deformation deviation.

[0040] When the deviation type is transport deformation deviation, a regionalized deformation model of the longitudinal beam is constructed by combining deflection curve deviation, torsional attitude deviation and planar positioning deviation.

[0041] The longitudinal beam regional deformation model is divided into three subdomains: weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain.

[0042] The first process relay compensation domain is determined by combining the weld zone compensation subdomain, the auxiliary support compensation subdomain, and the positioning constraint zone compensation subdomain.

[0043] In a preferred embodiment, the step of dividing the longitudinal beam regional deformation model to determine the weld zone compensation subdomain, the auxiliary support compensation subdomain, and the positioning constraint zone compensation subdomain includes:

[0044] The longitudinal beam skeleton curve is obtained by fitting the longitudinal beam regional deformation model.

[0045] The longitudinal beam skeleton curve is discretized to obtain the discrete points of the curve, and the curvature value of each discrete point is calculated to obtain the skeleton curvature distribution of the semi-finished longitudinal beam.

[0046] The rate of change of curvature between discrete points of two adjacent curves is calculated using the skeleton curvature distribution.

[0047] The region in the longitudinal beam regional deformation model whose curvature change rate is greater than or equal to the preset first threshold is taken as the weld zone compensation subdomain.

[0048] The region in the longitudinal beam regional deformation model whose rate of curvature change is greater than or equal to the preset second threshold and less than the preset first threshold is designated as the auxiliary support compensation subdomain.

[0049] The region in the longitudinal beam regional deformation model with a curvature change rate less than a preset third threshold is taken as the location constraint area compensation subdomain.

[0050] Among them, the preset first threshold is greater than the preset second threshold and the preset second threshold is greater than the preset third threshold.

[0051] In a preferred embodiment, generating a dynamic safety domain based on the operation data corresponding to the second target workstation includes:

[0052] Obtain the workstation geometry layout of the second target workstation;

[0053] Based on the operation data corresponding to the second target workstation, obtain the pose parameters and motion parameters of the robotic arm in the second target workstation, as well as the motion range of the second fixture and the movement position information of the operator.

[0054] An initial virtual fence boundary is constructed based on the workstation geometric layout, and the initial virtual fence boundary is divided into a core danger zone, a buffer warning zone, and a safe operation zone.

[0055] The initial virtual fence boundary is extended based on the operator's movement location information to obtain the first virtual fence range, which is located outside the buffer warning zone.

[0056] Using the first virtual fence range as a constraint benchmark, the initial virtual fence boundary is extended by combining the pose and motion parameters of the robotic arm to obtain the second virtual fence range. The extended boundary of the second virtual fence range does not break through the boundary of the first virtual fence range.

[0057] Using the first and second virtual fence ranges as dual constraints, the third virtual fence range is obtained by locally thickening the initial virtual fence boundary according to the movement range of the second clamp.

[0058] A dynamic security domain is generated by combining the first virtual fence range, the second virtual fence range, and the third virtual fence range.

[0059] In a preferred embodiment, the step of extending the initial virtual fence boundary to obtain the second virtual fence range, using the first virtual fence range as a constraint reference and combining the pose and motion parameters of the robotic arm, includes:

[0060] The spatial operation trajectory of the robotic arm is determined by the pose parameters of the robotic arm.

[0061] Using the first virtual fence range as a constraint benchmark, the extended boundary area of ​​the second virtual fence range is determined in combination with the spatial operation trajectory of the robotic arm;

[0062] The dynamic expansion rate of the extended boundary region is determined based on the robotic arm's motion parameters.

[0063] The range of the second virtual fence is determined by combining the dynamic expansion rate of the extended boundary area, wherein the extended boundary does not exceed the range of the first virtual fence.

[0064] In a preferred embodiment, determining the dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture based on the process rhythm parameters includes:

[0065] Obtain the current pose data of the robotic arm in the third target workstation;

[0066] The timing window for the robotic arm to receive the semi-finished longitudinal beam and the docking time window for the second fixture in the third target station are predicted using a pre-trained timing prediction model based on process rhythm parameters.

[0067] Based on the prediction time window and the current pose data of the robotic arm, a dynamic pre-positioning domain is generated;

[0068] Within the dynamic pre-positioning domain, the optimal pre-positioning point sequence of the robotic arm is calculated by combining process rhythm parameters;

[0069] Calculate the shortest movement time for the robotic arm to reach each optimal pre-position sequence;

[0070] Generate the time-matching pre-positioning trajectory of the robotic arm based on the optimal pre-positioning point sequence and the predicted time window;

[0071] The reachable pre-deployment point is determined by combining the shortest motion time with the time-matched pre-deployment trajectory;

[0072] The reachable pre-departure point with the smallest time difference between the shortest travel time and the predicted time window is selected as the dynamic pre-departure point;

[0073] Within the docking time window of the second fixture, the process rhythm parameters are input into the clamping compensation regression model to obtain the clamping compensation value on the second fixture.

[0074] Secondly, this application provides a machine-readable storage medium storing instructions for causing a machine to execute the aforementioned collaborative control method for an automotive longitudinal beam production line.

[0075] Thirdly, this application provides a collaborative control system for an automotive longitudinal beam production line, comprising:

[0076] The memory is configured to store instructions; and

[0077] The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the aforementioned collaborative control method for an automotive longitudinal beam production line.

[0078] The above technical solution first performs error-proofing checks on the production line to ensure that the fixtures and signal acquisition modules are in the correct state, avoiding safety accidents caused by human error and improving production efficiency. Then, a standardized preprocessing process ensures the consistency of each longitudinal beam base component, providing high-quality input for subsequent processes and reducing assembly problems caused by differences in base components. By acquiring the first posture data of the semi-finished longitudinal beam, the initial state information of the current semi-finished longitudinal beam can be grasped in a timely manner, facilitating adjustments in subsequent processes. Acquiring the operation data corresponding to the second target station and the second posture data of the semi-finished longitudinal beam on the second fixture allows for real-time understanding of the semi-finished longitudinal beam's state at the second station, facilitating subsequent compensation processing. By determining the relay compensation domain of the first process, deformation and positioning deviations during transfer can be understood in a timely manner, improving the accuracy of semi-finished longitudinal beam transfer between stations. A dynamic safety domain is generated from the operation data corresponding to the second target station, forming real-time constraints, reducing collision risks, and improving the collaborative efficiency of the automotive longitudinal beam production line. Next, by determining the process rhythm parameters, we can ensure coordinated movements between each process, optimize production rhythm, reduce waiting time and equipment idle time, and enable the production line to operate in the most efficient way while ensuring safety, thereby improving overall production efficiency. Determining the dynamic pre-positioning point of the robotic arm in the third target station and the clamping compensation value on the second fixture can make the movements of the robotic arm and fixture more coordinated, reduce conflicts and interference between equipment, and improve equipment synergy. By determining the motion reference template of the second fixture, we can reduce fixture wear and extend equipment life. Planning and generating a coordination strategy for the second fixture and the robotic arm in the third target station can make the equipment movements of the entire production line more coordinated, improving production efficiency and product quality.

[0079] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0080] Figure 1 A flowchart illustrating a collaborative control method for an automotive longitudinal beam production line, provided as an embodiment of this application;

[0081] Figure 2 A schematic diagram of a process for error-proofing verification of a longitudinal beam production line is provided for an embodiment of this application;

[0082] Figure 3 This is a structural schematic diagram of a longitudinal beam foundation component provided in an embodiment of this application. Detailed Implementation

[0083] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0084] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0085] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0086] Figure 1 This illustration schematically shows a flowchart of a collaborative control method for an automotive longitudinal beam production line according to an embodiment of this application. Figure 1 As shown in the figure, this application provides a collaborative control method for an automotive longitudinal beam production line. The automotive longitudinal beam production line includes at least two fixtures, workstations, and robotic arms. The method may include the following steps.

[0087] S101. Perform error-proofing verification on the longitudinal beam production line;

[0088] S102. After the longitudinal beam production line passes the error-proofing check, the longitudinal beam individual parts are pre-processed to obtain the longitudinal beam base parts.

[0089] S103. The longitudinal beam base component is locked by the first fixture and sent to the first target station to perform the welding process, so as to obtain the semi-finished longitudinal beam and acquire the first posture data of the semi-finished longitudinal beam.

[0090] S104. When the second target station receives the semi-finished longitudinal beam, obtain the operation data corresponding to the second target station and the second posture data of the semi-finished longitudinal beam on the second fixture.

[0091] S105. Determine the first process relay compensation domain based on the first attitude data and the second attitude data;

[0092] S106. Generate a dynamic safety domain based on the operation data corresponding to the second target workstation;

[0093] S107. Obtain the initial process rhythm reference for the second target station;

[0094] S108. Using the dynamic safety domain as a constraint, the initial process rhythm benchmark is corrected based on the operation data corresponding to the first process relay compensation domain and the second target station, and the process rhythm parameters are determined.

[0095] S109. Determine the dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture based on the process rhythm parameters.

[0096] S110. Determine the motion reference template of the second fixture by using the clamping compensation value on the second fixture;

[0097] S111. Based on the dynamic pre-positioning point and motion reference template, plan and generate collaborative strategies for the robotic arms in the second fixture and the third target workstation, respectively.

[0098] First, to ensure the safety and machining accuracy of the automotive longitudinal beam production line, a mistake-proofing check is performed before production. The fixture control module and signal acquisition module in the production line are initialized, and the signal acquisition effectiveness of the signal acquisition module is verified to ensure that the fixtures and sensors can respond normally to control commands and provide correct feedback. Then, a pre-set mistake-proofing sample is placed in the target tooling fixture, and it is determined whether the positioning point of the mistake-proofing sample matches the positioning point of the target tooling fixture to ensure correct workpiece clamping. If the positioning is confirmed to be consistent, the target tooling fixture is controlled to execute the clamping command, and the signal acquisition module is activated to collect clamping status signals and mistake-proofing identification signals. Based on the collected clamping status signals and mistake-proofing identification signals, it is determined whether the target tooling fixture and the mistake-proofing sample meet the preset error-proofing logic conditions. When the fixture and the mistake-proofing sample meet the preset conditions, the target tooling fixture is controlled to execute the reset command, completing the mistake-proofing check. This mistake-proofing check can promptly detect problems such as fixture abnormalities, signal failures, or incorrect workpiece loading before production, thereby effectively avoiding operational errors and improving the safety of the longitudinal beam production line.

[0099] After the longitudinal beam production line completes the error-proofing check and passes the verification, the individual longitudinal beam parts are then pre-processed to obtain the basic longitudinal beam components that can be used for subsequent welding processes. The individual longitudinal beam parts are placed in their respective pre-processing fixtures and stations for necessary welding, assembly, and positioning operations to generate spare parts for each part. Subsequently, multiple spare parts are placed in a pre-defined assembly fixture for welding or assembly according to a preset combination sequence, thus obtaining the longitudinal beam assembly. Finally, by further combining the longitudinal beam assemblies, the basic longitudinal beam component is obtained.

[0100] Subsequently, the longitudinal beam base component is placed in the first fixture, and the fixture is locked to maintain its fixed state during subsequent transport and processing. The control system drives the first fixture to transport the clamped longitudinal beam base component to a preset first target station, where a welding process corresponding to the vehicle model is performed to form a semi-finished longitudinal beam after preliminary welding. After the welding process is completed, the attitude acquisition device set at the first target station collects the first attitude data of the longitudinal beam base component. The first attitude data is characteristic data representing the spatial position, attitude angle, etc. of the semi-finished longitudinal beam, including but not limited to the lateral torsion angle (torsion angle along the width direction of the vehicle body) and vertical warpage height (warpage amount along the height direction of the vehicle body).

[0101] When the detection device determines that the second target station has received the semi-finished longitudinal beam transported from the first target station, the control system initiates the data acquisition process for that station. Specifically, it acquires the operation data corresponding to the second target station, including but not limited to the current process execution status of the station, the motion parameters of the tooling fixture, the robot's operating parameters, and equipment control information related to the process rhythm. Simultaneously, the attitude detection unit installed on the second fixture acquires the spatial attitude of the clamped semi-finished longitudinal beam to obtain the second attitude data of the semi-finished longitudinal beam in the second target station. The second attitude data is used to characterize the actual position and attitude angle of the semi-finished longitudinal beam, including but not limited to three-dimensional coordinate position data, pitch / yaw / roll angle data, and spatial offset of key positioning feature points.

[0102] After obtaining the first and second attitude data, the first process relay compensation domain is determined based on these data. This involves analyzing key points, identifying deviations, and modeling regional compensation for the attitude differences of the same half-finished longitudinal beam at the first and second target workstations to generate a spatial compensation range for the first process relay control. Specifically, firstly, the first weld positioning point data, the first fixture positioning center point data, and the first reinforcing rib support point data are extracted from the first attitude data to form a first key point dataset. Then, the second weld positioning point data, the second fixture positioning center point data, and the second reinforcing rib support point data are extracted from the second attitude data to form a second key point dataset. Subsequently, the first and second key point datasets are mapped to a preset reference coordinate system. By performing differential calculations on the corresponding key point data, the deflection curve deviation, torsional attitude deviation, and planar positioning deviation of the longitudinal beam are obtained to characterize the positional changes and deformation characteristics of the longitudinal beam during the cross-workstation transfer process. Based on the aforementioned deviation results, the deviation types are determined, including clamping and positioning inaccuracies and transport deformation deviations. When the deviation type is determined to be a transport deformation deviation, a regionalized deformation model of the longitudinal beam is further constructed based on the deflection curve deviation, torsional posture deviation, and planar positioning deviation. This model is then used to divide the longitudinal beam into weld zone compensation subdomains, auxiliary support compensation subdomains, and positioning constraint zone compensation subdomains. Finally, combining the spatial characteristics and deviation compensation requirements of each compensation subdomain, a first-process relay compensation domain is determined to guide the execution of the first-process relay action, thereby achieving precise compensation and workstation collaborative control for the initial welding process of the longitudinal beam.

[0103] Generating a dynamic safety domain based on the operational data corresponding to the second target workstation refers to dividing the workstation space into zones based on the real-time operating status and personnel distribution information of the second target workstation, in order to construct a safety protection domain that changes in real time with the operation process. Specifically, the geometric layout of the second target workstation forms a basic spatial model for constructing a virtual fence; then, based on the operational data, the pose and motion parameters of the robotic arm in the second target workstation, the range of motion of the second gripper, and the movement position information of the operator are obtained. An initial virtual fence boundary is constructed based on the workstation geometry, dividing the workstation space into a core danger zone, a buffer warning zone, and a safe operating zone. The initial virtual fence boundary is extended according to the operator's movement position information to obtain a first virtual fence range outside the buffer warning zone, which is used to ensure a safe distance when personnel approach the equipment area. On this basis, using the first virtual fence range as a constraint benchmark, the initial virtual fence boundary is dynamically adjusted in combination with the pose and motion parameters of the robotic arm to obtain a second virtual fence range, and its extended boundary does not exceed the boundary limit of the first virtual fence range. Furthermore, using the first and second virtual fence ranges as dual safety constraints, the initial virtual fence boundary is locally thickened according to the movement range of the second fixture to form a third virtual fence range. Finally, by integrating the first, second, and third virtual fence ranges, the hazard level of the workstation space is hierarchically fused to generate a dynamic safety domain for real-time safety management, thereby achieving dynamic risk protection and human-machine collaborative operation safety assurance for the second target workstation.

[0104] The initial process rhythm baseline for the second target workstation is obtained. In this embodiment, the initial process rhythm baseline refers to the initial set of parameters used to provide time and rhythm references for the motion planning of the second target workstation. It serves as the starting point for process rhythm parameters and is used for subsequent correction and optimization within the dynamic safety domain. The initial process rhythm baseline includes, but is not limited to, standard rhythm data of robotic arm operation, standard timing data of fixtures, and workstation operation interval data. Standard rhythm data of robotic arm operation describes the average running speed, acceleration, pose change time, and time interval between reaching each critical pre-contact point of the robotic arm during welding, handling, or assembly operations. Standard timing data of fixtures includes the sequence and duration of fixture gripping, clamping, releasing, and resetting actions, used to determine the coordination rhythm between the fixture and the robotic arm. Workstation operation interval data describes the relay time window of the semi-finished longitudinal beam between preceding and following processes, including handling time, dwell time, and process switching interval, providing a reference for rhythm parameters. This solution introduces an initial process rhythm benchmark as an external anchor point and limits the dynamic safety domain to an independent safety constraint generated based on the inherent operation data of the workstation. At the same time, it utilizes the first process relay compensation domain to carry the existing deviations across processes, thereby constructing the process rhythm parameter determination process into a one-way parameter correction process subject to safety constraints.

[0105] Using a dynamic safety domain as a constraint, and based on the operation data corresponding to the first process relay compensation domain and the second target workstation, the initial process rhythm benchmark is corrected to determine the process rhythm parameters. First, the initial process rhythm benchmark is mapped to the preliminary motion trajectory of the robotic arm. Specifically, this involves combining the velocity and acceleration in the initial process rhythm benchmark with the operational requirements of the third target workstation (such as the target position for grasping the semi-finished longitudinal beam and the assembly point). A polynomial interpolation algorithm is used to plan the preliminary motion trajectory of the robotic arm. The preliminary motion trajectory includes the complete path of the robotic arm from its current standby position to the material picking point next to the second fixture, and then to the work point of the third target workstation. The "time-space" mapping relationship of the trajectory is output, that is, the coordinates (x(t), y(t), z(t)) and attitude R(t) of the robotic arm end effector corresponding to each timestamp t. By solving the forward kinematics (using the DH parameter method), the spatial position of all links of the robotic arm at each moment on the preliminary trajectory is calculated, generating the preliminary motion envelope Einit of the robotic arm (including the entire spatial occupancy range of links, joints, and end effectors). Subsequently, by combining the dynamic safety domain to verify the safety of the initial trajectory, the initial motion envelope E of the robotic arm is... init Perform a geometric intersection operation with the boundary of the dynamic safety domain (such as the clamp's range of motion, safety redundancy distance boundary), if E init It has no intersection with the dynamic security domain (i.e., E) init ∩Dynamic safety domain = empty set), indicating that the initial trajectory is safe, and the following steps can be implemented. That is, (E ini (Some areas are outside the safety domain or intrude into the clamp's operating range), mark the coordinates of the interference points and their corresponding timestamps t. error (For example, the robotic arm link intrudes into the gripper's motion range at t=2.5s). Combining the attitude deviation value of the first process's relay compensation domain, determine whether the initial trajectory can adapt to the relay requirements. For example, if the semi-finished longitudinal beam has a 3° attitude deviation, it is necessary to verify whether the robotic arm's grasping posture in the initial trajectory can cover this deviation through compensation adjustment. If it cannot be covered, mark it as a "compensation adaptation insufficient" problem. Next, the initial process rhythm benchmark is corrected layer by layer. For the previously marked interference points, the trajectory is optimized by adding obstacle avoidance transition points. Specifically, transition points are added at reasonable positions before and after the interference points (e.g., adding an obstacle avoidance point 100mm before the robotic arm intrudes into the gripper's range), the obstacle avoidance path is replanned, or the motion envelope E of the optimized trajectory is recalculated. opt1 Then, an interference check is performed again with the dynamic safety domain. If the interference is eliminated, the process proceeds to the step of combining the first process relay compensation domain with the process requirements to finally solidify the process rhythm parameters. If interference still exists, the process rhythm parameters are adjusted appropriately to adapt to safety constraints. Parameter adjustments for interference issues specifically involve reducing the movement speed of the corresponding section (e.g., from (v...)) if the robotic arm cannot avoid obstacles in a narrow space due to excessive movement speed.init =0.2m / s decreases to v adj =0.15m / s), and at the same time, recalculate the movement time according to t=L / v (L is the length of the road segment) to ensure that the total time does not exceed the upper limit of the scheduling requirements; if the motion envelope suddenly intrudes into the safe domain due to excessive acceleration, reduce the acceleration to a adj ≤a max (a max (Maximum acceleration threshold for smooth robotic arm movement), avoiding sudden acceleration / deceleration. Parameter adjustments for relay compensation adaptability involve appropriately extending the "material handling dwell time" in the initial rhythm benchmark when posture deviations lead to insufficient robotic arm grasping time (e.g., from t...). stop,init =0.5s extended to t stop,adj =0.8s), to allow time for compensation and adjustment; based on the adjusted parameters, the optimized trajectory E is regenerated. opt2 The process is repeated until the dynamic safety domain constraints are met. Finally, combining the first process relay compensation domain with the process requirements, the process rhythm parameters are finalized to ensure that the corrected parameters are both safe and adaptable to relay requirements and process efficiency. The trajectory corresponding to the corrected rhythm parameters is matched with the allowable compensation range of the first process relay compensation domain to confirm that the robotic arm can complete the relay of the semi-finished longitudinal beam by adjusting its posture / position under this rhythm (e.g., the corrected gripping position deviation ≤ allowable deviation). The total time of all processes after correction (movement time + process time + dwell time) is calculated to ensure that the total time does not exceed the upper limit of the production line scheduling requirements, while the time of core processes (such as welding) is not less than the minimum process requirements. If the above verifications are all passed, the corrected speed, acceleration, movement time, dwell time and other parameters are integrated to form the final set of process rhythm parameters, i.e., process rhythm parameters.

[0106] Based on the process rhythm parameters, the dynamic pre-positioning point of the robotic arm in the third target station and the clamping compensation value on the second fixture are determined. In other words, for the robotic arm and fixture in the third target station, before receiving the semi-finished longitudinal beam, dynamic position pre-setting and clamping parameter optimization are achieved through prediction and planning. After obtaining the current pose data of the robotic arm in the third target station, combined with the process rhythm parameters, a pre-trained time-series prediction model predicts the time window for the robotic arm to receive the semi-finished longitudinal beam and the docking time window for the second fixture. Based on this, a dynamic pre-positioning domain is generated according to the predicted time window and the current pose data of the robotic arm. Within this domain, the optimal pre-positioning point sequence of the robotic arm is calculated in conjunction with the process rhythm parameters, and the shortest movement time for the robotic arm to reach each optimal pre-positioning point is also calculated. Subsequently, a time-matching pre-positioning trajectory of the robotic arm is generated based on the optimal pre-positioning point sequence and the predicted time window. The reachable pre-positioning point is determined by combining the shortest movement time with the time-matching pre-positioning trajectory. By selecting the point with the smallest time difference between the shortest movement time and the predicted time window among the reachable pre-positioning points, the dynamic pre-positioning point of the robotic arm in the third target station is finally determined. Within the docking time window of the second fixture, the process rhythm parameters are input into the clamping compensation regression model to obtain the clamping compensation value on the second fixture, thereby achieving precise control and collaborative operation of the robotic arm and fixture on the semi-finished longitudinal beam.

[0107] The clamping compensation value calculated based on process rhythm parameters and dynamic pre-alignment points is converted into an action plan that can be directly executed by the second fixture control system. The clamping compensation value on the second fixture includes, but is not limited to, information such as clamping force, clamping position, and clamping angle used to correct positional offsets, posture deviations, and local deformations of the semi-finished longitudinal beam during the relay process. Subsequently, based on the clamping compensation value of the second fixture, such as its structural parameters, joint constraints, and reachable space, the compensation amount is converted into displacement and torque commands for each joint or actuator of the fixture, and a continuous motion trajectory from the current position of the fixture to the target clamping position is planned. The motion trajectory includes time nodes, fixture posture, and corresponding clamping force parameters, forming a complete action sequence. Finally, this action sequence is organized into an action reference template for direct use by the second fixture control system, achieving precise clamping, posture compensation, and safe relay of the semi-finished longitudinal beam, thereby ensuring the stability and reliability of process collaborative control.

[0108] Subsequently, based on the dynamic pre-positioning point of the robotic arm at the third target station and the motion reference template of the second fixture, the movements of the robotic arm and fixture are jointly planned by combining process rhythm parameters, dynamic safety domain, and kinematic constraints of the robotic arm and fixture to achieve precise relay of the semi-finished longitudinal beam. After determining the optimal motion trajectory of the robotic arm from its current position to the dynamic pre-positioning point, the time sequence of clamping and positioning actions is planned according to the motion reference template of the fixture. Then, the robotic arm trajectory and the fixture motion sequence are aligned in time and space so that when the robotic arm reaches the dynamic pre-positioning point, the second fixture is already in the predetermined clamping state, ensuring the synchronicity and accuracy of the relay action. Finally, the coordinated motion sequence of the robotic arm and fixture is formed into an executable control strategy to guide the robotic arm and the second fixture at the third target station under the constraints of the dynamic safety domain, thereby improving the reliability and safety of process collaborative control.

[0109] Performing error-proofing checks on the production line ensures that fixtures and signal acquisition modules are in the correct state, preventing safety accidents caused by human error and improving production efficiency. Subsequently, a standardized pre-processing procedure ensures the consistency of each longitudinal beam base component, providing high-quality input for subsequent processes and reducing assembly problems caused by differences in base components. By acquiring the first posture data of the semi-finished longitudinal beam, the initial state information of the current semi-finished longitudinal beam can be grasped in a timely manner, facilitating adjustments in subsequent processes. Acquiring the operation data corresponding to the second target station and the second posture data of the semi-finished longitudinal beam on the second fixture allows for real-time understanding of the semi-finished longitudinal beam's state at the second station, facilitating subsequent compensation processing. By determining the relay compensation domain of the first process, deformation and positioning deviations during transfer can be understood in a timely manner, improving the accuracy of semi-finished longitudinal beam transfer between stations. Generating a dynamic safety domain from the operation data corresponding to the second target station creates real-time constraints, reducing collision risks and improving the collaborative efficiency of the automotive longitudinal beam production line. Next, by determining the process rhythm parameters, we can ensure coordinated movements between each process, optimize production rhythm, reduce waiting time and equipment idle time, and enable the production line to operate in the most efficient way while ensuring safety, thereby improving overall production efficiency. Determining the dynamic pre-positioning point of the robotic arm in the third target station and the clamping compensation value on the second fixture can make the movements of the robotic arm and fixture more coordinated, reduce conflicts and interference between equipment, and improve equipment synergy. By determining the motion reference template of the second fixture, we can reduce fixture wear and extend equipment life. Planning and generating a coordination strategy for the second fixture and the robotic arm in the third target station can make the equipment movements of the entire production line more coordinated, improving production efficiency and product quality.

[0110] Figure 2 This illustration schematically depicts a process diagram for error-proofing verification of a longitudinal beam production line according to an embodiment of this application. For example... Figure 2As shown, in one embodiment of this example, error-proofing verification of the longitudinal beam production line includes:

[0111] S210. Initialize the fixture control module and signal acquisition module in the production line;

[0112] S220. Verify the effectiveness of signal acquisition by the signal acquisition module;

[0113] S230. After the signal acquisition module passes the signal acquisition validity verification, the preset foolproof sample is placed in the target tooling fixture, and it is determined whether the positioning point of the preset foolproof sample is consistent with that of the target tooling fixture.

[0114] S240. When the preset foolproof sample and the target tooling fixture are aligned, control the target tooling fixture to execute the clamping command.

[0115] S250, the start signal acquisition module acquires the clamping status signal and the error-proof identification signal of the target tooling fixture;

[0116] S260. Determine whether the target tooling fixture and the preset error-proofing sample meet the preset error-proofing logic conditions based on the clamping status signal and the error-proofing identification signal.

[0117] S270. After the preset error-proofing logic conditions are met, the target tooling fixture is controlled to execute a reset command, thus completing the error-proofing verification.

[0118] Before the production line starts or the workpiece is in place, preprocessing operations are performed on the fixture control system and signal acquisition system to ensure that each module is in normal working condition and synchronized with the production line control system. The fixture control module is reset, including clearing historical status information, calibrating the initial zero point and joint posture of the fixture, detecting the operating status of actuators and sensors, and establishing a communication connection with the main production line control system. Subsequently, the signal acquisition module is initialized, including activating various sensors and performing self-tests, setting the data sampling frequency and buffer, synchronizing the clock, and confirming normal data transmission. This ensures that the fixture control module and signal acquisition module can stably and efficiently complete the tasks of fixture motion control and workpiece status monitoring during production line operation.

[0119] Verifying the signal acquisition module's effectiveness involves continuous data verification to ensure no packet loss or errors during data transmission; comparing measurements with reference sensor values ​​to ensure results are within reasonable ranges; checking the acquisition frequency and timestamps to guarantee real-time data updates and transmission to the control system; and detecting abnormal signals, including sudden noise changes, drift, or unreasonable fluctuations, issuing alarms or removing abnormal data as needed. This ensures the signal acquisition module provides reliable real-time data support during production line operation, providing accurate data for fixture control and robotic arm motion planning.

[0120] After the signal acquisition module verifies the signal acquisition validity, a pre-set foolproof sample is placed in the target fixture to verify the fixture's positioning accuracy. The foolproof sample, with pre-defined positioning points, is placed in the target fixture; these points can be set according to specific vehicle models or company requirements. Subsequently, the signal acquisition module acquires the position and orientation information of the foolproof sample within the fixture and compares it with the designed positioning points of the fixture to determine if the foolproof sample's positioning points match those of the target fixture. This operation verifies the fixture's clamping accuracy and positioning reliability, ensuring that the workpiece can be correctly clamped in subsequent production processes and preventing workpiece clamping deviations or welding quality problems caused by fixture positioning errors.

[0121] When the error-proof sample is confirmed by the signal acquisition module to have a position and orientation within the fixture that perfectly match the fixture's design positioning point, the control system sends a clamping control signal to the target tooling fixture, causing the fixture to perform a clamping action. The clamping action includes displacement control of the fixture actuator, application of clamping force, and orientation adjustment, fixing the error-proof sample or workpiece in a predetermined position and orientation within the fixture, thereby providing stable working conditions for subsequent machining processes.

[0122] During the clamping action of the target tooling fixture, the control system activates the signal acquisition module to acquire the clamping status of the fixture and the positioning status of the workpiece or mistaken-proof sample in real time. The clamping status signal includes information such as whether the fixture has completed clamping, whether the clamping force has reached the preset value, and whether the fixture actuator has reached the target position; the mistaken-proof identification signal is used to determine whether the workpiece or mistaken-proof sample is correctly positioned to ensure that the workpiece is in a safe and correct position before the fixture performs the clamping action.

[0123] After acquiring the clamping status signal of the target tooling fixture and the positioning identification signal of the mistake-proofing sample in real time, the control system makes a judgment based on the pre-set error-proofing logic rules. The error-proofing logic rules include determining that the fixture and the mistake-proofing sample meet the error-proofing conditions when the mistake-proofing sample is correctly placed, the fixture completes the clamping action, the clamping force reaches the preset value, and the fixture actuator is in the correct position; when any condition is not met, it is determined to be an abnormal state and the corresponding protection or alarm measures are triggered. In this embodiment, the error-proofing logic rules can be as follows: the workpiece or error-proofing sample must be placed at a predetermined positioning point of the fixture, and the position and posture must be confirmed to be correct by a sensor or signal acquisition module; the fixture must complete the clamping action and reach the preset clamping force or clamping stroke to ensure that the fixture actuator is in the target position; the fixture action must be executed in a predetermined sequence, and clamping is prohibited if the workpiece is not in place or the error-proofing identification fails; in the scenario of the robotic arm and fixture operating in coordination, the fixture clamping and the robotic arm movement must be synchronized to avoid action interference; when the workpiece is not in place, the clamping is abnormal, or the fixture action exceeds the safe area, the control system can trigger abnormal handling, alarm, or automatic stop action to ensure the safety of production line operation, etc.

[0124] After confirming that the fixture and workpiece or mistake-proofing sample meet the preset error-proofing conditions by real-time acquisition of clamping status signals and mistake-proofing identification signals, the control system sends a reset control signal to the target tooling fixture, restoring the fixture from the clamping state to the initial or standby state. The reset action includes, but is not limited to, releasing the fixture, returning the actuator to the initial position, and resetting the clamping force or position sensor, thereby completing the mistake-proofing verification operation of the fixture. This reset operation not only verifies the clamping accuracy of the fixture and the correctness of the workpiece positioning but also ensures that the fixture can safely and reliably return to its initial state, providing a guarantee for the smooth execution of production processes.

[0125] By performing error-proofing checks on the longitudinal beam production line, the accuracy of the production process can be effectively improved. This enables precise verification and error-proofing control of the production line, effectively avoiding erroneous operations during production, improving product quality and production efficiency, reducing production costs, and enhancing the stability and reliability of the production line.

[0126] Figure 3 This illustration schematically shows a structural diagram of a longitudinal beam foundation member according to an embodiment of this application. Figure 3 As shown, in one embodiment of this example, the longitudinal beam component includes a first longitudinal beam component, a second longitudinal beam component, a third longitudinal beam component, and a fourth longitudinal beam component. Preprocessing the longitudinal beam components yields the longitudinal beam base component, including:

[0127] S310. The first longitudinal beam component is placed in the first pre-processing station using the first pre-processing fixture and welded to obtain the first spare component.

[0128] S320. The second longitudinal beam component is placed in the second pre-processing station using the second pre-processing fixture and welded to obtain the second spare component.

[0129] S330. The first spare part and the second spare part are placed in the third pre-processing station using the third pre-processing fixture and welded to obtain the first longitudinal beam assembly.

[0130] S340. The third longitudinal beam component is placed in the fourth pre-processing station using the fourth pre-processing fixture and welded to obtain the third spare component.

[0131] S350. The fourth longitudinal beam component is placed in the fifth pre-processing station using the fifth pre-processing fixture and welded to obtain the fourth spare component.

[0132] S360. The third and fourth spare parts are placed in the sixth pre-processing station using the fifth pre-processing fixture and welded to obtain the second longitudinal beam assembly.

[0133] S370, The longitudinal beam base component is obtained through the first longitudinal beam assembly and the second longitudinal beam assembly.

[0134] In this embodiment, the longitudinal beam components include a first longitudinal beam component, a second longitudinal beam component, a third longitudinal beam component, and a fourth longitudinal beam component. These can be different structural sections of the longitudinal beam, reinforcing ribs, support plates, end plates, or auxiliary process components. The first longitudinal beam component to be processed is fixed on the first pre-processing fixture and placed at the first pre-processing station for welding. In this embodiment, the component is manually removed from the component box, loaded onto fixture 1, and the workstation start button is pressed. Welding begins at position H1, and after welding is completed, the first spare component is retrieved. The first spare component consists of different main sections, reinforcing ribs, and other components of the longitudinal beam after pre-processing and welding. The first pre-processing fixture is used to support and position the longitudinal beam component to ensure its correct spatial orientation and structural shape during welding. The welding operation can be completed using arc welding, spot welding, or other suitable welding methods. The first spare component obtained through this operation can be used as a workpiece for production line process verification, fixture calibration, or emergency replacement.

[0135] Similarly, the second longitudinal beam component to be processed is fixed on the second pre-processing fixture and placed at the second pre-processing station for welding. In this embodiment, the component is manually removed from the component box, loaded onto fixture 2, and the workstation start button is pressed. Welding begins at position H2. After welding is completed, a second spare component is retrieved. The second spare component can be a component of different end plates or support plates of the longitudinal beam after pre-processing and welding. The second pre-processing fixture is used to support and position the second longitudinal beam component to ensure its correct spatial orientation and structural shape during the welding process. The welding operation can be completed using arc welding, spot welding, or other suitable welding methods.

[0136] Subsequently, the first and second spare parts obtained above are fixed onto the third pre-processing fixture and placed at the third pre-processing station for welding. In this embodiment, the first and second spare parts are manually picked up and loaded onto fixture 3, the workstation start button is pressed, and welding begins at position H3. After welding is completed, the robot arm picks up the parts and places them into fixture 7. The third pre-processing fixture is used to support and position the first and second spare parts to ensure their correct relative spatial posture and structural position during the welding process; the welding operation can be completed using arc welding, spot welding, or other suitable welding methods. Through this operation, the obtained first longitudinal beam assembly forms a complete longitudinal beam structure, which can be used for formal production and subsequent assembly.

[0137] To pre-process the welding of the third longitudinal beam component, a fourth pre-processing fixture is first used to position, support, and clamp the component, ensuring a stable spatial posture and accurate clamping reference during subsequent welding. Then, the third longitudinal beam component, held by the fourth pre-processing fixture, is transported and placed at the fourth pre-processing station. At this station, the corresponding welding procedure is invoked to perform pre-set weld splicing, structural reinforcement, or local shaping treatments on the third longitudinal beam component. After welding is completed at the fourth pre-processing station, a third spare component that meets the requirements of the next stage of assembly is obtained, providing a stable and reliable foundation component for the subsequent welding of the overall longitudinal beam structure. In this embodiment, the component is manually removed from the component box, loaded onto fixture 4, and the workstation start button is pressed. Welding begins at position H4, and the third spare component is retrieved after welding is completed.

[0138] To achieve pre-treatment welding of the fourth longitudinal beam component, the fifth pre-treatment fixture is first used to precisely position and clamp the component, ensuring its posture stability and consistency of processing benchmarks during subsequent welding operations. Then, the fourth longitudinal beam component, held by the fifth pre-treatment fixture, is transported and placed at the fifth pre-treatment station. At this station, the corresponding welding program is invoked to perform preset weld splicing, structural connection, or local reinforcement processes. In this embodiment, the component is manually removed from the component box, loaded onto fixture 5, and the workstation start button is pressed. Welding begins at H5, and the component is removed for later use after welding. After welding at the fifth pre-treatment station, a structurally complete fourth spare component meeting the requirements of subsequent assembly is obtained, thus realizing step-by-step processing in the longitudinal beam pre-treatment stage and providing a reliable foundation component for the subsequent integrated welding of the longitudinal beam body.

[0139] To achieve the pre-assembly and welding of the second longitudinal beam assembly in the longitudinal beam structure, the third and fourth spare parts, obtained from the previous workstation, are first clamped sequentially in the fifth pre-processing fixture, ensuring precise positioning and relative constraint according to the preset assembly reference surface, weld seam alignment, and geometric relationships. Subsequently, the clamped third and fourth spare parts are placed as a whole in the sixth pre-processing station. At this station, the welding process corresponding to the target assembly is invoked to weld the butt welds, reinforcing structures, and connection points between the two spare parts. Specifically, the third and fourth spare parts are manually retrieved, mounted on fixture 6, and the workstation start button is pressed. Welding begins at position H6. After welding, the robotic arm retrieves the parts and places them in fixture 7. Through the welding process at the sixth pre-processing station, the third and fourth spare parts form a structurally integrated second longitudinal beam assembly, providing a stable and uniformly shaped longitudinal beam structural unit for the subsequent assembly stage.

[0140] The longitudinal beam base component is obtained by assembling the first and second longitudinal beam assemblies. After obtaining these components, the robotic arm grips the part in fixture 7 and moves it to position H8 to begin welding. After welding at position H8 is completed, the robotic arm places the product at position H9. The turntable at position H9 rotates to a welding position and welding begins. After welding is complete, the turntable rotates 180 degrees, the fixture opens, and the part is manually removed and inspected, thus obtaining the longitudinal beam base component.

[0141] By performing step-by-step pretreatment and welding operations on individual longitudinal beam components, efficient production of longitudinal beam foundation components is achieved. Based on the phased pretreatment and welding, welding quality can be effectively guaranteed, production efficiency can be improved, and the structural stability of longitudinal beam foundation components can be ensured.

[0142] In one embodiment of this example, determining the first process relay compensation domain based on the first attitude data and the second attitude data includes:

[0143] S410. Extract the first key point dataset based on the first posture data. The first key point dataset includes the first weld positioning point data, the first fixture positioning center point data, and the first reinforcing rib support point data.

[0144] S420. Extract the second key point dataset based on the second posture data. The second key point dataset includes the second weld positioning point data, the second fixture positioning center point data, and the second reinforcing rib support point data.

[0145] S430. Map the first key point dataset and the second key point dataset to the reference coordinate system, and calculate the deflection curve deviation, torsional attitude deviation and planar positioning deviation based on the first key point dataset and the second key point data.

[0146] S440. Determine the type of deviation by combining the deflection curve deviation, torsional attitude deviation and planar positioning deviation. The types of deviation include clamping positioning inaccuracy and transport deformation deviation.

[0147] S450. When the deviation type is transport deformation deviation, construct a regionalized deformation model of the longitudinal beam by combining deflection curve deviation, torsional attitude deviation and planar positioning deviation.

[0148] S460. Divide the longitudinal beam regional deformation model to determine the weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain.

[0149] S470. Combine the weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain to determine the first process relay compensation domain.

[0150] The first posture data contains multiple key point data of the semi-finished longitudinal beam, including but not limited to weld positioning points, fixture positioning center points, and stiffener support points. The first weld positioning point data, the first fixture positioning center point data, and the first stiffener support point data are extracted from the first posture data. The weld positioning point data characterizes the actual position of the weld in the reference coordinate system of the semi-finished longitudinal beam, facilitating subsequent processes to verify welding accuracy. The fixture positioning center point data characterizes the reference positioning of the semi-finished longitudinal beam in the clamped state, ensuring coordinate consistency during the transfer of the longitudinal beam between workstations. The stiffener support point data characterizes the support position of the stiffeners in the semi-finished longitudinal beam structure, facilitating the assessment of the stress state and deformation trend of the longitudinal beam during transport or clamping.

[0151] The second attitude data contains multiple key point data of the semi-finished longitudinal beam after it is received at the second target station. These key points include, but are not limited to, weld positioning points, fixture positioning center points, and reinforcing rib support points. By parsing the second attitude data, the second weld positioning point data, the second fixture positioning center point data, and the second reinforcing rib support point data can be extracted. Specifically, the weld positioning point data characterizes the actual position of the semi-finished longitudinal beam weld in the reference coordinate system of the second target station, facilitating comparison with the welding accuracy of the first station and deviation verification. The fixture positioning center point data characterizes the reference positioning of the semi-finished longitudinal beam in the second fixture locking state, ensuring the coordinate continuity and stability of the longitudinal beam during inter-station transfer. The reinforcing rib support point data characterizes the support position of the reinforcing ribs in the semi-finished longitudinal beam structure at the second station, facilitating the assessment of the stress state and deformation trend of the longitudinal beam during transfer, clamping, or secondary welding. Through the extraction and mapping of the above second key point dataset, a comparison can be made with the first key point dataset, providing a spatial reference and deviation identification basis for the subsequent construction of the process relay compensation domain.

[0152] By mapping the first and second key point datasets to the same reference coordinate system, position alignment and coordinate normalization can be performed on the longitudinal beam posture data collected from different processes or fixtures. The first key point dataset characterizes the spatial posture of the semi-finished longitudinal beam in the previous process (or the first fixture), while the second key point dataset characterizes the spatial posture of the semi-finished longitudinal beam in the current process (or the second fixture). After mapping the two sets of key point data to the reference coordinate system, based on the differences between corresponding points of the two sets of key points in the same coordinate system, the deflection curve deviation, torsional posture deviation, and planar positioning deviation of the longitudinal beam are calculated respectively: the deflection curve deviation reflects the bending deformation difference of the longitudinal beam along its length; the torsional posture deviation reflects the rotational offset of the longitudinal beam around its length axis; and the planar positioning deviation reflects the overall translational offset of the longitudinal beam in the horizontal plane. Specifically, along the length direction of the longitudinal beam, a point set formed by the weld positioning point and the stiffener support point is selected. Curve fitting is performed on the point set, using either polynomial fitting or spline curve fitting, to obtain the deflection curves for the first and second workstations. The difference between these two curves is then calculated to obtain the deflection curve deviation. Torsional attitude deviation is calculated by obtaining the attitude vectors of the first and second workstations from the first and second key point datasets, and calculating the difference between them. The attitude vectors of the first and second workstations characterize the spatial attitude changes of the semi-finished longitudinal beam between different workstations. Planar positioning deviation is calculated by extracting the two-dimensional coordinates of the fixture positioning center point in the reference coordinate system, calculating the planar position difference between the two coordinates, and thus obtaining the planar positioning deviation.

[0153] Subsequently, the deviation type is determined by combining the deflection curve deviation, torsional posture deviation, and planar positioning deviation. In this embodiment, these include clamping positioning inaccuracy and transport deformation deviation. When the planar positioning deviation of the longitudinal beam exceeds a preset threshold, while the deflection curve deviation and torsional posture deviation are both within the threshold range, the system classifies the deviation type as clamping positioning inaccuracy. This type of deviation is mainly caused by inaccurate positioning of the workstation fixture, and the overall longitudinal beam structure does not deform. When the deflection curve deviation or torsional posture deviation of the longitudinal beam exceeds a preset threshold, the system classifies the deviation type as transport deformation deviation. This type of deviation is mainly caused by bending, torsion, or local deformation of the longitudinal beam during handling or transportation, and the actual shape of the longitudinal beam structure changes. By determining the deviation type, a classification basis can be provided for the optimization of the robotic arm path, thereby realizing intelligent deviation correction of the longitudinal beam in the multi-station welding process.

[0154] When the deviation type is determined to be transport deformation deviation, the system combines the longitudinal beam's deflection curve deviation, torsional attitude deviation, and planar positioning deviation to construct a regionalized deformation model of the longitudinal beam. The longitudinal beam is divided into several local regions along its length and at key structural locations, with each region corresponding to a set of key points on the beam. Key points may include weld positioning points, fixture positioning center points, stiffener support points, end support points, and other geometric feature points characterizing the local structural attitude of the longitudinal beam. Within each region, the three-dimensional position information of the key points is locally fitted using MATLAB software to obtain the local deflection curve, local torsional angle, and local planar displacement, thus forming a regionalized deformation mesh model containing the three-dimensional deformation information of each local region. This regionalized deformation mesh model can quantify the local bending, torsional, and translational deviations of the longitudinal beam in different regions and provides accurate data support for subsequent weld zone compensation, auxiliary support compensation, and positioning constraint compensation.

[0155] The longitudinal beam's regional deformation model is divided to determine the weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain. Local areas along the longitudinal beam's length and at key structural locations are matched with the weld locations, auxiliary support points, and fixture positioning points marked on the design drawings to form compensation subdomains for each functional area. Specifically, the weld zone compensation subdomain compensates for local deflection and torsional deviations in the weld area to ensure welding accuracy; the auxiliary support compensation subdomain compensates for local translational or height deviations of the longitudinal beam's auxiliary support points to ensure the structural stability of the longitudinal beam during processing; and the positioning constraint zone compensation subdomain compensates for the overall planar deviation of the longitudinal beam at key fixture positioning points to ensure clamping accuracy at the workstation.

[0156] The compensation subdomains for the weld zone, auxiliary support, and positioning constraint zones are summarized to determine the first-stage relay compensation domain. Local deviation data from each compensation subdomain are mapped to the overall coordinate system of the longitudinal beam, and the effective range of each subdomain on the longitudinal beam is superimposed and integrated to form an overall compensation range encompassing the weld zone, auxiliary support zone, and positioning constraint zone. During the summarization process, local deviations in each region can be weighted to generate overall longitudinal beam deviation distribution information, and the required compensation type and magnitude for each region can be labeled. Specifically, during the summarization process, different weights can be set according to the functional importance of each subdomain to weight the local deviation data. For example, the weight of the weld zone compensation subdomain can be set to a higher value to prioritize welding accuracy; the weight of the auxiliary support compensation subdomain can be set to a medium value to ensure the structural stability of the longitudinal beam during processing; and the weight of the positioning constraint zone compensation subdomain can be set to a moderate or low value to ensure the overall clamping accuracy of the longitudinal beam. Simultaneously, the superimposed deviation data is fitted using MATLAB software to generate overall longitudinal beam deviation distribution information, and the comprehensive compensation amount for each region is obtained by combining this information with the overall longitudinal beam deviation distribution information. The first-stage relay compensation domain generated by this method can reflect the local deformation of the longitudinal beam in each functional area and provide an overall compensation strategy, thereby achieving precise local compensation of the longitudinal beam during multi-station welding. The first-stage relay compensation domain can serve as a unified compensation basis for subsequent stages, guiding fixture adjustments, robotic arm pre-positioning corrections, and weld processing deviation compensation. This ensures precise local compensation and overall assembly accuracy control of the longitudinal beam during multi-station welding, guaranteeing the continuity and stability of the longitudinal beam's structural accuracy between relay stages.

[0157] Through in-depth analysis of the first and second attitude data, the deviations in each stage of the longitudinal beam production process were accurately identified and the compensation strategy was optimized, thereby improving the precision and quality of longitudinal beam production, reducing the defect rate caused by deviations, and improving production efficiency and product consistency.

[0158] In one embodiment of this invention, the longitudinal beam regional deformation model is divided into weld zone compensation subdomains, auxiliary support compensation subdomains, and positioning constraint zone compensation subdomains, including:

[0159] S510. The longitudinal beam skeleton curve is obtained by fitting the longitudinal beam regional deformation model.

[0160] S520. Discretize the longitudinal beam skeleton curve to obtain the discrete points of the curve, and calculate the curvature value of each discrete point of the curve to obtain the skeleton curvature distribution of the semi-finished longitudinal beam.

[0161] S530. The rate of change of curvature between two adjacent discrete points of a curve is calculated by the skeleton curvature distribution.

[0162] S540. The region in the longitudinal beam regional deformation model whose curvature change rate is greater than or equal to the preset first threshold is taken as the weld zone compensation subdomain.

[0163] S550. The region in the longitudinal beam regional deformation model whose rate of curvature change is greater than or equal to the preset second threshold and less than the preset first threshold is taken as the auxiliary support compensation sub-domain.

[0164] S560. The region in the longitudinal beam regional deformation model with a curvature change rate less than the preset third threshold is taken as the location constraint area compensation subdomain.

[0165] Based on the regional deformation model of the longitudinal beam, a baseline curve for the longitudinal beam that can be used for process relay compensation calculation is obtained. To obtain a skeleton curve that continuously reflects the overall deformation trend of the longitudinal beam, this embodiment uses a three-dimensional spline fitting method to construct curves for discrete key points in the regional deformation model. Specifically, firstly, the key points of each local region in the regional deformation model are parametrically sorted according to the longitudinal beam length direction, and the spatial deviation of each key point is converted into a three-dimensional coordinate point set based on the longitudinal beam design baseline, resulting in a discrete sampling point sequence continuously distributed along the longitudinal beam length direction. Secondly, the parametric variables of the sampling point sequence are calculated; for example, a normalized length parameter is assigned to each sampling point using chord length parametric method to ensure that the fitted curve has higher descriptive accuracy in regions with large deformation gradients. Subsequently, based on the three-dimensional point coordinates and their corresponding parametric variables, three-dimensional spline interpolation equations in the x, y, and z dimensions are constructed respectively, and spline interpolation is performed on the discrete point sequence. During the interpolation process, a second derivative continuity constraint can be introduced to ensure that the obtained longitudinal beam skeleton curve maintains overall smoothness. Finally, by combining the interpolation results of the above three dimensions, a three-dimensional spline skeleton curve is obtained to characterize the overall shape and deformation trend of the longitudinal beam. Based on this, the longitudinal beam skeleton curve is obtained through a three-dimensional spline curve fitting method. The longitudinal beam skeleton curve obtained can be used to describe the overall bending, torsion, and local offset characteristics of the longitudinal beam during production, transportation, and welding, providing a unified global morphological benchmark for subsequent determination of the relay compensation domain, adjustment of the welding trajectory, and control of fixture compensation.

[0166] To obtain the curvature distribution of the semi-finished longitudinal beam in space, based on the longitudinal beam skeleton curve obtained by the aforementioned three-dimensional spline fitting, discretization sampling is performed on the skeleton curve. This can be achieved by segmenting the longitudinal beam skeleton curve along its length direction based on an equal-interval sampling rule for the curve parameters, thus obtaining several curve discrete points uniformly distributed in space. The equal-interval sampling rule can include, but is not limited to, a fixed-step parameter sampling rule, which involves uniformly dividing the parameter interval according to a set step size and calculating the coordinates of the curve point at each sampling parameter. Subsequently, for each curve discrete point, the first-order derivative vector and the second-order derivative vector of that point on the spline curve are calculated, and the curvature value corresponding to each discrete point is calculated according to the spatial curve curvature calculation formula. The specific expression for the spatial curve curvature is as follows:

[0167]

[0168] Where, k represents the spatial curvature value calculated at the discrete point of the curve, used to characterize the local bending degree of the semi-finished longitudinal beam; r(t) represents the parametric expression of the longitudinal beam skeleton curve obtained by fitting with three-dimensional splines, and t is the parameter variable of the curve; r′(t) represents the first derivative vector of the skeleton curve at parameter t, used to characterize the tangential direction and rate of change at that point; r′′(t) represents the second derivative vector of the skeleton curve at parameter t, used to characterize the bending direction and acceleration change of the curve;

[0169] Through the above steps, a skeleton curvature distribution sequence along the length of the longitudinal beam can be constructed to characterize the local bending trend and overall deformation characteristics of the semi-finished longitudinal beam, which can provide accurate curvature quantification basis for deviation type identification.

[0170] After obtaining the discrete point sequence of the semi-finished longitudinal beam skeleton curve and its corresponding curvature values, the curvature changes of adjacent discrete points can be quantitatively analyzed to delineate the structurally sensitive and straight regions of the longitudinal beam. Specifically, the discrete points of the curve are numbered according to the parameter order, and the corresponding curvature values ​​k are obtained. i and its arc length position s on the curve i Then, for two adjacent points Pi and Pi+1, calculate their rate of change of curvature Ci, as shown in the following formula:

[0171]

[0172] Where, k i+1 This indicates that at the discrete point P i Adjacent discrete points P i+1 The curvature value calculated at k; i This indicates that at the discrete point P i The curvature value calculated at that location is used to characterize the degree of local bending of the longitudinal beam skeleton curve at that position; si Represents discrete point P i The cumulative arc length position on the longitudinal beam skeleton curve; s i+1 Represents discrete point P i+1 The cumulative arc length position on the longitudinal beam skeleton curve; Ci represents the rate of curvature change. The larger the value, the steeper the deformation gradient in this area, which is a structurally sensitive area; the smaller the value, the more flat this area tends to be.

[0173] Subsequently, the system calculates the rate of curvature change between discrete points of the longitudinal beam. When the rate of curvature change of a certain continuous segment reaches or exceeds a preset first threshold, it indicates that the area is prone to significant bending or warping changes during production transfer, welding heat-affected zone, or assembly positioning. Since such areas typically correspond to structural connection points or areas of sudden stress change near the longitudinal beam weld, their geometric stability is most sensitive to welding quality and overall vehicle assembly accuracy. Therefore, these areas are directly defined as weld area compensation subdomains. By dividing these areas, higher-weighted deviation compensation can be applied to the weld area in the subsequent relay compensation logic, ensuring weld formation quality and longitudinal beam assembly accuracy, and improving the process stability of the production line. The preset first threshold is used to determine whether the rate of curvature change of the longitudinal beam reaches a significant deformation level that requires inclusion in the weld area compensation subdomain. By statistically analyzing the skeleton curvature distribution data collected during the welding, transfer, and clamping processes of historical batches of longitudinal beams, typical numerical ranges of the rate of curvature change in the weld area, stiffener connection area, and stable section are extracted, and the lower quartile of the rate of curvature change in the weld area is used as the preset first threshold.

[0174] To further refine the compensation strategy of the longitudinal beam regional deformation model, local areas of the longitudinal beam with curvature change rates between a preset first threshold and a preset second threshold are designated as auxiliary support compensation subdomains. For continuous segments with curvature change rates greater than or equal to the preset second threshold and less than the preset first threshold, their deformation gradient is considered moderate; they are neither highly sensitive areas of the weld zone nor completely straight stable segments. These areas are typically located at longitudinal beam stiffener connections, secondary stress points, or structural transition points. Their geometric stability has a certain impact on welding quality and assembly accuracy, but is less sensitive than critical areas of the weld zone. The purpose of designating these areas as auxiliary support compensation subdomains is to apply appropriate compensation or support to these areas during subsequent relay compensation and fixture control. This reduces the cumulative deviation of the overall longitudinal beam posture caused by moderate deformation areas, while avoiding over-compensation that leads to welding stress concentration or redundant fixture movements, thereby improving the stability of the production process. The preset second threshold is used to identify areas where the longitudinal beam curvature change rate is moderately sensitive, i.e., the auxiliary support compensation subdomains. Statistical analysis was conducted on the skeleton curvature distribution data collected during the welding, transportation, and clamping processes of the longitudinal beams to extract the curvature change rate distribution in the highly sensitive area of ​​the weld zone and the curvature change rate distribution in the stable area of ​​the straight zone. Subsequently, based on the distribution characteristics, a typical curvature change rate interval below the weld zone threshold but above the stable zone threshold was selected as the preliminary second threshold range.

[0175] When the rate of curvature change of a continuous segment is lower than a preset third threshold, it indicates that the region remains almost straight with minimal curvature and relatively stable geometry during production, assembly, and welding. This type of region is typically located in the non-weld connection area of ​​the longitudinal beam, in a stable section far from structural abrupt changes. Its main function is to serve as a reference or constraint for longitudinal beam assembly positioning. In subsequent relay compensation processes, dividing this region into a positioning constraint compensation sub-domain aims to provide stable reference constraints for the weld area and auxiliary support area, ensuring the overall stability of the longitudinal beam's posture. Simultaneously, it avoids applying unnecessary compensation actions to the straight and stable region, thereby improving the overall accuracy and reliability of the production line process and realizing a hierarchical strategy for regionalized longitudinal beam compensation. The preset third threshold is used to identify stable regions with a small rate of curvature change in the longitudinal beam, i.e., the positioning constraint compensation sub-domain. The distribution of the rate of curvature change in each segment of the longitudinal beam during welding, transfer, and clamping can be statistically analyzed, and the average value of the curvature change rate characteristics in the straight area or non-stress-sensitive area can be extracted. Subsequently, the average curvature change rate is used as the preset third threshold.

[0176] In this embodiment, a first threshold, a second threshold, and a third threshold constitute a hierarchical classification system for the longitudinal beam curvature change rate, wherein the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold. This means that: the first threshold corresponds to the region with the highest curvature change rate, typically the weld seam compensation sub-domain, indicating that the longitudinal beam is prone to significant bending or warping in this region, which is a highly sensitive area; the second threshold corresponds to the region with a moderate curvature change rate, i.e., the auxiliary support compensation sub-domain, indicating that the longitudinal beam bends relatively significantly in this region, but is less sensitive than the weld seam area; the third threshold corresponds to the region with the lowest curvature change rate, i.e., the positioning constraint compensation sub-domain, indicating that the longitudinal beam is nearly straight in this region, with high structural stability. By setting the relative values ​​of these three thresholds, a hierarchical classification of the longitudinal beam's regional deformation can be achieved, providing a layered and scientific compensation strategy for process relay compensation. This ensures that the weld seam area receives focused correction, the auxiliary support area receives appropriate support, and the positioning constraint area maintains baseline stability, achieving high precision and reliability in the overall production process.

[0177] By intelligently dividing the longitudinal beam's regional deformation model based on the skeleton curvature, the precise definition and differentiated processing of compensation subdomains were achieved. This significantly improved the targeting of the compensation strategy, effectively suppressing welding deformation, optimizing support layout, and enhancing positioning accuracy. Consequently, while ensuring the structural integrity of the longitudinal beam, assembly efficiency was increased and the rework rate was reduced.

[0178] In one embodiment of this invention, a dynamic safety domain is generated based on the operation data corresponding to the second target workstation, including:

[0179] S610, Obtain the workstation geometry layout of the second target workstation;

[0180] S620. Based on the operation data corresponding to the second target workstation, obtain the pose parameters and motion parameters of the robotic arm in the second target workstation, as well as the motion range of the second fixture and the movement position information of the operator.

[0181] S630: Construct an initial virtual fence boundary based on the workstation geometric layout. The initial virtual fence boundary is divided into a core danger zone, a buffer warning zone, and a safe work zone.

[0182] S640. The initial virtual fence boundary is extended according to the operator's movement location information to obtain the range of the first virtual fence, which is located outside the buffer warning zone.

[0183] S650. Using the first virtual fence range as a constraint reference, and combining the pose parameters and motion parameters of the robotic arm, the initial virtual fence boundary is extended to obtain the second virtual fence range. The extended boundary of the second virtual fence range does not break through the boundary of the first virtual fence range.

[0184] S660. Using the first and second virtual fence ranges as dual constraints, the initial virtual fence boundary is locally thickened according to the movement range of the second fixture to obtain the third virtual fence range.

[0185] S670. A dynamic security domain is generated by combining the first virtual fence range, the second virtual fence range, and the third virtual fence range.

[0186] It provides a precise spatial reference basis for the subsequent construction of dynamic safety domains and collaborative planning of processes. The data acquisition module obtains the workstation geometry layout of the second target workstation. The workstation geometry layout includes, but is not limited to, the actual position and size of various equipment, fixtures, robotic arms and auxiliary facilities in space.

[0187] Based on the operational data corresponding to the second target workstation, the pose and motion parameters of the robotic arm in the second target workstation, as well as the motion range of the second fixture and the operator's movement position information, can be obtained. Specifically, the robotic arm's pose parameters include three-dimensional spatial information such as the position, attitude angle, and joint angles of the end effector; the robotic arm's motion parameters include velocity, acceleration, jerk, and dynamic trajectory characteristics during the execution of the process; the motion range of the second fixture includes the spatial range in which the fixture can move, the clamping and releasing strokes, and related constraints; the operator's movement position information includes their real-time position within the workstation, their range of motion, and the areas where they may interact with the robotic arm and fixture. The purpose of obtaining the above information is to provide accurate spatial and motion constraint data for constructing a dynamic safety domain, robotic arm motion planning, and fixture motion control, so as to predict potential collision risks, operational interference, and safety constraints within the workstation, ensuring the safety, coordination, and high precision of the production process.

[0188] An initial virtual fence boundary is constructed based on the workstation geometry layout of the second target workstation. According to the workstation geometry layout, the workstation is divided into zones according to spatial safety levels. The maximum range of motion of the robotic arm and the range of motion of the gripper are defined as the core hazard zone, a buffer zone a certain distance outside the core hazard zone is defined as the buffer warning zone, and the remaining areas are defined as safe working zones. Subsequently, the boundary point coordinates of each zone are spatially modeled, and a three-dimensional spatial boundary model is generated using meshing or voxelization. Boundary constraint algorithms are then used to stitch the boundaries of each zone together to form the initial virtual fence boundary. Specifically, the distance between boundary points of adjacent zones is calculated to identify overlapping or intersecting parts, and gaps and conflicts between boundaries are eliminated using interpolation or weighted averaging methods. Furthermore, meshing or voxel modeling is used to uniformly map the boundaries of each zone to the same spatial coordinate system, and the boundaries of the core hazard zone, buffer warning zone, and safe working zone are layered and superimposed using boundary stitching rules to form a complete virtual fence structure. The final virtual fence boundary accurately reflects the safety zoning of the workstation in three-dimensional space, providing a unified geometric benchmark for subsequent dynamic expansion, real-time monitoring, and scheduling control. The core hazard zone corresponds to the space where the robotic arm, fixtures, and workpieces may pose a high risk during the execution of processes. When operators or other equipment enter this area, there is a high risk of collision or injury. The buffer warning zone is located outside the core hazard zone and is used to provide early warning of operations that may enter the core hazard zone, providing a response buffer for the system or operators. The safe working zone is located outside the buffer warning zone and provides a safe working space for operators and auxiliary equipment, which is basically unaffected by the movement of the robotic arm and fixtures.

[0189] Subsequently, the system acquires the real-time position and movement trajectory of operators within the workstation, and appropriately expands the safety boundaries of the core danger zone and buffer warning zone of the initial fence to form a first virtual fence range. This range is located outside the buffer warning zone, meaning it does not affect the normal operating area of ​​the robotic arm and grippers. Specifically, positioning sensors are used to collect the position coordinates and movement trajectory information of operators within the workstation. Next, the position information is matched and analyzed with the spatial boundaries of the initial virtual fence to identify the spatial relationship between the operator's activity range and the fence boundary. Then, the safety boundaries of the core danger zone and buffer warning zone of the initial fence are appropriately expanded. The expansion process includes: determining the operator's maximum activity range through their movement trajectory, and adding a safety margin along the longitudinal and lateral directions outside this range. The safety margin can be set according to the company's production requirements. The expanded boundary is then spliced ​​and merged with the original fence boundary to form a continuous and complete first virtual fence range. The first virtual fence range generated in this way always remains outside the buffer warning zone, does not affect the normal operating area of ​​the robotic arm and grippers, and dynamically reflects the actual activity space of the operators, providing boundary constraints and safety benchmarks for the subsequent construction of the second virtual fence. By considering the potential activity range of operators, additional spatial buffers are provided for workstation safety management, preventing personnel from accidentally entering high-risk areas while ensuring that the operation of robotic arms and fixtures is not restricted. The first virtual fence range serves as the primary constraint boundary of the dynamic safety domain. In subsequent processes, it can be combined with the robotic arm's range of motion, the fixture's range of motion, and process rhythm parameters to guide robotic arm trajectory planning and fixture motion control. This achieves a balance between personnel safety and equipment collaboration within the workstation, improving the safety and reliability of the production process.

[0190] Using the first virtual fence range as a spatial constraint benchmark, and while maintaining the integrity of the personnel safety activity area, the initial virtual fence boundary is dynamically expanded to form a second virtual fence range. Based on the robot arm's pose parameters in the second target workstation, the three-dimensional spatial operation trajectory of the robot arm is obtained, and combined with the robot arm's motion parameters, the maximum dynamic workspace that the robot arm may reach in each movement stage is determined. Subsequently, using the robot arm's operation trajectory and its dynamic work space as the extension basis, the safety domain corresponding to the initial virtual fence boundary is expanded to obtain the boundary area to be extended for the second virtual fence range. Furthermore, the system uses the outer boundary of the first virtual fence range as an insurmountable constraint condition to spatially prune the boundary area to be extended, ensuring that the final extension boundary of the second virtual fence range is limited to the interior of the first virtual fence range in any direction and must not exceed its outer boundary. Through the above processing, the second virtual fence range fully represents the dynamic operation range of the robot arm while preventing it from encroaching on the personnel safety domain, thereby ensuring the continuous stability of the spatial isolation relationship between the robot arm's dynamic safety domain and the personnel safety domain, and improving the reliability of the overall dynamic safety domain planning. The first virtual fence is used to limit the safe activity area of ​​the operator, and the second virtual fence is used to limit the maximum movement space of the robotic arm.

[0191] Subsequently, the system first obtains the motion range parameters of the second fixture in the current process, including the fixture's maximum opening stroke, maximum closing stroke, flipping angle, fine-tuning displacement, and dynamic offset that may occur during the motion. Based on the above motion range parameters, the system constructs the maximum motion envelope model of the fixture. A parametric simulation method based on kinematic chains can be used to generate a set of achievable poses in the workstation coordinate system according to the joint constraints, link dimensions, and pose constraints of the fixture and actuator, and then mesh and merge them to form the envelope. Next, the maximum motion envelope of the fixture and the initial virtual fence boundary are unified to the workstation coordinate system to complete the registration and timestamp synchronization. Then, the minimum distance is calculated by subtracting the envelope from the fence boundary. When the minimum distance is less than a preset distance, the corresponding fence boundary segment can be marked as a candidate for thickening. The preset distance can be set based on standard safety requirements. Next, Boolean intersection and penetration depth analysis is performed on the candidate areas to be thickened to identify the intersection, proximity, and potential conflict areas between the envelope and the fence boundary. Based on the consistency of local curvature and normal, the direction and magnitude of boundary thickening are evaluated, thus forming a spatially continuous local boundary region and obtaining the target locally thickened region, thereby creating a thickened boundary with redundant thickness. Using the first and second virtual fence ranges as constraints, without exceeding their boundaries, the virtual 3D spatial model is updated based on the thickened boundary region to obtain a third virtual fence range. This range completely covers all possible reachable areas of the fixture under maximum range of motion, while maintaining non-overlap with the personnel safety zone and the robotic arm's movement zone, achieving enhanced safety isolation under multi-subject collaborative operation conditions.

[0192] Finally, the first, second, and third virtual fence ranges are combined to generate a dynamic safety domain. Specifically, the first virtual fence range is formed by extending the initial virtual fence boundary based on the operator's movement position information, reflecting the operator's real-time activity space. The second virtual fence range, constrained by the first virtual fence, dynamically expands the boundary based on the robotic arm's pose parameters and motion amplitude, reflecting the equipment's operating space. The third virtual fence range, based on the gripper's motion range, motion amplitude, and dynamic offset during clamping / releasing, locally thickens the corresponding area of ​​the initial virtual fence boundary to provide redundant width and safety margin. By spatially fusing and stitching the boundaries of these three types of virtual fence ranges, the system can form a dynamic overall safety domain that changes dynamically with the operator's position, equipment movement, and gripper operation. This dynamic safety domain not only reflects the comprehensive risk distribution within the workstation in real time but also improves workstation space utilization and human-machine collaboration efficiency while ensuring safety.

[0193] By comprehensively considering the geometric layout of the second target workstation, the range of motion of the robotic arm and fixture, and the movement position information of the operator, a highly safe safety domain is dynamically generated. This domain can adapt to various dynamic changes within the workstation in real time, effectively avoiding human-machine collisions and equipment damage, ensuring the safety of operators and equipment, and improving production efficiency and operational flexibility.

[0194] In one embodiment of this invention, using the first virtual fence range as a constraint reference, and combining the pose and motion parameters of the robotic arm, the initial virtual fence boundary is extended to obtain a second virtual fence range, including:

[0195] S710. Determine the spatial operation trajectory of the robotic arm through its pose parameters.

[0196] S720. Using the first virtual fence range as a constraint benchmark, and combining it with the spatial operation trajectory of the robotic arm, determine the extended boundary area of ​​the second virtual fence range.

[0197] S730: Determine the dynamic expansion rate of the extended boundary region based on the motion parameters of the robotic arm;

[0198] S740. Determine the range of the second virtual fence by combining the dynamic expansion rate of the extended boundary area, wherein the extended boundary does not exceed the range of the first virtual fence.

[0199] In this embodiment, under the constraint of the first virtual fence range, the spatial operation trajectory of the robotic arm undergoes constrained morphological expansion processing. Specifically, the spatial operation trajectory of the robotic arm is determined based on the pose parameters of the robotic arm end effector at each key workstation. The pose parameters include the position coordinates and orientation information of the robotic arm end effector in three-dimensional space, which describe the specific position and direction of the robotic arm end effector in space. The spatial operation trajectory of the robotic arm can be generated using the position coordinates and orientation information of the robotic arm end effector in three-dimensional space through a spatial interpolation algorithm. The spatial operation trajectory can accurately reflect the motion state of the robotic arm end effector over time and is used to drive the robotic arm to complete workpiece gripping, handling, assembly, or other operation tasks according to a predetermined path, thereby achieving high-precision positioning and reliable operation of the robotic arm.

[0200] Using the first virtual fence area as a constraint benchmark, and combining it with the spatial operation trajectory of the robotic arm, the extended boundary area of ​​the second virtual fence area is determined. Specifically, the first virtual fence area is formed by extending the operator's movement position information, used to limit the safe activity space of personnel, and is the basic safety boundary of the power equipment. Under this constraint, key points are screened through the spatial operation trajectory of the robotic arm. Key points can be the initial standby position of the robotic arm before operation, key insulation parts of the power equipment (insulator string skirts, switch cabinet busbar joints, cable terminations, surge arrester surfaces, etc.), the intermediate positions passed by the robotic arm to avoid obstacles in the power equipment (such as switch cabinet doors, busbar supports), etc., which can be screened and set according to the specific situation in implementation. These key points are then formed into a set of key point coordinates. Subsequently, the three-dimensional bounding box of all key points in the robotic arm trajectory is calculated using forward kinematics and MATLAB software. This involves processing all key trajectory points in the spatial operation trajectory using MATLAB software to extract the position coordinates of each point in three-dimensional space. Based on these position coordinates, the maximum and minimum values ​​of each trajectory point along the three-dimensional coordinate axes are determined, thus constructing a three-dimensional bounding box that covers all key trajectory points. The three-dimensional bounding box is used to characterize the minimum spatial range that the robotic arm's end effector may involve within the prediction time window. Next, geometric intersection operations are performed based on the first virtual fence and the 3D bounding box to obtain candidate regions located inside the first virtual fence and covering key points of the robotic arm trajectory. Then, the boundary cells of each candidate region are analyzed one by one. The minimum spatial distance between each boundary cell and the key points of the trajectory is calculated using Euclidean distance, and boundary cells with distances less than a preset trajectory influence threshold are marked as extendable boundary cells. The set of all marked boundary cells constitutes the extended boundary region of the second virtual fence, while unmarked boundary cells retain their original shape and do not participate in the expansion. The preset trajectory influence threshold characterizes the maximum spatial distance by which the robotic arm's spatial operation trajectory affects the surrounding area. Specifically, it is obtained by comprehensively analyzing the robotic arm's own characteristics, the operating object, the working environment, task requirements, and relevant safety standards, combined with empirical judgment, risk assessment, simulation, and experimental verification. Through the above processing, it is ensured that the extended boundary region covers the space that the robotic arm trajectory may occupy without exceeding the range of the first virtual fence, thereby achieving safe adaptive adjustment of the second virtual fence and providing a basic geometric basis for subsequent dynamic expansion and safety domain updates.

[0201] Subsequently, the control system determines the dynamic expansion rate of the extension boundary region based on the real-time motion parameters of the robotic arm. By analyzing these motion parameters, the control system calculates the expansion rate of the extension boundary region. Specifically, when the robotic arm's speed is high, the extension boundary thickens at a high expansion rate. The expansion rate is positively correlated with the robotic arm's movement speed, being twice the speed. The specific multiplier can be set by the company to ensure that the robotic arm does not exceed the safety boundary during high-speed movement. When the robotic arm's speed is low or it is stationary, the expansion rate of the extension boundary remains constant. The dynamic expansion rate updates the extension boundary in real time through periodic or continuous control, enabling the extension boundary to adaptively thicken according to the robotic arm's movement state, forming a reliable dynamic safety protection zone.

[0202] Finally, based on the real-time motion parameters of the robotic arm and the dynamic expansion rate of the extended boundary region, the control system thickens the extended boundary to generate a second virtual fence range. Throughout the extension process, the system consistently uses the first virtual fence range as a constraint benchmark to limit the extended boundary, ensuring it does not exceed the first virtual fence range and thus guaranteeing that the personnel safety zone remains unaffected. Through this process, the actual operating space of the robotic arm can be dynamically reflected while ensuring personnel safety, enabling real-time adjustment of the virtual fence boundary and safeguarding both production and personal safety.

[0203] The second virtual fence range is determined by analyzing the pose and motion parameters of the robotic arm, and its extension boundary area is also defined. Then, a dynamic expansion rate is calculated based on the robotic arm's motion parameters, allowing the fence boundary to be adjusted in real-time according to the robotic arm's actual speed and acceleration. Finally, the second virtual fence range is extended by combining the dynamic expansion rate of the extension boundary area, while strictly ensuring that the extension boundary does not exceed the first virtual fence range, thus ensuring the integrity and reliability of the safety domain. This not only improves the dynamic adaptability of the safety domain but also effectively balances production efficiency and operational safety, providing more precise safety assurance for automated production under complex working conditions and reducing safety hazards in human-machine interaction.

[0204] In one embodiment of this invention, determining the dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture based on process rhythm parameters includes:

[0205] S810: Obtain the current pose data of the robotic arm in the third target station;

[0206] S820. Using the process rhythm parameters, a pre-trained timing prediction model is used to predict the prediction time window for the robotic arm to receive the semi-finished longitudinal beam in the third target station and the docking time window for the second fixture.

[0207] S830: Generate a dynamic pre-positioning domain based on the prediction time window and the current pose data of the robotic arm;

[0208] S840. Within the dynamic pre-positioning domain, calculate the optimal pre-positioning point sequence of the robotic arm by combining the process rhythm parameters;

[0209] S850: Calculate the shortest movement time for the robotic arm to reach each optimal pre-position sequence;

[0210] S860: Generates the time-matching pre-positioning trajectory of the robotic arm based on the optimal pre-positioning point sequence and the predicted time window;

[0211] S870: Determine the reachable pre-deployment point by combining the shortest motion time with the time-matched pre-deployment trajectory;

[0212] S880. Select the reachable pre-departure point with the smallest time difference between the shortest motion time and the predicted time window as the dynamic pre-departure point.

[0213] S890. Within the docking time window of the second fixture, input the process rhythm parameters into the clamping compensation regression model to obtain the clamping compensation value on the second fixture.

[0214] First, the current pose data of the robotic arm at the third target workstation is acquired to describe the real-time position and attitude of the end effector in space. The pose data includes the three-dimensional spatial coordinates and attitude information of the robotic arm's end effector, comprehensively reflecting its actual state at that workstation. The control system calculates the spatial pose of the end effector by reading the joint angles, link lengths, and kinematic model of the robotic arm. The acquired pose data can be used for trajectory planning, dynamic safety domain generation, and collision detection of the robotic arm, ensuring that it accurately reaches the target workstation and maintains the correct attitude, while also providing real-time reference for path adjustment and safety control.

[0215] Secondly, a pre-trained time-series prediction model is used to predict the predicted time window for the robotic arm to receive the semi-finished longitudinal beam and the docking time window for the second fixture at the third target station. In this embodiment of the invention, the predicted time window for the semi-finished longitudinal beam and the docking time window for the second fixture are used to establish the process rhythm matching relationship and clamping compensation control logic. Process rhythm parameters are used to describe the time and production pattern characteristics between various processes in the production line, including but not limited to the transmission time of the semi-finished product from the previous station to the third target station, processing cycle, and waiting time. The pre-trained time-series prediction model is trained using historical production data and can predict the time sequence characteristics of the robotic arm and fixture movements, thereby outputting the time interval for the robotic arm to receive the semi-finished product and the time interval for the second fixture to dock with the semi-finished product. By arranging the movements of the robotic arm and fixture in advance according to the predicted time windows, the movements of the two at the third target station are coordinated, ensuring that the robotic arm can receive the semi-finished longitudinal beam on time and that the fixture completes the docking operation within the predetermined time window, thereby improving the synchronization, efficiency, and reliability of the production line operation. In this embodiment, the pre-trained time-series prediction model can be a Long Short-Term Memory (LSTM) network model. The pre-trained model is trained using historical production data. Specifically, historical production line data is collected as input features, including the completion time of the robotic arm's actions at the previous workstation, the transfer time of semi-finished products from the previous workstation to the current workstation, the start and end times of fixture docking at the current workstation, the average time taken for the robotic arm to receive semi-finished products in the most recent N times, the operating speed of the robotic arm at the current workstation, the load status or load level of the robotic arm at the current workstation, the historical production cycle time of the current workstation, and the cycle time deviation between adjacent workstations. The data is preprocessed, including normalization, standardization, and time-series feature construction. Features may include the completion time of the previous workstation, the interval time at the current workstation, the historical average receiving time of the robotic arm, the historical docking time window of the fixture, and the production cycle time. Subsequently, the training data, after timestamp alignment, normalization, and time series construction, is divided into training, validation, and test sets, and input into the selected time series prediction model for training. The model can be a Long Short-Term Memory (LSTM) network. During training, a loss function (such as mean squared error or mean absolute error) is used to measure the deviation between the predicted time window and the actual time, and backpropagation and optimization algorithms (such as Adam or SGD) are used to iteratively update the model parameters to minimize the prediction error. After validation set evaluation and hyperparameter tuning, the output of the time series prediction model is the time prediction result, which represents the predicted time window for the robotic arm to receive semi-finished products or for the fixture to complete docking at the target workstation. The trained model can accurately predict the time window for the robotic arm to receive semi-finished products and for the fixture to dock based on real-time process rhythm parameters, achieving motion coordination and production rhythm optimization.

[0216] In this embodiment, a dynamic pre-alignment domain is generated based on the prediction time window and the current pose data of the robotic arm, specifically including the following steps:

[0217] Construct a global coordinate system for the workstation and convert the current pose data of the robotic arm into the global coordinate system for the workstation;

[0218] Within the prediction time window, the robot arm's motion trajectory is determined using the robot arm's current pose data, and the robot arm's motion trajectory is discretized into a pose sequence.

[0219] The spatial sweep volume of the robotic arm within the prediction time window is calculated based on the pose sequence, and the spatial sweep volume is used as the boundary of the pre-access domain.

[0220] The dynamic pre-fitting domain is obtained by correcting the spatial boundary of the pre-fitting domain using process rhythm parameters.

[0221] First, a global coordinate system is constructed at the third target workstation. The origin is the workstation reference point, which is a fixed mechanical positioning pin or optical calibration mark. The X, Y, and Z axis directions and reference posture in three-dimensional space are defined to obtain the global coordinate system. The current pose data of the robotic arm includes the three-dimensional position and posture information of the end effector. This information is transformed into the global coordinate system using the ROS tf2 library, a core library for managing and processing coordinate transformations.

[0222] Secondly, within the prediction time window, the robot arm's current pose data is interpolated using joint space to determine its motion trajectory. For example, using inverse kinematics, the current end effector pose is converted into the corresponding current joint angle, and then the target end effector pose is also converted into the corresponding target joint angle. Interpolation is then performed in joint space (a vector space composed of all joint angles) to obtain the robot arm's motion trajectory. Next, a series of discrete points on the robot arm's trajectory are extracted according to the system-defined time intervals. For each sampled point, the system calculates its corresponding three-dimensional position coordinates (x, y, z) and attitude information (Euler angles or quaternions) to form complete pose data. These discrete pose points are arranged in chronological or path order to form an ordered pose sequence, which is used to describe the robot arm's motion process within the prediction time window.

[0223] The system calculates the spatial sweep volume of the robotic arm within the prediction time window based on the pose sequence, and uses the spatial sweep volume as the spatial boundary of the pre-attachment domain. Specifically, firstly, the pose sequence of the robotic arm within the prediction time window is obtained. This sequence is discretized from a continuous motion trajectory and contains multiple ordered three-dimensional positions and attitude points. Then, based on the geometric models of each link and end effector of the robotic arm, a corresponding spatial instance is generated at each attitude point. All instances are spatially fused using Boolean union operations to form the overall spatial envelope that the robotic arm may occupy within the time window, i.e., the spatial sweep volume. Further, the boundary of the spatial sweep volume is extracted to obtain a complete spatial boundary model, and this boundary is used as the spatial boundary of the pre-attachment domain to limit the spatial range that the robotic arm may reach in future actions.

[0224] The pre-positioning domain spatial boundary is corrected using process rhythm parameters to obtain a dynamic pre-positioning domain. Specifically, firstly, the spatial sweep volume of the robotic arm is calculated based on its pose sequence within the prediction time window, and the outer boundary of this sweep volume is used as the pre-positioning domain spatial boundary. Then, process rhythm parameters are acquired, including the workstation processing cycle time, action sequence, process duration, and equipment operating frequency, to characterize the movement patterns and rhythmic features of the robotic arm at different process stages. Further, the system performs correlation analysis between the process rhythm parameters and the pre-positioning domain spatial boundary, dynamically correcting the boundary according to the speed of the process rhythm. For example, the boundary margin is appropriately expanded during high-frequency action stages, and the boundary range is contracted during low-frequency or stationary stages, thus forming a dynamic pre-positioning domain that changes in real time with the process rhythm. Through this correction process, the pre-positioning domain spatial boundary not only reflects the geometric range of the robotic arm but can also be dynamically adjusted in conjunction with the temporal characteristics of the process rhythm, thereby achieving accurate prediction of the robotic arm's operating space.

[0225] After generating a dynamic pre-positioning domain to constrain the robotic arm's approach to the semi-finished longitudinal beam, the control system further calculates the optimal pre-positioning point sequence for the robotic arm based on process rhythm parameters. The rhythm parameters corresponding to the current process are obtained, including but not limited to: the predicted completion time of the preceding workstation, the relay start time of the third target workstation, the motion time model of the robotic arm in different postures, and the synchronization action time of the fixture. Subsequently, within the dynamic pre-positioning domain, the controller optimizes the pre-positioning points of the robotic arm based on the process rhythm parameters, generating a set of candidate pre-positioning points according to the reachability range and joint limitations of the robotic arm. The generation of candidate points can combine uniform sampling or gridding methods to ensure coverage of the entire feasible space. For each candidate pre-positioning point, rhythm matching calculations are performed on the robotic arm based on the process rhythm parameters to ensure that the time for the robotic arm to reach each pre-positioning point and the final target point is synchronized with the production rhythm. In specific implementation, the motion time model of the robotic arm is used to predict the movement time of the robotic arm from the previous pre-positioning point to the current candidate point, and the expected arrival time of the robotic arm is calculated by combining the completion time of the preceding workstation and the relay start time of the target workstation. Subsequently, the estimated arrival time is compared with the target arrival time to obtain the rhythm deviation. Points with a rhythm deviation less than a preset deviation value are selected as the optimal pre-positioning point sequence. The motion time consumption model is obtained by collecting historical motion data of the robotic arm under different tasks and postures, including the initial and final poses of each joint, the corresponding actual motion time, and constraint parameters such as fixture actions and workpiece states. Feature processing is performed on the collected data, constructing input feature vectors from information such as joint angle differences, joint velocity and acceleration constraints, path complexity, and interference constraints, with motion time consumption as the output label. A neural network model is then trained, and loss functions such as mean squared error are used to optimize the model parameters, enabling the model to accurately predict the motion time of the robotic arm under arbitrary postures and constraints. After training, the robotic arm motion time consumption model can be used to predict the movement time between candidate pre-positioning points in real time and combine it with process rhythm parameters to calculate the rhythm deviation, further optimizing the optimal pre-positioning point sequence. The preset deviation value can be set based on the precision provided by the robotic arm manufacturer, the fixture installation error, and the production line debugging experience. By collecting historical operation data, the error distribution of the robotic arm's actual arrival at the target pose is calculated, and a certain confidence interval is taken as the preset deviation value.

[0226] Calculate the shortest movement time for the robotic arm to reach each optimal pre-docking point sequence, obtain the current position of the robotic arm, and calculate the shortest movement distance using the distance between the current position of the robotic arm and each optimal pre-docking point sequence. Alternatively, calculate the shortest movement time for each optimal pre-docking point sequence using the robotic arm's movement speed from its pose data.

[0227] Based on the optimal pre-docking point sequence and the predicted time window, the time-matching pre-docking trajectory of the robotic arm is generated, and the control system is based on the optimal pre-docking point sequence of the robotic arm. The time-matching pre-positioning trajectory of the robotic arm is generated within the predicted time window. The optimal pre-positioning point sequence is calculated by comprehensively considering factors such as robotic arm joint limitations, reachability, collision risk, and gripper motion synchronization within the dynamic pre-positioning domain. Each pre-positioning point corresponds to the spatial position and attitude information of the robotic arm's end effector. The predicted time window is determined by the process rhythm parameters, the predicted completion time of the preceding station, and the relay start time of the target station, and is used to provide the ideal time range for the robotic arm to reach the target station [t]. start , t end It allows for a certain amount of time flexibility to adapt to clamping errors or environmental disturbances. During implementation, based on the spatial location information of the optimal pre-positioning point sequence and the corresponding shortest movement time, the estimated time for the robotic arm to reach each pre-positioning point is calculated, forming a spatiotemporal matching trajectory that includes spatial location and time information. Through this trajectory, the robotic arm can synchronize its movement time with the process rhythm while ensuring the optimal spatial path. This allows it to continuously, efficiently, and safely complete the pre-positioning action within the dynamic pre-positioning domain, improving the automation level and operational efficiency of the production line. At the same time, it enhances the adaptability to fixture movements, clamping errors, and dynamic environments.

[0228] By combining the shortest motion time of the robotic arm to each optimal pre-positioning point sequence with the time-matched pre-positioning trajectory, reachable pre-positioning points are determined. Reachable pre-positioning points are those that the robotic arm can reach both spatially and temporally within the dynamic pre-positioning domain, meeting the requirements of the process rhythm. Subsequently, the control system compares the shortest motion time of each candidate pre-positioning point with the corresponding estimated arrival time in the trajectory to determine whether the robotic arm can reach the pre-positioning point within a specified time range, under the constraints of joint speed, acceleration, and safety distance. Only when the shortest motion time of a candidate pre-positioning point meets the time-matching requirements is the pre-positioning point determined as reachable and included in the final optimal pre-positioning point sequence. Through this method, the control system can ensure optimal spatial path for the robotic arm while synchronizing motion time with the process rhythm, ensuring that the robotic arm continuously, efficiently, and safely completes pre-positioning actions within the dynamic pre-positioning domain, thereby improving the automation level and operational accuracy of the production line.

[0229] Within the dynamic pre-positioning domain generated by combining process rhythm parameters, the system first calculates the shortest movement time for the robotic arm to reach each candidate pre-positioning point and matches this shortest movement time with the target reception time window output by the time-series prediction model. Subsequently, the system compares the time difference between each candidate pre-positioning point, i.e., the degree of deviation between the actual reachable time of the robotic arm and the predicted time window, and selects the pre-positioning point with the smallest deviation as the final dynamic pre-positioning point.

[0230] When the second clamp is within the docking time window for aligning with the semi-finished longitudinal beam, input variables composed of process rhythm parameters are invoked. These parameters are input into a pre-trained clamping compensation regression model. This model performs regression calculations on the mapping relationship between the input parameters and historical clamping deviation data, outputting the corresponding clamping compensation value. This clamping compensation value is used to correct the positional deviation, angular deviation, or clamping force deviation of the second clamp during the actual clamping process, thereby ensuring that the clamp can complete the clamping operation on the semi-finished product with optimal compensation within the docking time window. The clamping compensation regression model is a supervised learning regression model. It collects clamping action data of the second clamp under different process rhythm parameter conditions. This historical clamping action data includes input features such as the process cycle time corresponding to the clamping action, the time interval from the completion of the preceding station to the start of the clamping action, the operating speed, acceleration, and load status of the robotic arm at the start of the clamping action, as well as corresponding output parameters such as clamping force, clamping position deviation, and clamping angle deviation. Then, the collected data is subjected to outlier removal, normalization, and clamping action time window alignment to build a stable training sample set. The training sample set is then input into the regression model, and the model parameters are optimized by minimizing the error between the predicted clamping compensation value and the actual clamping compensation target value, thereby obtaining a clamping compensation regression model that can predict the clamping compensation value based on process rhythm parameters and equipment operating status.

[0231] By determining the dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture, the position of the robotic arm can be determined in advance, improving operating efficiency. This not only improves the collaborative operation efficiency of the robotic arm and fixture, ensuring precise matching of production cycle time, but also significantly enhances the automation and intelligence level of the production process.

[0232] This application also provides a machine-readable storage medium storing instructions for causing a machine to execute the aforementioned collaborative control method for an automotive longitudinal beam production line.

[0233] This application also provides a collaborative control system for an automotive longitudinal beam production line, comprising:

[0234] The memory is configured to store instructions; and

[0235] The processor is configured to retrieve instructions from memory and, when executing instructions, to implement the aforementioned collaborative control method for an automotive longitudinal beam production line.

[0236] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0237] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0238] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0239] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0240] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0241] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0242] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0243] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0244] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A collaborative control method for an automotive longitudinal beam production line, characterized in that, Applied to an automotive longitudinal beam production line, the automotive longitudinal beam production line includes at least two fixtures, workstations, and a robotic arm, including: Perform error-proofing checks on the longitudinal beam production line; After the longitudinal beam production line passes the error-proofing test, the longitudinal beam individual parts are pre-processed to obtain the longitudinal beam base parts; The longitudinal beam base component is locked by the first clamp and sent into the first target station to perform the welding process, thereby obtaining a semi-finished longitudinal beam and acquiring the first posture data of the semi-finished longitudinal beam. When the second target station receives the semi-finished longitudinal beam, it acquires the operation data corresponding to the second target station and the second posture data of the semi-finished longitudinal beam on the second fixture. The first process relay compensation domain is determined based on the first attitude data and the second attitude data. A dynamic safety domain is generated based on the operation data corresponding to the second target workstation. Obtain the initial process rhythm baseline for the second target station; Using the dynamic safety domain as a constraint, the initial process rhythm benchmark is corrected based on the operation data corresponding to the first process relay compensation domain and the second target station, and the process rhythm parameters are determined. The dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture are determined based on the process rhythm parameters. The motion reference template of the second fixture is determined by the clamping compensation value on the second fixture. Based on the dynamic pre-positioning point and motion reference template, collaborative strategies are planned and generated for the robotic arms in the second fixture and the third target workstation, respectively. The determination of the dynamic pre-position of the robotic arm in the third target station and the clamping compensation value on the second fixture based on the process rhythm parameters includes: Obtain the current pose data of the robotic arm in the third target workstation; The timing window for the robotic arm to receive the semi-finished longitudinal beam and the docking time window for the second fixture in the third target station are predicted using a pre-trained timing prediction model based on process rhythm parameters. Based on the prediction time window and the current pose data of the robotic arm, a dynamic pre-positioning domain is generated; Within the dynamic pre-positioning domain, the optimal pre-positioning point sequence of the robotic arm is calculated by combining process rhythm parameters; Calculate the shortest movement time for the robotic arm to reach each optimal pre-position sequence; Generate the time-matching pre-positioning trajectory of the robotic arm based on the optimal pre-positioning point sequence and the predicted time window; The reachable pre-deployment point is determined by combining the shortest motion time with the time-matched pre-deployment trajectory; The reachable pre-departure point with the smallest time difference between the shortest travel time and the predicted time window is selected as the dynamic pre-departure point; Within the docking time window of the second fixture, the process rhythm parameters are input into the clamping compensation regression model to obtain the clamping compensation value on the second fixture.

2. The method according to claim 1, characterized in that, The error-proofing verification of the longitudinal beam production line includes: Initialize the fixture control module and signal acquisition module in the production line; Verify the effectiveness of signal acquisition by the signal acquisition module; After the signal acquisition module passes the signal acquisition validity verification, the preset foolproof sample is placed in the target tooling fixture, and it is determined whether the positioning point of the preset foolproof sample is consistent with that of the target tooling fixture. When the preset error-proof sample and the target tooling fixture are aligned at their positioning points, control the target tooling fixture to execute the clamping command. The signal acquisition module is activated to acquire the clamping status signal and error-proof identification signal of the target tooling fixture; Based on the clamping status signal and the error prevention identification signal, determine whether the target tooling fixture and the preset error prevention sample meet the preset error prevention logic conditions. After the preset error-proofing logic conditions are met, the target tooling fixture is controlled to execute a reset command, thus completing the error-proofing verification.

3. The method according to claim 1, characterized in that, The longitudinal beam components include a first longitudinal beam component, a second longitudinal beam component, a third longitudinal beam component, and a fourth longitudinal beam component. The preprocessing of the longitudinal beam components to obtain the longitudinal beam base components includes: The first longitudinal beam component is placed in the first pre-processing station using the first pre-processing fixture and welded to obtain the first spare component. The second longitudinal beam component is placed in the second pretreatment station using the second pretreatment fixture and welded to obtain the second spare component. The first spare part and the second spare part are placed in the third pretreatment station using the third pretreatment fixture and welded to obtain the first longitudinal beam assembly; The third longitudinal beam component is placed in the fourth pre-processing station using the fourth pre-processing fixture and welded to obtain the third spare component. The fourth longitudinal beam component is placed in the fifth pre-processing station using the fifth pre-processing fixture and welded to obtain the fourth spare component. The third and fourth spare parts are placed in the sixth pretreatment station using the fifth pretreatment fixture and welded to obtain the second longitudinal beam assembly. The longitudinal beam base component is obtained by using the first longitudinal beam assembly and the second longitudinal beam assembly.

4. The method according to claim 1, characterized in that, The determination of the first process relay compensation domain based on the first attitude data and the second attitude data includes: A first key point dataset is extracted based on the first posture data. The first key point dataset includes the first weld positioning point data, the first fixture positioning center point data, and the first reinforcing rib support point data. The second key point dataset is extracted based on the second posture data. The second key point dataset includes the second weld positioning point data, the second fixture positioning center point data, and the second reinforcing rib support point data. The first and second key point datasets are mapped to the reference coordinate system, and the deflection curve deviation, torsional attitude deviation and planar positioning deviation are calculated based on the first and second key point datasets. The deviation type is determined by combining the deflection curve deviation, torsional attitude deviation, and planar positioning deviation. The deviation types include clamping positioning inaccuracy and transport deformation deviation. When the deviation type is transport deformation deviation, a regionalized deformation model of the longitudinal beam is constructed by combining deflection curve deviation, torsional attitude deviation and planar positioning deviation. The longitudinal beam regional deformation model is divided into three subdomains: weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain. The first process relay compensation domain is determined by combining the weld zone compensation subdomain, the auxiliary support compensation subdomain, and the positioning constraint zone compensation subdomain.

5. The method according to claim 4, characterized in that, The process of dividing the longitudinal beam regional deformation model to determine the weld zone compensation subdomain, auxiliary support compensation subdomain, and positioning constraint zone compensation subdomain includes: The longitudinal beam skeleton curve is obtained by fitting the longitudinal beam regional deformation model. The longitudinal beam skeleton curve is discretized to obtain the discrete points of the curve, and the curvature value of each discrete point is calculated to obtain the skeleton curvature distribution of the semi-finished longitudinal beam. The rate of change of curvature between discrete points of two adjacent curves is calculated using the skeleton curvature distribution. The region in the longitudinal beam regional deformation model whose curvature change rate is greater than or equal to the preset first threshold is taken as the weld zone compensation subdomain. The region in the longitudinal beam regional deformation model whose rate of curvature change is greater than or equal to the preset second threshold and less than the preset first threshold is designated as the auxiliary support compensation subdomain. The region in the longitudinal beam regional deformation model with a curvature change rate less than a preset third threshold is taken as the location constraint area compensation subdomain. Among them, the preset first threshold is greater than the preset second threshold and the preset second threshold is greater than the preset third threshold.

6. The method according to claim 1, characterized in that, The generation of a dynamic safety domain based on the operation data corresponding to the second target workstation includes: Obtain the workstation geometry layout of the second target workstation; Based on the operation data corresponding to the second target workstation, obtain the pose parameters and motion parameters of the robotic arm in the second target workstation, as well as the motion range of the second fixture and the movement position information of the operator. An initial virtual fence boundary is constructed based on the workstation geometric layout, and the initial virtual fence boundary is divided into a core danger zone, a buffer warning zone, and a safe operation zone. The initial virtual fence boundary is extended based on the operator's movement location information to obtain the first virtual fence range, which is located outside the buffer warning zone. Using the first virtual fence range as a constraint benchmark, the initial virtual fence boundary is extended by combining the pose and motion parameters of the robotic arm to obtain the second virtual fence range. The extended boundary of the second virtual fence range does not break through the boundary of the first virtual fence range. Using the first and second virtual fence ranges as dual constraints, the third virtual fence range is obtained by locally thickening the initial virtual fence boundary according to the movement range of the second clamp. A dynamic security domain is generated by combining the first virtual fence range, the second virtual fence range, and the third virtual fence range.

7. The method according to claim 6, characterized in that, The process of extending the initial virtual fence boundary to obtain the second virtual fence range, using the first virtual fence range as a constraint benchmark and combining the pose and motion parameters of the robotic arm, includes: The spatial operation trajectory of the robotic arm is determined by the pose parameters of the robotic arm. Using the first virtual fence range as a constraint benchmark, the extended boundary area of ​​the second virtual fence range is determined in combination with the spatial operation trajectory of the robotic arm; The dynamic expansion rate of the extended boundary region is determined based on the robotic arm's motion parameters. The range of the second virtual fence is determined by combining the dynamic expansion rate of the extended boundary area, wherein the extended boundary does not exceed the range of the first virtual fence.

8. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform the collaborative control method for an automotive longitudinal beam production line according to any one of claims 1 to 7.

9. A collaborative control system for an automobile longitudinal beam production line, characterized in that, include: The memory is configured to store instructions; as well as The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the collaborative control method for an automotive longitudinal beam production line according to any one of claims 1 to 7.

Images