A simulation dispensing method for a fastener assembly process

Abstract

The present application relates to the technical field of fastener assembly, in particular to a simulation distribution method for fastener assembly process, comprising: obtaining position coordinates of ideal assembly points and actual assembly points respectively through CAD model and three-coordinate measuring machine, and calculating deviation vectors of actual assembly points and ideal assembly points, taking the deviation vectors as sample points, fitting deviation function to output error tensor field; correcting the position of ideal assembly point of each fastener to generate a corrected point position, generating a process path sequence based on the corrected point position, constructing a node set and an edge set, calculating the scheduling priority weight of each node, and constructing a task cooperation graph based on the node set and the edge set; defining an assembly total scheduling cost function under the constraints of the task cooperation graph, and outputting a task distribution sequence by using a heuristic search strategy, converting each assembly point position and torque setting value of the task distribution sequence into control coordinates and equipment control format under the reference system of the work station respectively to generate complete action instructions.

Description

Technical Field

[0001] This invention relates to the field of fastener assembly technology, and more particularly to a simulation allocation method for fastener assembly processes. Background Technology

[0002] Fasteners, as critical connectors in industrial manufacturing, directly impact product structural safety and performance stability through their assembly quality. This is particularly true in sectors like automotive, rail transportation, home appliances, and high-end equipment manufacturing, where the precision and efficiency of fastener assembly are crucial for ensuring production cycle time and product reliability. With the widespread adoption of automated assembly lines and the increasing number of multi-model mixed-line production, fastener assembly processes are gradually shifting from traditional manual methods to simulation-driven intelligent approaches. However, existing process simulations often rely on idealized models and standard operating conditions, neglecting factors commonly encountered in actual production, such as hole position deviations, part deformations, material batch differences, and equipment station characteristics. This leads to significant discrepancies between simulation results and actual execution, manifesting as difficulties in hole alignment, abnormal assembly torque, and disordered assembly cycle time. In severe cases, this can result in rework or even line stoppage. Furthermore, current simulation results, when transmitted to the assembly allocation stage, often only provide a fixed sequence and static resource arrangement, failing to fully consider logical dependencies between workstations, spatial interference, and cycle time conflicts in multi-workstation collaborative assembly. This can easily lead to workstation idleness, task congestion, or resource waste.

[0003] While attempts have been made to improve the process through empirical rules or local optimization strategies, the lack of high-fidelity assembly error modeling and scalable task scheduling mechanisms has prevented the formation of a self-consistent "modeling-simulation-scheduling-execution" closed-loop process. Consequently, it is difficult to effectively support the dual requirements of high precision and high efficiency for fastener assembly production lines. Summary of the Invention

[0004] The purpose of this invention is to provide a simulation allocation method for fastener assembly processes to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A simulation allocation method for fastener assembly processes includes:

[0007] The position coordinates of the ideal assembly point and the actual assembly point are obtained by using a CAD model and a coordinate measuring machine, respectively. The deviation vector between the actual assembly point and the ideal assembly point is calculated. Radial basis function interpolation is used to fit the deviation vector as sample points and the deviation function is fitted. The error tensor field is output through the deviation function.

[0008] The ideal assembly point of each fastener is corrected to generate a correction point. A process path sequence is generated based on the correction point. The process path sequence includes multiple assembly tasks. Each assembly task consists of an assembly point and a corresponding torque setting value.

[0009] Construct a node set and an edge set. Each node in the node set corresponds to an assembly task. The assembly point and torque setting value are embedded as node attributes. Calculate the scheduling priority weight for each node. The edge set includes multiple types of edge relationships. Each type of relationship is generated by an independent judgment logic and is marked as a different edge weight category in the graph. Construct a task collaboration graph based on the node set and edge set.

[0010] The total scheduling cost function for assembly is defined under the constraints of the task collaboration graph, and a heuristic search strategy is used to output the task allocation sequence, which includes multiple tasks, and each task includes nodes inherited from the task collaboration graph.

[0011] Each assembly point and torque setpoint in the task assignment sequence is converted into control coordinates and equipment control format under the workstation reference system. Combined with the response model of the actual equipment, a complete action command including activation signal, parameter setting and duration is generated and sent to the workshop control system through the industrial communication bus.

[0012] Preferably, the heuristic search strategy includes:

[0013] The task collaboration diagram is sorted topologically by workstation number. For each node in the task collaboration diagram, the executable assembly time point is enumerated in the node candidate workstation set.

[0014] Calculate the minimum combination that satisfies the total scheduling cost function at the corresponding work sites;

[0015] If multiple combinations have the same cost, the combination with the lowest torque switching cost should be selected first.

[0016] If scheduling fails, the previous task is rolled back and its scheduling assignment is replaced, and the process is repeated.

[0017] Preferably, the assembly tasks are arranged in the process path sequence in the order of assembly execution.

[0018] Preferably, the multiple types of edge relationships include:

[0019] Physical exclusion edges are used to forcibly add a unidirectional edge if the correction space positions of two assembly tasks are too close and their normal directions are consistent, thus indicating the execution order of the two assembly tasks.

[0020] Structural dependency edges are used to restrict the structural execution order of fasteners in a given location.

[0021] Workstation conflict edges are used to generate edges in the order of the process path sequence if two assembly tasks are mapped to the same workstation number, thus determining the execution order of the two assembly tasks.

[0022] Preferably, the task collaboration diagram includes multiple nodes, each node including location, torque, workstation and priority.

[0023] Preferably, the action command includes spatial control coordinates, torque setting value, workstation number, start time, and preload control signal.

[0024] Preferably, after sending the action commands to the workshop control system via the industrial communication bus, the process also includes:

[0025] If it is an automated workstation, the PLC or robot controller will parse and execute the commands.

[0026] For manual assisted workstations, the interaction is converted to graphic prompts, voice prompts, and confirmation buttons through the human-machine interface.

[0027] Preferably, after generating a complete action command including an enable signal, parameter settings, and duration by combining the response model of the actual device, the method further includes:

[0028] All variables from the simulation, modeling, and scheduling phases are encapsulated into data structures that can be directly parsed and executed by the control system.

[0029] Preferably, before the process is parsed and executed by the PLC or robot controller, the following steps are also included:

[0030] The system defines the execution window value for task initiation and preloads the action control structure corresponding to the task within the execution window value before the corresponding execution time of the task arrives.

[0031] Preferably, the execution window value is derived from the task buffer characteristics of the field PLC system and the initialization cycle of the tightening gun.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] This invention proposes a simulation allocation method for fastener assembly processes, constructing an overall scheme consisting of five stages: assembly error modeling, path simulation generation, collaboration graph construction, task allocation optimization, and instruction issuance. By introducing a spatial error tensor field to quantitatively model the geometric deviations of actual parts, the simulation input is made closer to real working conditions. Based on this, the generated process path not only corrects the assembly points but also dynamically optimizes the torque and sequence by combining error distribution and process library data, ensuring the executability of the path.

[0034] The path is further transformed into an assembly task collaboration diagram, comprehensively expressing the process logic, spatial constraints, and workstation resource characteristics. A scheduling optimization model incorporating error intensity, torque switching, and dependency complexity is designed on this diagram structure to generate workstation-level task allocation results. Finally, through workstation coordinate transformation and structured instruction encapsulation, the scheduling results are converted into execution instructions that can be directly issued to the workshop control system. This method, for the first time, tightly integrates the entire process of assembly error modeling, process path optimization, and task allocation and execution, forming a complete link from virtual simulation to actual process control. It effectively solves the problems of simulation and execution disconnect and task allocation mismatch with on-site conditions in traditional technologies, achieving an overall improvement in the accuracy, cycle time, and resource utilization of fastener assembly processes. Attached Figure Description

[0035] Figure 1 This is a flowchart of a simulation allocation method for fastener assembly process in a specific embodiment of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] refer to Figure 1 As shown, this application proposes a simulation allocation method for fastener assembly processes, including:

[0038] Step 1: Obtain the position coordinates of the ideal assembly point and the actual assembly point using the CAD model and a coordinate measuring machine, respectively. Calculate the deviation vector between the actual assembly point and the ideal assembly point. Use radial basis function interpolation, taking the deviation vector as sample points, to fit a deviation function. Output the error tensor field through the deviation function. Specifically, this includes:

[0039] In fastener assembly processes, common problems include hole misalignment and unevenness of reference surfaces in actual manufacturing. If ideal CAD models are still used for simulation, the results often fail to reflect reality, leading to issues such as "hole misalignment" and "abnormal bolt force" during actual tightening. Therefore, the first step must be to measure and model the workpiece to obtain an error field that reflects the actual assembly conditions.

[0040] The input is a CAD model from the design department. This model contains the ideal coordinates of all fastener hole positions. Taking a car engine block as an example, it has hundreds of bolt holes. Each hole has clear center coordinates and diameter on the design drawings, but in actual parts, the actual position and shape of each hole are often affected by casting and machining errors.

[0041] To obtain this error data, a coordinate measuring machine (CMM) can be used to inspect the actual workpiece. Pre-set inspection points around each bolt hole; for example, measure the coordinates of one point at each of the four directions of 0°, 90°, 180°, and 270° around the hole. Simultaneously, several inspection points are also arranged on a reference plane. The CMM will output the actual spatial coordinates of these points. By comparing with the corresponding points in the CAD model By comparison, the deviation vector at that point can be obtained:

[0042] ;

[0043] in The coordinates of the points were obtained by measuring a coordinate measuring machine. These are the ideal coordinates in the CAD model. This is the deviation vector. Taking a bolt hole as an example, if the design value is (100.000, 50.000, 0.000), and the measured value is (100.120, 49.950, 0.010), then... This is (0.120, -0.050, 0.010). These specific values ​​directly reflect the actual drift of the hole position in three directions.

[0044] Since it is impossible to measure every single point, the error distribution across the entire assembly area must be estimated using a finite number of detection points. Therefore, a continuous error tensor field needs to be constructed. A common method is to use radial basis function interpolation to interpolate each... Using these sample points, a function is fitted that can provide a deviation value for any spatial location:

[0045] ;

[0046] in This is the total number of testing sites. These are the weighting coefficients obtained from the fitting process. It uses a radial basis function (such as a Gaussian kernel function), which ensures more accurate deviations in areas closer to the measurement point. This allows for direct input of the design coordinates when simulating any unmeasured bolt hole location. and through The corresponding deviation correction is obtained. For example, at the position of bolt hole No. 37 in the cylinder block (200,80,0), the calculation result may be (0.085,-0.030,0.012), and this vector can be used as the input for subsequent simulation path correction.

[0047] Step 2: Correct the ideal assembly point of each fastener to generate correction points. Based on the correction points, generate a process path sequence. The process path sequence includes multiple assembly tasks, each consisting of an assembly point and a corresponding torque setting value. Specifically, it includes:

[0048] The spatial error tensor field obtained in step one Having already been able to provide deviation correction values ​​for any design coordinates, this step generates a sequence of process paths that can be directly used for assembly execution. The path not only corrects the spatial position of each fastener but also generates torque settings and assembly sequences that match the current part state.

[0049] First, determine the ideal assembly point for each fastener. Make position corrections to obtain the actual assembly points. :

[0050] ;

[0051] In the formula, The first one pre-set in the process library The coordinates of each design point (imported from the design model and fixed before production). The deviation vector of the error tensor field given in step one at this coordinate is (established by interpolation of coordinate measuring data and radial basis functions). These are the correction points used to generate the path. In actual implementation, the program adjusts the point set... Call one by one Complete batch corrections, such as design points. through The deviation vector is obtained after calculation. The corrected point is then... (The values ​​are for illustrative purposes only and will be calculated directly by the system on-site.)

[0052] Subsequently, from the integrated perspective of "assembly sequence - torque setting", the correction point set was targeted. Find a path that satisfies process feasibility. (Using a sequence...) Indicates the assembly sequence of fasteners ( For the first (the corresponding fastener index), to Indicates the first The torque setting value for each fastener. Define an objective function to optimize both sequence and torque simultaneously:

[0053] ;

[0054] The meanings and implementation methods of each term in the formula are as follows: The deviation strength term for this assembly site is directly derived from step one. Calculated at the corresponding design points, the larger the deviation, the more difficult the assembly alignment becomes; This is the term representing the intensity of local changes in the deviation field. It is obtained by performing analytical or differential differentiation on the radial basis function interpolation model in step one, and is used to penalize locations with severe local fluctuations (avoiding the initial installation of areas prone to deformation accumulation). This is the weight of the item; Number the fasteners according to their specifications in the process library. The preset recommended torque (obtained from the process parameter table of diameter, material, surface condition and target preload; the lookup process is a deterministic table lookup and does not involve model inference) is set up. The extent to which the penalty deviates from the recommended value This is the weight of the item; These are sequence constraints, derived from the adjacency matrix generated from the design model (established based on the distance between holes and the affiliation of the same flange). If If a pattern of "continuous tightening of adjacent holes" or "local panel warping caused by tightening from the inside out" occurs, a discrete penalty is applied to the corresponding adjacent steps. This penalty is determined by the offline calculated adjacency matrix and flange partition table (implemented in the program as applying a penalty to adjacent steps). Query adjacent relationships and accumulate a fixed penalty.

[0055] In practice, an initial order is first generated using a greedy strategy of "adjacent priority dispersion" (preferably starting with "diagonal dispersion" within the same flange), followed by local swapping and small-amplitude operations using 2-opt. Adjust the alternating descent objective function. Updates only move within the safe zone allowed by the process library, and follow... The size is slightly adjusted (a slightly higher setting is used when the deviation is large, and a closer setting is used when the deviation is small). The entire solution process converges to a stable sequence and a set of torque settings. Example: When the center distance between two holes is very close and they belong to the same flange, the adjacency matrix determines them as "adjacent." If the initial sequence is adjacent continuous assembly, then... The algorithm penalizes adjacent steps by attempting to separate them into non-adjacent steps using 2-opt, and for delayed aperture positions, if their... and If the torque is relatively large, the torque setting will be slightly increased within the allowable range to enhance the ability to resist assembly clearances.

[0056] The output of this step is a process path sequence. Each of them It consists of a pair of data: the corrected spatial location. With the corresponding torque setting The order of the sequence determines the assembly execution order. Path sequence In the next step, it will be directly converted into a task collaboration graph for scheduling.

[0057] Step 3: Construct a node set and an edge set. Each node in the node set corresponds to an assembly task. The assembly point and torque setting are embedded as node attributes. Calculate the scheduling priority weight for each node. The edge set includes multiple types of edge relationships, each generated by an independent judgment logic and labeled with different edge weight categories in the graph. Construct a task collaboration graph based on the node set and edge set, specifically including:

[0058] After the calculations in step two, the process path sequence has been obtained. Each assembly action unit From a corrected spatial point and optimized torque setting value The path is linear and sequential, designed to meet simulation executability requirements. However, on real fastener assembly lines, tasks are often completed synchronously and in parallel by multiple workstations. Therefore, this linear path structure must be transformed into a multi-workstation schedulable directed graph structure, i.e., an assembly task collaboration graph. This makes each task The relationships between them can be expressed in a structured way and have contextual information that can be used for scheduling optimization.

[0059] The input is the output path from the previous step. Every single one must be used to its fullest extent. and And maintain the consistency of its calculation source (i.e. It is designed by point and error tensor field Calculated It is a combination (The results are derived from joint optimization of the process library and deviation gradient). The core idea of ​​this step is to combine each... Transform into graph nodes And establish a set of directed edges between nodes. It is used to describe the dependencies between tasks, physical constraints, and scheduling priorities.

[0060] First, construct the node set. Each node Corresponding to one assembly task ,in and It is embedded as a node attribute. The workstation number is also attached. (through spatial location) (Obtained by matching with the assembly line workstation layout drawing), assembly normal direction (From CAD design model) Extraction) and scheduling priority weight The latter will be defined in this step.

[0061] Then, construct the edge set. We designed three types of edge relationships, each generated by an independent judgment logic and labeled with different edge weight categories in the graph for use by the scheduling optimization algorithm.

[0062] 1. Physically exclusive edges (cannot be parallelized): If two tasks... , If the correction space positions are too close and the normal directions are consistent (e.g., adjacent holes in the same direction), it may cause tool interference or worker operation conflicts. In this case, a unidirectional edge should be forcibly added. ,express Must precede Execution. The judgment condition is:

[0063] ;

[0064] in It is the operating safety distance threshold (set by tightening the gun arm length and the assembly tool guard distance). , yes , Assembly normal direction, It is the normal consistency threshold (e.g., 0.95 corresponds to an included angle of less than 18°), which reflects the very common but often ignored factor of "spatial interference" in fastener field.

[0065] 2. Structural Dependency Edges (Process Sequence): Some fasteners must be assembled strictly according to the structural sequence, such as "main connection first, then secondary connection" or "inner before outer." This is extracted from the assembly hierarchy structure of the CAD model and statically imported into the system, without the need for algorithmic reasoning. For each set of dependencies... Directly establish directed edges The edge weights are labeled as "structural dependencies".

[0066] 3. Workstation Conflict Edge (Scheduling Control): If two tasks are mapped to the same workstation number... Furthermore, since there is no explicit structural dependency and no forced sorting due to physical interference, to ensure the uniqueness of the execution order of the control system, the path is followed. The order of edge generation , indicating the system First of all, After.

[0067] Next, we need to calculate the scheduling priority weight for each node. In this step, we designed a priority function specifically for fastener scenarios. This function not only considers node deviations and the number of preceding tasks, but also introduces a "torque stability" regularization term. This term penalizes locations with drastic torque changes, preventing frequent switching of process parameters due to continuous assembly of different types of fasteners at the same workstation. The definition is as follows:

[0068] ;

[0069] in It is the deviation intensity of the error tensor field at the original design point (established in step one, the value can be directly looked up). It is a node The number of incoming edges (i.e., the number of preceding dependencies) in a graph reflects its bottleneck status in the graph structure. This represents the current node and its previous node along its path. The torque difference, i.e. This is used to penalize devices for frequently changing settings. Three coefficients. , , The values ​​are set based on engineering experience, for example (0.6, 0.3, 0.1), which represents a scheduling strategy that prioritizes deviation and takes scheduling structure and equipment stability as secondary factors.

[0070] The output is the completed task collaboration graph. Each node has a location. Torque workstation and priority Each edge encodes scheduling constraints such as space conflicts, process logic, or resource competition.

[0071] Step 4: Define the total assembly scheduling cost function based on the constraints of the task collaboration graph, and output the task allocation sequence using a heuristic search strategy. The task allocation sequence includes multiple tasks, and each task includes nodes inherited from the task collaboration graph, specifically including:

[0072] In the previous step, a task collaboration graph was constructed that includes actual spatial location, torque parameters, workstation mapping, assembly dependencies, and scheduling priorities. This diagram structure represents the original linear process path. This is transformed into a multi-station collaborative and schedulable logical network, serving as a bridge connecting the simulation process and actual execution resources. The core task of this step is to use the collaboration graph... As input, and under the premise of satisfying its topological and physical constraints, generate scheduling and allocation results oriented towards workstation resources. Each of them This indicates when and by which workstation a task is performed, ensuring the efficient, conflict-free, and stable operation of the assembly system in the actual workshop.

[0073] Since this application focuses on fastener assembly processes, in multi-station synchronous operation scenarios, there are three types of highly coupled scheduling challenges among tasks: First, due to structural dependencies and physical exclusions in the diagram (such as spatial conflicts and structural pre-positioning), some tasks must be scheduled sequentially; second, the cost of station switching is nonlinear, especially in tightening equipment, where frequent changes in torque settings will lead to fluctuations in station load; third, the scheduling strategy has extremely high requirements for cycle time robustness, and the maximum execution time window of the system must be controlled. To address these issues, this step designs an assembly task allocation optimization model for fastener process scenarios. By simultaneously considering priority weights, equipment switching penalties, resource utilization, and assembly cycle time compression targets, the optimal solution for task allocation is achieved within a low-complexity, achievable algorithm framework.

[0074] First, under the premise of satisfying the topological constraints of the collaboration graph, define the overall assembly scheduling cost function. :

[0075] ;

[0076] in: Indicates task Actual startup time Compared to the earliest start time The difference is used to penalize dependency chain waits; It is the torque switching overhead function, if the task Compared to the previous task In the same workstation Above, and torque difference Exceeding the device's acceptable switching threshold If the cost is 1, then the cost is 0; otherwise, the cost is 0. This indicates the time at which the workstation is located. The load rate is calculated from the workstation scheduling table and is used to suppress excessive resource stacking. As a special regularization term, stability control is added for the "high error, high coupling assembly point" scenario in this application, and its definition is as follows:

[0077]

[0078] in These are experience-weighted coefficients. It is the deviation strength of the error tensor field defined in step one from the original design point. It is a node The graph in-degree reflects the complexity of its preceding dependencies. This innovative approach combines assembly errors and collaborative complexity to prioritize the stable allocation of complex process points, avoiding excessive parallelism or resource jumps.

[0079] To solve this objective function minimization problem, a heuristic search strategy of "hierarchical decoupling + priority greedy algorithm" is adopted. The specific process is as follows:

[0080] 1. Perform topological sorting on the collaboration graph to ensure that all dependencies satisfy the sequential logic;

[0081] 2. In the sorted task sequence, according to Tasks are scheduled in descending order of priority. ;

[0082] 3. For each In its candidate work station set In the middle, enumerate the executable time points. Find the workstation that satisfies the dependency constraints. smallest combination;

[0083] 4. If multiple combinations have the same cost, prioritize the combination with the lowest torque switching cost;

[0084] 5. If scheduling fails (due to deadlock caused by constraints in the diagram), roll back the previous task and replace its scheduling assignment, repeating the process.

[0085] Output the task allocation sequence after scheduling is complete. Each of them Indicates task Assigned to work station And plans to do so at a specific time. Execution ensures that all nodes satisfy dependencies, avoids spatial interference, and controls workstation switching costs and peak resource load.

[0086] Step 5: Convert each assembly point and torque setpoint in the task allocation sequence into control coordinates and equipment control format under the workstation reference system. Combined with the actual equipment response model, generate complete action instructions including activation signals, parameter settings, and duration. Send these action instructions to the workshop control system via the industrial communication bus. Specifically, this includes:

[0087] The core task of this step is to process the task scheduling results output from step four. Directly convert it into a sequence of control commands that can be sent to the actual assembly system in the workshop. Each task Includes from collaboration graph Inherited task nodes (Including actual assembly points) Setting torque Priority (Information such as...) In this step, the consistency and continuity of these variables must be maintained to ensure that the assembly system execution process truly connects with the previous modeling and scheduling results.

[0088] The input is the output of step four. In this sequence, each Indicates task Dispatch to workstation It is expected that in time Execution. The task... This comes from step three, which structurally corresponds to an assembly action. Both of these variables come from the preceding simulation path generation process. It was through Corrected assembly points, This is the torque value optimized based on the process database and the coupling of deviations. Therefore, in this step, the complete usage must be employed. , , and Four pieces of information.

[0089] The specific process for generating execution instructions is as follows:

[0090] First, each assembly point needs to be... Converted to control coordinates in the workstation reference system To ensure that equipment such as robotic arms and tightening guns are in place at the workstation The system can identify and accurately move to that point. This transformation process relies on the pre-calibrated homogeneous coordinate transformation matrix for each workstation. The formula is as follows:

[0091] ;

[0092] in, It is the pose transformation matrix obtained through laser calibration or vision calibration system during the installation and commissioning phase at the workstation. It is the simulation path generated in step two that passes through the error field. The obtained spatial positions are corrected. The control coordinates that the tightening gun tip should reach are then obtained through matrix multiplication. These coordinates will be used to generate motion control commands for the servo system.

[0093] Then, the torque value The code is converted into a device control format and combined with the actual device's response model to generate a complete action command structure that includes an enable signal, parameter settings, and duration. The allowable torque step size and maximum setting frequency are usually provided by the equipment manufacturer. For example, the tightening control module may support setting updates every 10ms in 0.1Nm steps, and the system can set accordingly. The target value and transition time.

[0094] To address potential cycle time drift or upstream delays during actual execution, this step also defines a flexible execution window for task initiation to ensure the on-site system has a certain degree of fault tolerance. Specifically, the system will... Corresponding execution time Before arrival Within a given timeframe, the action control structure corresponding to the task is preloaded for the controller to execute in a buffered manner. The system supports a preloading window of [number missing]. This parameter is derived from the task buffer characteristics of the field PLC system and the initialization cycle of the tightening gun.

[0095] In summary, each generated execution instruction Includes the following components: spatial control coordinates ,Depend on Calculated torque setpoint This is the output result of the simulation path and scheduling optimization; workstation number. Used for instruction routing; startup time It is directly provided by the scheduling output; the preloaded control signal is used for clock cycle fault tolerance.

[0096] These instructions will be sent to the workshop control system via industrial communication buses (such as EtherCAT, ModbusTCP, or OPCUA). If it is an automated workstation, it will be parsed and executed by the PLC or robot controller; if it is a manually assisted workstation, it will be converted into interactive forms such as graphic prompts, voice prompts, and confirmation buttons through a human-machine interface (HMI).

[0097] For example, regarding scheduling results ,in , Nm, For the transformation matrix of workstation A03, the system will... The control structure is loaded in the first 0.5 seconds, and control commands are sent on the control bus. The instructions contain control coordinates and target torque .

[0098] The output is a structured sequence of control instructions. It is entirely based on the scheduling result of the previous step. The information in the data is fully encapsulated into a data structure that can be directly parsed and executed by the control system.

[0099] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A simulation dispensing method for a fastener assembly process, characterized by, The method comprises the following steps: Obtaining the position coordinates of ideal assembly points and actual assembly points through CAD models and three-coordinate measuring machines respectively, and calculating the deviation vector of the actual assembly points and the ideal assembly points, using radial basis function interpolation, taking the deviation vector as a sample point, fitting a deviation function, and outputting an error tensor field through the deviation function; Correcting the position of each ideal assembly point of the fastener to generate a corrected point position, and generating a process path sequence based on the corrected point position, wherein the process path sequence comprises a plurality of assembly tasks, and each assembly task is composed of an assembly point position and a corresponding torque setting value; Constructing a node set and an edge set, wherein each node in the node set corresponds to an assembly task, the assembly point position and the torque setting value are embedded as node attributes, the scheduling priority weight of each node is calculated, the edge set comprises a plurality of edge relationships, each relationship is generated by an independent judgment logic, and is marked as a different edge weight category in the graph, and a task cooperation graph is constructed based on the node set and the edge set; Defining an assembly total scheduling cost function under the constraints of the task cooperation graph, and outputting a task allocation sequence by using a heuristic search strategy, wherein the task allocation sequence comprises a plurality of tasks, and each task inherits a node from the task cooperation graph; Converting each assembly point position and torque setting value in the task allocation sequence into control coordinates and device control formats in a work station reference system, combining an actual device response model to generate complete action instructions containing an enable signal, a parameter setting and a duration, and sending the action instructions to a workshop control system through an industrial communication bus.

2. The simulation dispensing method for a fastener assembly process according to claim 1, wherein, The heuristic search strategy comprises: Topologically sorting the task cooperation graph according to the work station number, enumerating executable assembly time points in the node candidate work station set for each node in the task cooperation graph; Calculating the minimum combination that satisfies the total scheduling cost function at the corresponding work station point; If the costs of multiple combinations are the same, the combination with the lowest torque switching cost is preferentially selected; If the scheduling fails, the previous task is rolled back and its scheduling allocation is replaced, and the process is repeated.

3. The simulation dispensing method for a fastener assembly process according to claim 1, wherein, The order of the assembly tasks in the process path sequence is the assembly execution order.

4. The simulation dispensing method for a fastener assembly process according to claim 1, wherein, The plurality of edge relationships comprise: A physical repulsion edge, which is used to add a single edge to indicate the execution order of two assembly tasks if the modified spatial positions of the two assembly tasks are too close and the normal directions are consistent; A structure-dependent edge, which is used to limit the structure execution order of the part fastener; A work station conflict edge, which is used to generate an edge to determine the execution order of two assembly tasks if the two assembly tasks are mapped to the same work station number according to the order of the process path sequence.

5. The simulation dispensing method for fastener assembly process according to claim 1, wherein, The task cooperation graph comprises a plurality of nodes, and each node comprises a point position, a torque, a work station and a priority.

6. The simulation dispensing method for a fastener assembly process according to claim 1, wherein, The action instructions comprise spatial control coordinates, torque setting values, work station numbers, start times and preloading control signals.

7. The simulation dispensing method for fastener assembly process according to claim 1, wherein, After sending the action instructions to the workshop control system through the industrial communication bus, the method further comprises the following steps: If it is an automatic work station, the PLC or robot controller is used to analyze and execute; If it is a manual auxiliary work station, the man-machine interface is used to convert the action instructions into a graphical prompt, a voice prompt and a confirmation button form for interaction.

8. The simulation dispensing method for fastener assembly process according to claim 1, wherein, After generating the complete action instruction containing the enable signal, parameter setting and duration based on the response model of the actual equipment, the method further comprises: Encapsulating all variables in the simulation, modeling and scheduling stages into a data structure that can be directly parsed and executed by the control system.

9. The simulation dispensing method for a fastener assembly process according to claim 7, wherein, Before being parsed and executed by the PLC or robot controller, the method further comprises: The system defines an execution window value for the task start, and preloads the action control structure corresponding to the task within the execution window value before the execution time corresponding to the task arrives.

10. The simulation dispensing method for a fastener assembly process according to claim 9, wherein, The execution window value is derived from the task buffer characteristics of the on-site PLC system and the initialization cycle of the tightening gun.

Images