A numerical control method for large-area multi-laser synchronous splicing and melting forming

By employing a large-area multi-laser synchronous splicing and melting forming CNC control method, the problems of low efficiency and quality defects in traditional single-laser beam forming have been solved, achieving efficient and stable large-area component forming.

CN123113671APending Publication Date: 2026-11-17深圳市金石三维打印科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611299636.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-26
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Traditional single-laser beam serial scanning melting forming process is inefficient in forming large-area components and is prone to quality defects such as warping, interlayer cracking, and excessive internal porosity, which cannot meet the requirements of high-efficiency and high-consistency industrial mass production.

Method used

A large-area multi-laser synchronous splicing and melting forming CNC control method is adopted. The three-dimensional model of the component to be formed is preprocessed, and the error features are extracted by using the CNN-LSTM-Transformer model. The nonlinear mapping relationship between the thermal field and process parameters is established, and the processing is iteratively optimized by collecting data in real time.

Benefits of technology

It improves the forming efficiency and splicing metallurgical quality of large-area components, reduces the forming scrap rate, and enhances process stability and anti-interference ability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113671A_ABST
    Figure CN123113671A_ABST
Patent Text Reader

Abstract

This invention relates to the field of laser additive manufacturing technology and discloses a CNC control method for large-area multi-laser synchronous splicing and melting forming. The method involves preprocessing a three-dimensional model of the large-area component to be formed to obtain the initial CNC scanning path and basic CNC process parameters of the laser units, and collecting error data to form a target dataset. This target dataset is then input into a CNN-LSTM-Transformer model to extract error spatial features, temporal cumulative features, and splicing area association weights. Iterative calculations are performed on the temporal compensation values, position correction values, and speed adaptation values ​​of each laser unit. A nonlinear mapping relationship between the thermal field and process parameters is established to generate a pre-control optimization scheme for CNC parameters. This pre-control optimization scheme is then distributed to the execution unit, which collects real-time data on the thermal field, molten pool, and trajectory conditions. Iterative calculations optimize the overall melting forming process of the large-area component. This invention improves forming efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of laser additive manufacturing technology, specifically to a numerical control method for large-area multi-laser synchronous splicing and melting forming. Background Technology

[0002] Laser melting additive manufacturing technology, with its outstanding advantages such as near-net-shape forming without molds, high material utilization, excellent metallurgical bonding performance of components, and high degree of freedom in structural design, has gradually replaced traditional processes such as integral forging, segmented welding and splicing, and mechanical milling and cutting. It has become the mainstream key technology supporting the integrated manufacturing of large core metal components for high-end major equipment. With the advancement of industrial upgrading in aerospace and new energy equipment towards large-scale production, the market requirements for the surface area size, forming efficiency, overall dimensional accuracy, uniformity of microstructure and properties, and metallurgical quality of spliced ​​areas of formed components have increased significantly. Traditional single-laser beam serial scanning melting forming process has exposed its fundamental technical shortcomings. The effective scanning area of ​​a single laser is limited, and the forming of large-area components requires long-term, layer-by-layer, and area-by-area serial processing, which increases the forming cycle by several times and results in extremely low production efficiency. Moreover, during the long-term continuous forming process, the heat accumulation in the component matrix is ​​severe, and the temperature gradient across the entire area is continuously unbalanced, which can easily lead to a series of quality defects such as overall warping deformation, interlayer cracking, excessive internal porosity, and uneven grain size. It is completely unable to meet the rigid requirements of large-scale, high-efficiency, and high-consistency industrial mass production of ultra-large-area components. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing a numerical control method for large-area multi-laser synchronous splicing and melting forming.

[0004] The first aspect of this invention provides a numerical control method for large-area multi-laser synchronous splicing and melting forming, the method comprising the following steps: The three-dimensional model of the large surface area component to be formed is preprocessed to obtain the initial CNC scanning path of the laser unit and the basic CNC process parameters, and error data is collected to form the target dataset; The target dataset is input into the CNN-LSTM-Transformer model to extract error spatial features, temporal cumulative features and splicing area association weights, and to iteratively calculate the temporal compensation value, position correction value and velocity adaptation value of each laser unit. Establish a nonlinear mapping relationship between the thermal field and process parameters to generate a pre-control and optimization scheme for CNC parameters; The pre-adjustment and optimization scheme of CNC parameters is sent to the execution unit, and the thermal field, molten pool and trajectory working condition data are collected in real time. Iterative calculation and optimization are performed to complete the overall melting and forming of large area components.

[0005] Optionally, in a first implementation of the first aspect of the present invention, the step of performing three-dimensional model preprocessing on the large surface area component to be formed, obtaining the initial CNC scanning path of the laser unit and basic CNC process parameters, and collecting error data to form a target dataset includes: A three-dimensional digital solid model of the large surface area component to be formed is obtained, and the overall forming working surface of the large surface area is divided into a uniform and seamless region to obtain the laser forming sub-processing region. The initial CNC scanning path and basic CNC process parameters of the laser unit are determined based on the machining area of ​​the forming sub-sub-unit. Collect dynamic error data of mechanical transmission clearance of each CNC servo motion axis, dynamic response delay fluctuation data of servo motor, differential data of laser generator light emission trigger hysteresis, and initial position error data of scanning trajectory. The collected data is cleaned, standardized, integrated, and summarized to form the target dataset.

[0006] Optionally, in a second implementation of the first aspect of the present invention, determining the initial CNC scanning path and basic CNC process parameters of the laser unit based on the forming sub-processing area includes: Based on the three-dimensional digital solid model, according to the effective scanning area of ​​a single laser of the multi-laser melting and forming equipment, the large-area overall forming working surface is divided into equal and seamless regions to obtain the laser forming sub-processing area. For a fixed-width splicing overlap transition buffer area reserved between two adjacent laser forming sub-regions, the initial numerical control scanning path corresponding to the laser emitting unit is generated based on the boundary coordinates of the laser forming sub-regions and the forming structure features completed by partitioning. The basic CNC process parameters are obtained by acquiring the reference laser power, reference scanning feed speed, interlayer tool lifting height, and initial light output sequence of each laser unit.

[0007] Optionally, in a third implementation of the first aspect of the present invention, the step of inputting the target dataset into the CNN-LSTM-Transformer model, extracting error spatial features, temporal cumulative features, and splicing region association weights, and iteratively calculating the temporal compensation value, position correction value, and velocity adaptation value of each laser unit includes: The target dataset is input into the CNN-LSTM-Transformer model. The CNN branch extracts spatial dimension features from the error data of the target dataset, identifies and outputs spatial correlation features. The spatial correlation features are transmitted to the LSTM branch to iterate on the temporal evolution pattern of error data, the cumulative and progressive change characteristics of error, and the temporal decay characteristics of equipment operation error, so as to obtain the temporal dimension error evolution features. Input the temporal dimension error evolution features into the Transformer branch to strengthen the association weight features of the splicing region; Based on spatial correlation features, temporal dimension error evolution features, and correlation weight features, the temporal compensation value, position correction value, and velocity adaptation value corresponding to each laser unit are iteratively calculated, and finally the corrected quantization parameters of each laser unit are output.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, the step of inputting the temporal dimension error evolution features into the Transformer branch to strengthen the association weight features of the splicing region includes: Input the temporal dimension error evolution features into the Transformer branch and perform position encoding operations on the temporal dimension error evolution features; Initialize the basic weight matrix for multi-head attention operation, and preset the basic weights corresponding to the splicing transition area according to the partition layout of the large-area multi-laser forming structure. Correlation calculation is performed on the error correlation relationships of different time nodes and different laser units in the temporal dimension error evolution characteristics to generate a global feature correlation matrix, and weight bias assignment is performed on the feature nodes corresponding to the splicing boundary transition region. Based on the weight bias assignment results, attention weight weights are weighted and summed, and after normalized linear transformation, the final weighted and enhanced key association features of the splicing region are output.

[0009] Optionally, in the fifth implementation of the first aspect of the present invention, the step of establishing a nonlinear mapping relationship between the thermal field and process parameters and generating a pre-control optimization scheme for numerical control parameters includes: Obtain the corrected quantization parameters of the laser unit output by the CNN-LSTM-Transformer model, and extract the basic correlation features of the global thermal field spatial distribution corresponding to the corrected parameters; Based on the spatial distribution-based correlation characteristics, synchronous temporal feature alignment calculations are performed on the dynamic change time series process and the thermal field state time series evolution process. Global focusing calibration is performed on the correlation weights of the modified quantization parameters and the correlation characteristics of the local coupling of the thermal field to strengthen the corresponding correlation between the parameters of the key forming area and the thermal field, and the correlation characteristics are obtained. Based on the spatial distribution basic correlation characteristics, temporal characteristics and correlation characteristics, a nonlinear mapping is established between the corrected quantization parameters, basic CNC process parameters and the real-time thermal field distribution state of the whole domain. Based on the preset thermoforming safety threshold, the required adjustment level of process parameters for each forming area is determined, and the results of the control parameters of all laser units are integrated to generate a pre-control optimization scheme for CNC parameters.

[0010] Optionally, in the sixth implementation of the first aspect of the present invention, the step of sending the pre-adjustment and optimization scheme of the numerical control parameters to the execution unit, collecting real-time data on the thermal field, molten pool, and trajectory conditions, and iteratively calculating and optimizing to complete the overall melting and forming process of the large-area component includes: The CNC parameter pre-adjustment optimization scheme is divided into high-efficiency forming zone, splicing and overlapping heat-sensitive transition zone, and edge high-precision boundary sensitive zone, and the operation instructions are issued to the execution unit. Real-time data on thermal field, molten pool, and trajectory are collected and input into the CNN-LSTM-Transformer model. The real-time data is then compared and verified against safety thresholds. Based on comparison, verification, and iterative optimization of CNC control parameters, the large-area component is melt-formed and processed as a whole.

[0011] A second aspect of the present invention provides a large-area multi-laser synchronous splicing and melting forming CNC control system, the system comprising: The preprocessing module is used to preprocess the three-dimensional model of the large surface area component to be formed, obtain the initial CNC scanning path of the laser unit and the basic CNC process parameters, and collect error data to form the target dataset; The computation module is used to input the target dataset into the CNN-LSTM-Transformer model, extract error spatial features, temporal cumulative features and splicing area association weights, and iteratively calculate the temporal compensation value, position correction value and velocity adaptation value of each laser unit. The generation module is used to establish the nonlinear mapping relationship between the thermal field and process parameters, and to generate a pre-control and optimization scheme for CNC parameters. The execution module is used to send the pre-adjustment and optimization scheme of CNC parameters to the execution unit, collect real-time data on the thermal field, molten pool, and trajectory conditions, and iteratively calculate and optimize to complete the overall melting and forming of large-area components. A third aspect of the present invention provides a large-area multi-laser synchronous splicing and melting forming CNC control device, the large-area multi-laser synchronous splicing and melting forming CNC control device comprising a memory and at least one processor, the memory storing instructions; the at least one processor calling the instructions in the memory to cause the large-area multi-laser synchronous splicing and melting forming CNC control device to execute the various steps of the large-area multi-laser synchronous splicing and melting forming CNC control method as described in any of the preceding claims.

[0012] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of a large-area multi-laser synchronous splicing and melting forming CNC control method as described in any of the preceding claims.

[0013] In the technical solution provided by this invention, a three-dimensional model preprocessing is performed on the large-area component to be formed to obtain the initial CNC scanning path and basic CNC process parameters of the laser unit, and error data is collected to form a target dataset. The target dataset is input into the CNN-LSTM-Transformer model to extract error spatial features, temporal cumulative features and splicing area association weights, and iteratively calculates the temporal compensation value, position correction value and speed adaptation value of each laser unit. A nonlinear mapping relationship between the thermal field and process parameters is established to generate a CNC parameter pre-control optimization scheme. The CNC parameter pre-control optimization scheme is sent to the execution unit, and the thermal field, molten pool and trajectory working condition data are collected in real time. Iterative calculation and optimization are performed to complete the overall melting and forming of the large-area component. This invention uses the CNN-LSTM-Transformer model as the core control carrier, which breaks through the technical limitations of single algorithm control prediction lag and feature extraction one-sidedness. It simultaneously realizes error spatial feature mining, temporal evolution memory and splicing area weight focusing processing, improves forming efficiency and splicing metallurgical quality, has strong anti-interference ability and high process stability, and effectively reduces the forming scrap rate. Attached Figure Description

[0014] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0015] Figure 1 A flowchart of a large-area multi-laser synchronous splicing and melting forming CNC control method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a large-area multi-laser synchronous splicing and melting forming CNC control equipment provided in an embodiment of the present invention. Detailed Implementation

[0016] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0017] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1The flowchart of the large-area multi-laser synchronous splicing and melting forming CNC control method provided in this embodiment of the invention includes the following steps: Step 101: Perform 3D model preprocessing on the large surface area component to be formed to obtain the initial CNC scanning path and basic CNC process parameters of the laser unit, and collect error data to form the target dataset; In this embodiment, the three-dimensional digital solid model of the target component to be formed is retrieved. Based on the effective scanning area of ​​the single laser of the multi-laser melting forming equipment, an adaptive equalization partitioning algorithm is used to divide the overall forming working surface of the large area into equal and seamless regions. This results in the formation of multiple independent laser forming sub-processing areas with uniform specifications and regular edge boundaries. A fixed-width splicing and overlapping transition buffer area is reserved between two adjacent independent forming sub-areas to complete the structured partitioning preprocessing of the large area forming region. Based on the boundary coordinates and forming structure characteristics of each independent laser forming sub-area after partitioning, a trajectory continuous planning algorithm is used to generate a dedicated CNC scanning motion path for each laser emitting unit. This ensures that the scanning trajectories of each sub-area are connected continuously and smoothly, and that the overlap of the scanning trajectories in the splicing and overlapping transition areas meets the standard. Simultaneously, a complete set of basic CNC process parameters, such as the reference laser power, reference scanning feed speed, interlayer tool lifting height, and initial light output sequence, are set for each laser unit. All laser emitting units, CNC servo motion platforms, and supporting multi-sensor monitoring components underwent unified installation, alignment, and initial factory calibration and debugging. All laser CNC equipment was brought into a stable, no-load trial operation state. Dynamic error data of mechanical transmission clearance, servo motor dynamic response delay fluctuation data, laser generator light emission trigger hysteresis differentiation data, and initial position deviation data of the scanning trajectory were collected for each CNC servo motion axis using position encoders and photoelectric sensing modules. All collected mechanical motion errors, servo response errors, laser emission errors, and other multi-dimensional basic raw error data were uniformly classified, cleaned, and invalid data was removed. The data was standardized and integrated according to the model input data format specifications, ultimately forming a structured target dataset that can be directly input into composite machine learning models.

[0018] Step 102: Input the target dataset into the CNN-LSTM-Transformer model, extract the error space features, temporal accumulation features and splicing area association weights, and iteratively calculate the temporal compensation value, position correction value and velocity adaptation value of each laser unit; In this embodiment, the preprocessed target dataset is input into the CNN-LSTM-Transformer composite model. The CNN convolutional neural network branch of the model performs comprehensive spatial dimension feature extraction operations on the original error data of various laser devices contained in the target dataset. It performs deep feature mining and feature filtering on the spatial position deviation distribution information, the error distribution difference information of the forming area, and the spatial error correlation information of the splicing area corresponding to different laser units, accurately identifies the spatial distribution correspondence between different error data, and purifies and outputs stable and reliable multi-laser error spatial correlation features. The multi-laser error spatial correlation features extracted by the CNN branch are completely transmitted to the LSTM long short-term memory neural network branch of the model. The LSTM branch performs iterative learning and feature memory solidification processing on the dynamic temporal evolution law of error in the entire multi-laser forming process, the progressive accumulation and change trend of error layer by layer, and the temporal decay change process of long-term operation error of equipment according to the time series progressive operation logic. This completes the deep mining of temporal dimension error information and feature temporal correlation fitting, and generates temporal dimension error evolution features that can reflect the dynamic change law of error. The temporal dimension error evolution features output from the LSTM branch are input into the Transformer attention mechanism branch of the model. The Transformer branch, relying on multi-head attention operation logic, calculates the weight allocation of the error correlation degree at different positions in the overall forming area. It specifically increases the feature operation weight ratio corresponding to the laser splicing transition boundary area, strengthens the focus and purification of the multi-laser collaborative error coupling correlation weight features in the splicing area, and highlights the error correlation influence of the key splicing area. Using the spatial correlation features, temporal dimension error evolution features, and enhanced splicing area correlation weight features extracted by the model as the common operation basis, according to the multi-laser spatiotemporal collaborative matching iterative operation logic, the temporal compensation adjustment value, trajectory position deviation correction value, and motion speed collaborative adaptation value corresponding to each laser unit are subjected to multi-dimensional iterative quantitative calculation, repeatedly optimizing the matching accuracy of each parameter, and finally outputting the synchronous correction quantitative parameters exclusively matched for each laser unit.

[0019] In this embodiment, the temporal dimension error evolution features obtained by the LSTM branch are completely input into the Transformer model branch. The Transformer branch first performs position encoding operation on the received temporal dimension error evolution feature data. According to the order of the temporal nodes and the corresponding number of each laser unit, a unique position temporal identifier is independently assigned to each group of temporal error evolution features, completing the accurate binding and labeling of the position information of the temporal feature sequence, and clarifying the position correspondence of different error features in the overall temporal sequence. After the Transformer branch completes the position encoding operation, it initializes the basic weight matrix required for the multi-head attention mechanism operation. Combined with the overall structural partitioning layout of large-area multi-laser synchronous splicing and melting forming, according to the difference in process importance between each forming sub-region and the splicing overlapping transition region, the basic weight initial assignment of the splicing transition region is set in advance and entered, completing the pre-weight configuration preparation work for the attention operation. Based on the core computational mechanism of the multi-head attention algorithm, a comprehensive correlation quantification calculation is performed on the correlation of data changes at different time-series nodes within the temporal dimension error evolution features and the correlation of mutual influence of errors between different laser unit devices. Simultaneously, a global feature correlation matrix covering all time-series nodes and all laser units is generated. During the matrix operation, targeted weight bias assignment is performed on the feature nodes corresponding to the splicing boundary transition region to enhance the priority of feature operation in the splicing region. Based on the computation results after the weight bias assignment of the splicing region, the Transformer branch performs a weighted summation and fusion operation of the multi-head attention weights. The basic features of global temporal error and the enhanced weighted correlation features of the splicing region are deeply fused and superimposed. Then, the fused features are uniformly regularized and purified through normalized linear transformation operation to complete redundant feature filtering and effective feature enhancement. Finally, the key correlation feature data of the splicing region after weighted enhancement and optimization is output.

[0020] Step 103: Establish the nonlinear mapping relationship between the thermal field and process parameters, and generate a pre-control optimization scheme for CNC parameters; In this embodiment, the corrected quantization parameters of the laser unit output by the CNN-LSTM-Transformer model are obtained, and the basic correlation features of the global thermal field spatial distribution corresponding to the corrected parameters are extracted. Based on the basic correlation features of the spatial distribution, the dynamic change time series process and the thermal field state time series evolution process are synchronized with the time series feature alignment operation. The correlation weight of the corrected quantization parameters and the local coupling correlation features of the thermal field are globally focused and calibrated to strengthen the correspondence between the parameters of the key forming area and the thermal field, and the correlation features are obtained. According to the basic correlation features of the spatial distribution, the time series features and the correlation features, a nonlinear mapping between the corrected quantization parameters, the basic CNC process parameters and the global real-time thermal field distribution state is established. Based on the preset thermoforming safety threshold, the adjustment requirement of the process parameters corresponding to each forming area is determined, and the results of the partition control parameters of all laser units are integrated to generate a CNC parameter pre-control optimization scheme.

[0021] Step 104: Send the CNC parameter pre-adjustment and optimization scheme to the execution unit, collect the thermal field, molten pool and trajectory working condition data in real time, and iteratively calculate and optimize to complete the overall melting and forming of large area components.

[0022] In this embodiment, the final output CNC parameter pre-adjustment optimization scheme is matched to three types of forming areas according to the three-level forming zone control division standard: core high-efficiency forming area, splicing overlap heat-sensitive transition area, and edge high-precision boundary sensitive area. This completes the classification and distribution of exclusive differentiated CNC control parameters for different areas and the loading and configuration of CNC system parameters. The CNC core controller synchronously sends synchronous start operation commands to all laser emission units and CNC servo motion platform, controlling all laser units to run synchronously according to differentiated adaptation control parameters, and carrying out large-area zone collaborative synchronous splicing melt additive forming basic processing operation. The entire forming process is synchronously initiated with a multi-sensor real-time monitoring and acquisition process. Infrared full-domain thermal imaging temperature measurement sensors collect real-time temperature distribution data of the entire thermal field. High-definition visual monitoring cameras for the molten pool collect real-time images of the molten pool morphology and splicing seam forming status. Servo position real-time feedback encoders calibrate the actual running position data of the laser scanning trajectory. All real-time collected thermal field, molten pool, and trajectory data are transmitted back to the CNC core controller and simultaneously input into the CNN-LSTM-Transformer composite machine learning fusion prediction model. The model performs real-time updates and iterative comparison calculations of the working condition features, and accurately compares and verifies the real-time working condition data with pre-trained safety thresholds and standard benchmark features one by one. Based on real-time comparison and verification results, the model iteratively optimizes the CNC control parameters as needed, dynamically and in real-time fine-tunes the laser output power, scanning feed speed, and synchronous timing compensation, completing the fully closed-loop adaptive dynamic correction and control operation of the forming process to ensure stable and compliant single-layer forming conditions. After the single-layer fusion forming operation is completed, the model automatically performs verification operations on the fusion quality of the single-layer splicing seam, the uniformity of interlayer temperature, and the accuracy of the forming position. The verification results are fed back to the composite machine learning model to complete the iterative update of the model parameters and optimize the CNC control reference parameters for the next layer forming. The entire process of synchronous calibration, thermal field prediction, zoned differentiated forming, and closed-loop dynamic correction is repeatedly performed, layer by layer, to complete the fusion forming operation until the integrated splicing and fusion forming of the overall large-area component is fully completed.

[0023] Figure 2 This is a schematic diagram of the structure of a large-area multi-laser synchronous splicing and melting forming CNC control device 300 provided in an embodiment of the present invention. This large-area multi-laser synchronous splicing and melting forming CNC control device 300 can vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the large-area multi-laser synchronous splicing and melting forming CNC control device 300. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute a series of instruction operations in the storage media 330 on the large-area multi-laser synchronous splicing and melting forming CNC control device 300 to implement the method provided in the above embodiment.

[0024] A large-area multi-laser synchronous splicing and melting forming CNC control device 300 may further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 2 The structure of the large-area multi-laser synchronous splicing and melting forming CNC control equipment shown does not constitute a limitation on the computer equipment provided by the present invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0025] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the various steps of the large-area multi-laser synchronous splicing and melting forming CNC control method provided in the above embodiments.

[0026] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus / unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0027] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0028] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended technical solutions and their equivalents.

Claims

1. A large area multi-laser synchronous splicing fused forming numerical control regulation method, characterized in that, The method includes the following steps: The three-dimensional model of the large surface area component to be formed is preprocessed to obtain the initial CNC scanning path of the laser unit and the basic CNC process parameters, and error data is collected to form the target dataset; The target dataset is input into the CNN-LSTM-Transformer model to extract error spatial features, temporal cumulative features and splicing area association weights, and to iteratively calculate the temporal compensation value, position correction value and velocity adaptation value of each laser unit. Establish a nonlinear mapping relationship between the thermal field and process parameters to generate a pre-control and optimization scheme for CNC parameters; The pre-adjustment and optimization scheme of CNC parameters is sent to the execution unit, and the thermal field, molten pool and trajectory working condition data are collected in real time. Iterative calculation and optimization are performed to complete the overall melting and forming of large area components.

2. The large area multi-laser synchronous splicing fused forming numerical control method according to claim 1, characterized in that, The process involves preprocessing a 3D model of the large-area component to be formed, obtaining the initial CNC scanning path and basic CNC process parameters for the laser unit, and collecting error data to form a target dataset, including: A three-dimensional digital solid model of the large surface area component to be formed is obtained, and the overall forming working surface of the large surface area is divided into a uniform and seamless region to obtain the laser forming sub-processing region. The initial CNC scanning path and basic CNC process parameters of the laser unit are determined based on the machining area of ​​the forming sub-sub-unit. Collect dynamic error data of mechanical transmission clearance of each CNC servo motion axis, dynamic response delay fluctuation data of servo motor, differential data of laser generator light emission trigger hysteresis, and initial position error data of scanning trajectory. The collected data is cleaned, standardized, integrated, and summarized to form the target dataset.

3. The large area multi-laser synchronous splicing fused forming numerical control method according to claim 2, characterized in that, The determination of the initial CNC scanning path and basic CNC process parameters of the laser unit based on the forming sub-processing area includes: Based on the three-dimensional digital solid model, according to the effective scanning area of ​​a single laser of the multi-laser melting and forming equipment, the large-area overall forming working surface is divided into equal and seamless regions to obtain the laser forming sub-processing area. For a fixed-width splicing overlap transition buffer area reserved between two adjacent laser forming sub-regions, the initial numerical control scanning path corresponding to the laser emitting unit is generated based on the boundary coordinates of the laser forming sub-regions and the forming structure features completed by partitioning. The basic CNC process parameters are obtained by acquiring the reference laser power, reference scanning feed speed, interlayer tool lifting height, and initial light output sequence of each laser unit.

4. The large area multi-laser synchronous splicing fused forming numerical control method according to claim 1, characterized in that, The process of inputting the target dataset into the CNN-LSTM-Transformer model, extracting error spatial features, temporal cumulative features, and splicing region association weights, and iteratively calculating the temporal compensation value, position correction value, and velocity adaptation value of each laser unit includes: The target dataset is input into the CNN-LSTM-Transformer model. The CNN branch extracts spatial dimension features from the error data of the target dataset, identifies and outputs spatial correlation features. The spatial correlation features are transmitted to the LSTM branch to iterate on the temporal evolution pattern of error data, the cumulative and progressive change characteristics of error, and the temporal decay characteristics of equipment operation error, so as to obtain the temporal dimension error evolution features. Input the temporal dimension error evolution features into the Transformer branch to strengthen the association weight features of the splicing region; Based on spatial correlation features, temporal dimension error evolution features, and correlation weight features, the temporal compensation value, position correction value, and velocity adaptation value corresponding to each laser unit are iteratively calculated, and finally the corrected quantization parameters of each laser unit are output.

5. The large area multi-laser synchronous splicing fused forming numerical control method according to claim 4, characterized in that, The step of inputting the temporal dimension error evolution features into the Transformer branch to strengthen the association weight features of the splicing region includes: Input the temporal dimension error evolution features into the Transformer branch and perform position encoding operations on the temporal dimension error evolution features; Initialize the basic weight matrix for multi-head attention operation, and preset the basic weights corresponding to the splicing transition area according to the partition layout of the large-area multi-laser forming structure. Correlation calculation is performed on the error correlation relationships of different time nodes and different laser units in the temporal dimension error evolution characteristics to generate a global feature correlation matrix, and weight bias assignment is performed on the feature nodes corresponding to the splicing boundary transition region. Based on the weight bias assignment results, attention weight weights are weighted and summed, and after normalized linear transformation, the final weighted and enhanced key association features of the splicing region are output.

6. The large area multi-laser synchronous coalescence fusion forming numerical control method according to claim 1, characterized in that, The process of establishing a nonlinear mapping relationship between the thermal field and process parameters, and generating a pre-control and optimization scheme for numerical control parameters, includes: Obtain the corrected quantization parameters of the laser unit output by the CNN-LSTM-Transformer model, and extract the basic correlation features of the global thermal field spatial distribution corresponding to the corrected parameters; Based on the spatial distribution-based correlation characteristics, synchronous temporal feature alignment calculations are performed on the dynamic change time series process and the thermal field state time series evolution process. Global focusing calibration is performed on the correlation weights of the modified quantization parameters and the correlation characteristics of the local coupling of the thermal field to strengthen the corresponding correlation between the parameters of the key forming area and the thermal field, and the correlation characteristics are obtained. Based on the spatial distribution basic correlation characteristics, temporal characteristics and correlation characteristics, a nonlinear mapping is established between the corrected quantization parameters, basic CNC process parameters and the real-time thermal field distribution state of the whole domain. Based on the preset thermoforming safety threshold, the required adjustment level of process parameters for each forming area is determined, and the results of the control parameters of all laser units are integrated to generate a pre-control optimization scheme for CNC parameters.

7. The large area multi-laser synchronous tiled fusion forming numerical control method according to claim 1, characterized in that, The process of sending the pre-adjustment and optimization scheme of CNC parameters to the execution unit, collecting real-time data on the thermal field, molten pool, and trajectory conditions, and iteratively calculating and optimizing to complete the overall melting and forming of large-area components includes: The CNC parameter pre-adjustment optimization scheme is divided into high-efficiency forming zone, splicing and overlapping heat-sensitive transition zone, and edge high-precision boundary sensitive zone, and the operation instructions are issued to the execution unit. Real-time data on thermal field, molten pool, and trajectory are collected and input into the CNN-LSTM-Transformer model. The real-time data is then compared and verified against safety thresholds. Based on comparison, verification, and iterative optimization of CNC control parameters, the large-area component is melt-formed and processed as a whole.

8. A system for implementing a large area multi-laser synchronous splicing fused forming numerical control method as claimed in claim 1, characterized in that, The system includes: The preprocessing module is used to preprocess the three-dimensional model of the large surface area component to be formed, obtain the initial CNC scanning path of the laser unit and the basic CNC process parameters, and collect error data to form the target dataset; The computation module is used to input the target dataset into the CNN-LSTM-Transformer model, extract error spatial features, temporal cumulative features and splicing area association weights, and iteratively calculate the temporal compensation value, position correction value and velocity adaptation value of each laser unit. The generation module is used to establish the nonlinear mapping relationship between the thermal field and process parameters, and to generate a pre-control and optimization scheme for CNC parameters. The execution module is used to send the pre-adjustment and optimization scheme of CNC parameters to the execution unit, collect thermal field, molten pool and trajectory working condition data in real time, and iteratively calculate and optimize to complete the overall melting and forming of large area components.

9. A large area multi-laser synchronous splicing fused forming numerical control regulation device, characterized in that, The large-area multi-laser synchronous splicing and melting forming CNC control device includes a memory and at least one processor. The memory stores instructions. The at least one processor calls the instructions in the memory to cause the large-area multi-laser synchronous splicing and melting forming CNC control device to execute each step of the large-area multi-laser synchronous splicing and melting forming CNC control method as described in any one of claims 1-7.

10. A computer-readable storage medium having stored thereon instructions, the instructions comprising, When the instructions are executed by the processor, they implement each step of the numerical control control method for large-area multi-laser synchronous splicing and melting forming as described in any one of claims 1-7.