Fine control method and system for feeding and discharging of numerical control machine tool

Abstract

The invention provides a numerical control machine tool feeding and discharging fine control method and system, and relates to the technical field of laser cladding machining.The method comprises the steps that based on an image and a point cloud data set, coating fitting analysis is carried out, a first sub-stage coating three-dimensional model is generated, deviation comparison is carried out, and a second sub-stage coating deviation data set is determined; the preset powder feeding flow is corrected and analyzed in combination with the real-time powder amount of the powder cylinder so as to meet the requirement for the coating deviation data set of the second sub-stage, and the optimized powder feeding flow is obtained; and generating a second sub-stage optimization control strategy to execute feeding control of the second sub-stage scanning time window. By means of the method, the technical problems that according to an existing repeated laser cladding machining method, due to the fact that the cladding deviation of the previous machining stage cannot be accurately analyzed, targeted compensation cannot be conducted on the cladding deviation in time in the next machining stage, machining deviation is accumulated, and the overall cladding quality of a workpiece is poor can be solved; and the effect of improving the overall cladding quality of the workpiece can be achieved.

Description

Technical Field

[0001] This application relates to the field of laser cladding technology, and in particular to a method and system for precise control of material feeding and unloading in CNC machine tools. Background Technology

[0002] Laser cladding is an advanced material surface modification technology that uses a high-energy laser beam to melt powder materials and deposit them onto the surface of a substrate material, forming a coating with specific properties. This technology has many advantages, such as achieving high precision, high speed, low heat-affected zone, good material compatibility, and controllability, and is widely used in aerospace, automotive manufacturing, energy, and other industrial fields.

[0003] Currently, when performing repeated laser cladding processes, the existing technology cannot accurately analyze the cladding deviation in the previous processing stage, which leads to the inability to promptly compensate for the cladding deviation in the next processing stage. This results in the accumulation of processing deviations and a poor overall cladding quality of the workpiece. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for precise control of the feeding and unloading of CNC machine tools, in order to solve the technical problem that when performing repeated laser cladding processing in the existing process, the cladding deviation in the previous processing stage cannot be accurately analyzed, resulting in the inability to timely compensate for the cladding deviation in the next processing stage, leading to the accumulation of processing deviation and poor overall cladding quality of the workpiece.

[0005] In view of the above problems, this application provides a method and system for precise control of material feeding and unloading in CNC machine tools.

[0006] In a first aspect, this application provides a method for precise control of material feeding and unloading in a CNC machine tool. The method is implemented through a precise control system for material feeding and unloading in a CNC machine tool. The method includes: acquiring melting and processing indicators of a target workpiece, wherein the melting and processing indicators include a preset coating thickness, a preset coating width, and a preset coating shape; performing melting and processing of the target workpiece according to a preset melting and processing scheme, wherein the preset melting and processing scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder feed rate, and a predetermined number of repetitions; dividing the melting and processing indicators according to the predetermined number of repetitions to determine multiple stages of melting and processing indicators, and selecting a second stage of melting and processing indicators, wherein the second stage of melting and processing indicators includes a second coating thickness, a second coating width, and a second coating shape; and collecting data on the melting and processing process of the target workpiece using a dual-source monitoring device within a preset period to obtain a dual-source monitoring dataset, wherein the dual-source monitoring dataset includes an image dataset and a point cloud dataset. Based on the image dataset and the point cloud dataset, a coating fitting analysis is performed to generate a first-stage coating 3D model. A second-stage standard coating 3D model is constructed based on the second coating thickness, the second coating width, and the second coating shape. The first-stage coating 3D model and the second-stage standard coating 3D model are compared for mesh deviation to determine the second-stage coating deviation dataset. Real-time powder quantity in the powder cylinder is monitored and acquired to meet the second-stage coating deviation dataset. The predetermined powder feeding flow rate is then corrected based on the real-time powder quantity to obtain an optimized powder feeding flow rate. The scanning time for the second-stage is predicted based on the predetermined scanning speed and the mapped workpiece area of ​​the first stage to determine the second-stage scanning time window. An optimized control strategy for the second stage is generated based on the optimized powder feeding flow rate and the second-stage scanning time window, and the feeding control within the second-stage scanning time window is executed according to the optimized control strategy.

[0007] Secondly, this application also provides a precision control system for feeding and unloading CNC machine tools, used to execute a precision control method for feeding and unloading CNC machine tools as described in the first aspect, wherein the system includes: a melting processing index acquisition module, used to acquire melting processing indexes of a target workpiece, wherein the melting processing indexes include a preset coating thickness, a preset coating width, and a preset coating shape; a melting processing module, used to execute the melting processing of the target workpiece according to a preset melting processing scheme, wherein the preset melting processing scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder feeding flow rate, and a predetermined number of repetitions; a second-stage melting index selection module, used to divide the melting processing indexes according to the predetermined number of repetitions, determine multiple stages of melting indexes, and select a second-stage melting index, wherein the second-stage melting indexes include a second coating thickness, a second coating width, and a second coating shape; and a data acquisition module, used to acquire data on the melting processing process of the target workpiece using a dual-source monitoring device within a preset period, to obtain a dual-source monitoring dataset, wherein the dual-source monitoring dataset includes an image dataset and a point cloud. The system comprises the following modules: a coating deviation dataset determination module, used to perform coating fitting analysis based on the image dataset and the point cloud dataset, generate a first sub-stage coating 3D model, construct a second-stage standard coating 3D model based on the second coating thickness, the second coating width, and the second coating shape, and compare the mesh deviation between the first sub-stage coating 3D model and the second-stage standard coating 3D model to determine the second sub-stage coating deviation dataset; a correction analysis module, used to monitor and acquire the real-time powder quantity of the powder cylinder to meet the second sub-stage coating deviation dataset, and perform correction analysis on the predetermined powder feeding flow rate based on the real-time powder quantity to obtain an optimized powder feeding flow rate; a scanning time prediction module, used to predict the scanning time of the second sub-stage based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, and determine the second sub-stage scanning time window; and a feeding control module, used to generate a second sub-stage optimized control strategy based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and execute the feeding control within the second sub-stage scanning time window according to the second sub-stage optimized control strategy.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The melting and processing parameters of the target workpiece are obtained, including a preset coating thickness, a preset coating width, and a preset coating shape. The melting and processing of the target workpiece is performed according to a preset melting and processing scheme, including a preset laser power, a preset scanning speed, a preset powder flow rate, and a preset number of repetitions. The melting and processing parameters are divided into multiple stages based on the predetermined number of repetitions, and a second stage melting and processing parameter is selected, including a second coating thickness, a second coating width, and a second coating shape. Within a preset period, data is collected from the melting and processing process of the target workpiece using a dual-source monitoring device to obtain a dual-source monitoring dataset, including an image dataset and a point cloud dataset. Coating fitting analysis is performed based on the image dataset and the point cloud dataset to generate a first sub-coating. A staged coating 3D model is constructed. Based on the second coating thickness, second coating width, and second coating shape, a standard 3D model for the second stage coating is built. The mesh deviation of the first sub-stage coating 3D model and the second stage standard 3D model is compared to determine the second sub-stage coating deviation dataset. Real-time powder quantity in the powder cylinder is monitored and acquired. To meet the second sub-stage coating deviation dataset, the predetermined powder feeding flow rate is corrected and analyzed based on the real-time powder quantity to obtain an optimized powder feeding flow rate. The scanning time for the second sub-stage is predicted based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, determining the second sub-stage scanning time window. An optimized control strategy for the second sub-stage is generated based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and the feeding control within the second sub-stage scanning time window is executed according to the optimized control strategy. In other words, by using dual-source monitoring equipment to acquire images and point clouds during the repeated melting process of the workpiece, and then performing coating fitting analysis based on the image and point cloud datasets, a first-stage coating 3D model is generated. On the other hand, a second-stage standard coating 3D model is constructed based on the processing indicators of the second stage, and the mapping deviation of the first-stage coating 3D model and the second-stage standard coating 3D model is compared to obtain the second-stage coating deviation dataset. Then, with the goal of satisfying the second-stage coating deviation dataset, the predetermined powder feeding flow rate is corrected and analyzed in conjunction with the real-time powder cylinder powder quantity to obtain the optimized powder feeding flow rate. Based on the optimized powder feeding flow rate and the second-stage scanning time window, a second-stage optimized control strategy is generated. Finally, the feeding control of the second-stage scanning time window is executed according to the second-stage optimized control strategy, and the feeding optimization compensation of subsequent stages is performed using the same method until the workpiece completes the melting process. This can improve the precision and accuracy of the stage melting deviation analysis during the melting process, thereby enabling timely and effective compensation for stage melting deviations, reducing or avoiding the accumulation of deviations during the melting process, and achieving the technical effect of improving the overall melting quality of the workpiece.

[0009] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating a method for precise control of material feeding and unloading in a CNC machine tool, as described in this application. Figure 2 This is a schematic diagram of the process for generating the first sub-stage three-dimensional model of the coating in a refined control method for feeding and unloading of a CNC machine tool according to this application; Figure 3 This is a schematic diagram of the structure of a precision control system for feeding and discharging materials in a CNC machine tool according to this application.

[0012] Explanation of reference numerals in the attached figures: The module includes: 11 for obtaining melting processing indexes, 12 for melting processing, 13 for selecting melting indexes in the second stage, 14 for data acquisition, 15 for determining coating deviation dataset, 16 for correction analysis, 17 for scanning time prediction, and 18 for feeding control. Detailed Implementation

[0013] This application provides a method and system for precise control of material feeding and unloading in CNC machine tools. It solves the technical problem in existing processes of repeated laser cladding where the inability to accurately analyze cladding deviations from previous stages leads to a lack of timely and targeted compensation for these deviations in subsequent stages, resulting in accumulated processing deviations and poor overall cladding quality. This method improves the precision and accuracy of stage-specific cladding deviation analysis during the cladding process, enabling timely and effective compensation for these deviations, reducing or avoiding the accumulation of deviations, and ultimately improving the overall cladding quality of the workpiece.

[0014] The technical solutions in this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0015] Example 1 Please see the appendix Figure 1 This application provides a method for precise control of material feeding and unloading in a CNC machine tool. The method is applied to a precise control system for material feeding and unloading in a CNC machine tool, and specifically includes the following steps: Step 1: Obtain the melting and processing parameters of the target workpiece, wherein the melting and processing parameters include the preset coating thickness, preset coating width, and preset coating shape.

[0016] Specifically, firstly, based on the melting and cladding requirements of the target workpiece, melting and cladding parameters are obtained. These parameters include a preset coating thickness, a preset coating width, and a preset coating shape. The preset coating thickness refers to the thickness of the cladding layer formed on the workpiece during the melting and cladding process, determined according to specific application requirements and material properties. The preset coating width refers to the width of the cladding layer on the workpiece along a certain direction. The preset coating shape refers to the geometric shape of the cladding layer on the substrate, typically determined based on the specific shape of the workpiece and usage requirements. For example, in some applications, a smooth curved surface or a specific contour is required. Obtaining these melting and cladding parameters provides data support for subsequent melting and cladding deviation analysis.

[0017] Step 2: Perform the melting process on the target workpiece according to the preset melting process scheme, wherein the preset melting process scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder flow rate, and a predetermined number of repetitions.

[0018] Specifically, based on the material properties of the target workpiece, melting processing parameters, and the equipment performance of the CNC laser cladding machine tool, a melting processing scheme analysis is conducted to obtain a preset melting processing scheme. For example: First, the material properties of the target workpiece are analyzed, including its melting point, thermal conductivity, specific heat capacity, coefficient of thermal expansion, chemical composition, and microstructure. These properties will affect the formation of the molten pool, the size of the heat-affected zone, stress distribution, and the final coating quality during the cladding process. Next, the performance of the CNC laser cladding machine tool is evaluated, including the maximum power and minimum power adjustment range of the laser, the accuracy and speed of the scanning system, and the stability of the powder feeding system. Equipment performance will limit the range of selectable processing parameters. Finally, based on the material properties, The melting process parameters and equipment performance are considered. Appropriate process parameters such as laser power, scanning speed, and powder flow rate are selected, and a processing strategy is formulated, including multi-layer cladding, scanning path, and number of repetitions. The choice of processing strategy will affect the uniformity and overall performance of the coating. A preset melting process plan is generated, which includes a predetermined laser power, a predetermined scanning speed, a predetermined powder flow rate, and a predetermined number of repetitions. The predetermined powder flow rate can be calculated based on the thickness and width of the cladding layer, as well as the laser power and scanning speed, to ensure the uniformity of the cladding layer. The predetermined number of repetitions can be determined based on the thickness of the cladding layer and the thickness of a single cladding pass.

[0019] Step 3: Divide the melting process index according to the predetermined number of repetitions, determine multiple stages of melting indexes, and select the second stage melting indexes, wherein the second stage melting indexes include the second coating thickness, the second coating width, and the second coating shape.

[0020] Specifically, the melting process index is then divided according to the predetermined number of repetitions, that is, the entire processing process is decomposed into multiple stages, each of which has its own specific melting index, resulting in multiple stage melting indices; then, the second stage melting index is selected from the multiple stage melting indices. The second stage melting index refers to the specific index that needs to be achieved during the second repetition of laser cladding. The second stage melting index includes the second cladding thickness, the second cladding width, and the second cladding shape.

[0021] Step 4: Within a preset period, data is collected on the melting process of the target workpiece using a dual-source monitoring device to obtain a dual-source monitoring dataset, which includes an image dataset and a point cloud dataset.

[0022] Specifically, a preset period is configured, which is the time period corresponding to a sub-stage in each processing stage. Each processing stage includes multiple sub-stages with the same processing time, and can be set according to the actual situation. For example, the preset period can be set to 0.5 seconds. Then, within the preset period of the first sub-stage of the first stage, data on the melting process of the target workpiece is collected by a dual-source monitoring device. The dual-source monitoring device includes multiple image sensors at different angles and a laser point cloud scanning device to obtain a dual-source monitoring dataset. The dual-source monitoring dataset includes an image dataset and a point cloud dataset of the target workpiece completing the cladding area within the preset period.

[0023] Step 5: Perform overlay fitting analysis based on the image dataset and the point cloud dataset to generate a first sub-stage overlay 3D model. Construct a second-stage standard overlay 3D model based on the second overlay thickness, the second overlay width, and the second overlay shape. Compare the mesh deviations of the first sub-stage overlay 3D model and the second-stage standard overlay 3D model to determine the second sub-stage overlay deviation dataset.

[0024] Specifically, image fusion of the cladding area is performed based on the image dataset, and feature extraction is performed on the image fusion results to obtain the cladding contour of the cladding area. Then, using the cladding contour as a constraint, cladding fitting analysis is performed on the point cloud dataset to obtain the first sub-stage cladding point cloud dataset. Next, in a 3D simulation platform, 3D simulation modeling is performed based on the first sub-stage cladding point cloud dataset to generate the first sub-stage cladding 3D model. By performing cladding fitting analysis on the point cloud dataset, it can be ensured that the cladding layer matches the contour shape of the workpiece substrate, thereby improving the precision and accuracy of the first sub-stage cladding 3D model construction.

[0025] On the other hand, within the 3D simulation platform, 3D simulation modeling is performed based on the second coating thickness, the second coating width, and the second coating shape to generate a second-stage standard coating 3D model; a preset mesh size is obtained, which can be set according to the deviation comparison accuracy requirements, wherein the higher the deviation comparison accuracy requirements, the smaller the preset mesh size; based on the mapping of the first sub-stage coating 3D model, a second sub-stage standard coating 3D model corresponding to the workpiece area in the second-stage standard coating 3D model is obtained; based on the preset mesh size, the first sub-stage coating 3D model and the second sub-stage standard coating 3D model are meshed, and based on the meshing results, the first sub-stage coating 3D model and the second sub-stage standard coating 3D model are meshed to obtain a second sub-stage coating deviation dataset, wherein the second sub-stage coating deviation dataset includes a thickness deviation dataset, a width deviation dataset, and a shape deviation dataset.

[0026] Step Six: Monitor and acquire the real-time powder quantity of the powder cylinder. With the aim of meeting the coating deviation dataset of the second sub-stage, combine the real-time powder quantity of the powder cylinder with the predetermined powder delivery flow rate for correction analysis to obtain the optimized powder delivery flow rate.

[0027] Specifically, firstly, sensors, such as weight sensors, volume sensors, or optical sensors, are installed in the powder cylinder to monitor and acquire the real-time powder quantity in the powder cylinder, where the real-time powder quantity in the powder cylinder is the powder quantity in the powder cylinder at the start of the second sub-stage processing node. Next, with the aim of satisfying the coating deviation dataset of the second sub-stage, the predetermined powder feeding flow rate is corrected based on the coating deviation dataset of the second sub-stage, that is, the predetermined powder feeding flow rate is adjusted to compensate for the processing deviation caused by the first sub-stage, and a corrected powder feeding flow rate is obtained. Then, the corrected powder feeding flow rate is corrected a second time based on the real-time powder quantity in the powder cylinder to obtain an optimized powder feeding flow rate. There is a correlation between the powder quantity in the powder cylinder and the powder feeding flow rate. As the powder quantity in the powder cylinder decreases, the ability of the negative pressure device to pick up powder will gradually decrease, resulting in a reduction in the conveying volume.

[0028] Step 7: Based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, predict the scanning time of the second sub-stage and determine the scanning time window of the second sub-stage.

[0029] Step 8: Generate a second sub-stage optimized control strategy based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and execute the feeding control within the second sub-stage scanning time window according to the second sub-stage optimized control strategy.

[0030] Specifically, based on the predetermined scanning speed, the scanning time for the second sub-stage is predicted for the mapped workpiece area of ​​the first sub-stage. First, based on the processing results of the first sub-stage, the workpiece area to be processed in the second sub-stage is mapped. Then, based on the mapped workpiece area and the determined scanning speed, the scanning time for the second sub-stage is calculated. The scanning time can be obtained by calculating the area or volume of the workpiece area and multiplying it by the scanning speed, thus obtaining the scanning time window for the second sub-stage. This second sub-stage scanning time window represents the expected processing time cycle for the second sub-stage. Finally, based on the optimized powder feed rate and the second sub-stage scanning time window, an optimized control strategy for the second sub-stage is obtained, and powder feed rate control within the second sub-stage scanning time window is executed according to this optimized control strategy.

[0031] The aforementioned method for precise control of CNC machine tool feeding and unloading is applied to a precision control system for CNC machine tool feeding and unloading. It can solve the technical problem that when performing repeated laser cladding processing in existing processes, the inability to accurately analyze the cladding deviation of the previous processing stage leads to the inability to timely compensate for the cladding deviation in the next processing stage, resulting in the accumulation of processing deviations and poor overall cladding quality of the workpiece. First, the melting and processing parameters of the target workpiece are obtained, including a preset coating thickness, a preset coating width, and a preset coating shape. Then, the melting and processing of the target workpiece is performed according to a preset melting and processing scheme, including a preset laser power, a preset scanning speed, a preset powder flow rate, and a preset number of repetitions. Next, the melting and processing parameters are divided according to the predetermined number of repetitions to determine multiple stages of melting parameters, and a second stage melting parameter is selected, including a second coating thickness, a second coating width, and a second coating shape. Then, within a preset period, data is collected from the melting and processing process of the target workpiece using a dual-source monitoring device to obtain a dual-source monitoring dataset, including an image dataset and a point cloud dataset. Next, coating fitting analysis is performed based on the image dataset and the point cloud dataset to generate the first... A first-stage cladding 3D model is constructed. Based on the second cladding thickness, second cladding width, and second cladding shape, a second-stage standard 3D cladding model is built. The mesh deviation of the first-stage cladding 3D model and the second-stage standard 3D cladding model is compared to determine the second-stage cladding deviation dataset. Further, the real-time powder quantity in the powder cylinder is monitored and acquired. To meet the second-stage cladding deviation dataset, the predetermined powder feeding flow rate is corrected and analyzed based on the real-time powder quantity to obtain an optimized powder feeding flow rate. In addition, the scanning time of the second-stage is predicted based on the predetermined scanning speed and the mapped workpiece area of ​​the first stage to determine the second-stage scanning time window. Finally, an optimized control strategy for the second stage is generated based on the optimized powder feeding flow rate and the second-stage scanning time window, and the feeding control within the second-stage scanning time window is executed according to the optimized control strategy. This method improves the precision and accuracy of stage cladding deviation analysis during the cladding process, enabling timely and effective compensation for stage cladding deviations, reducing or avoiding deviation accumulation during the cladding process, and ultimately improving the overall cladding quality of the workpiece.

[0032] Further details are attached. Figure 2 As shown, step five of this application includes: A first data acquisition node is set based on the preset cycle, wherein the first data acquisition node is the end node of the first preset cycle; under the first data acquisition node, the dual-source monitoring device is used to acquire data on the melting process of the target workpiece to obtain an image dataset and a point cloud dataset, wherein the image dataset consists of multiple images from different angles; the image dataset is fused based on an image fusion strategy to generate a fused overlay image, and the fused overlay image is input into a preset contour extraction channel to obtain an overlay contour, wherein the preset contour extraction channel is constructed based on a convolutional neural network; using the overlay contour as a constraint, an overlay fitting analysis is performed on the point cloud dataset to obtain the three-dimensional overlay model of the first sub-stage.

[0033] Specifically, firstly, a first data acquisition node is set based on the preset period, wherein the first data acquisition node is the end node of the first preset period. Then, under the first data acquisition node, data is acquired from the cladding area of ​​the target workpiece during the melting process using the dual-source monitoring device, resulting in an image dataset and a point cloud dataset. The image dataset and the point cloud dataset represent the image dataset and point cloud dataset of the target workpiece within the preset period after the cladding area is completed. The image dataset consists of multiple images from different angles. Next, image fusion is performed on the image dataset based on an image fusion strategy. First, the multiple images from different angles are registered to ensure precise spatial alignment. Registration can be achieved through feature matching, transformation calculation, etc. Then, based on the image registration result, image fusion technology is used to merge the multiple images into a single comprehensive image. Common image fusion techniques include weighted averaging, pyramid fusion, and multi-scale fusion, which can be set according to the actual situation, resulting in a fused cladding image.

[0034] Convolutional Neural Networks (CNNs) are deep learning models specifically designed to process data with a grid structure, such as images, audio, and text. By using components like convolutional layers, pooling layers, and fully connected layers, CNNs can automatically learn data features and be used for tasks such as classification, recognition, and detection. A preset contour extraction channel is constructed based on a CNN. This preset contour extraction channel is a CNN model that can be iteratively optimized in machine learning, obtained through supervised training using a training dataset. The input data for the preset contour extraction channel is the overlay image, and the output is the overlay contour. Then, a sample training dataset is collected online to supervise the training of the preset contour extraction channel until a preset contour extraction channel that meets preset convergence constraints is obtained. These preset convergence constraints can be set according to actual needs, such as setting a threshold for expected output accuracy or expected number of training iterations. Next, the fused overlay image is input into the preset contour extraction channel, and the overlay contour is output. By constructing a preset contour extraction channel based on a CNN, the efficiency and accuracy of obtaining the overlay contour can be improved, thereby improving the accuracy of the subsequent first sub-stage overlay 3D model construction.

[0035] Further, using the overlay contour as a constraint, an overlay fitting analysis is performed on the point cloud dataset to obtain the overlay point cloud fitting result, and a first sub-stage overlay 3D model is constructed based on the overlay point cloud fitting result.

[0036] Furthermore, this application also includes the following steps: Using the overlay contour as a constraint, the point cloud dataset is randomly fitted to obtain a first fitting result, and a first fitting convergence of the first fitting result is calculated, wherein the first fitting convergence is the ratio of the number of point cloud data falling within the overlay contour to the total number of point cloud data in the point cloud dataset; it is determined whether the first fitting convergence meets a preset convergence, wherein the preset convergence is set based on the image acquisition accuracy and point cloud acquisition accuracy; if not, the point cloud dataset is iteratively randomly fitted again using the overlay contour as a constraint until the preset convergence or a preset fitting number threshold is met, and the current point cloud fitting result is output; in the 3D visualization platform, 3D simulation modeling is performed based on the current point cloud fitting result to generate the first sub-stage overlay 3D model.

[0037] Specifically, the method for obtaining the first sub-stage cladding 3D model by performing cladding fitting analysis on the point cloud dataset with the cladding contour as a constraint is as follows: First, the point cloud dataset is randomly fitted with the cladding contour as a constraint, that is, the point cloud data in the point cloud dataset is randomly placed into the cladding contour to obtain a first fitting result; then, the first fitting convergence of the first fitting result is calculated, where the first fitting convergence is the ratio of the number of point cloud data falling into the cladding contour to the total number of point cloud data in the point cloud dataset. The higher the first fitting convergence, the better the matching degree between the fitted contour and the actual cladding contour; if the first fitting convergence is low, it indicates that there is a large difference between the randomly fitted contour and the actual cladding contour, and it is necessary to reselect points for fitting.

[0038] The image acquisition accuracy of the image sensor and the point cloud acquisition accuracy of the laser point cloud device in the dual-source monitoring device are obtained. A preset convergence degree is then set based on these two accuracy values. This preset convergence degree is directly proportional to the image acquisition accuracy and the point cloud acquisition accuracy; that is, the higher the image acquisition accuracy and the higher the point cloud acquisition accuracy, the larger the corresponding preset convergence degree. Next, it is determined whether the first fitting convergence degree is greater than or equal to the preset convergence degree. If so, the first fitting result is output, and three-dimensional simulation modeling is performed based on the first fitting result to generate the three-dimensional model of the first sub-stage overlay.

[0039] If not, i.e., the first fitting convergence is less than the preset convergence, then the point cloud dataset is iteratively and randomly fitted again using the overlay contour as a constraint until the preset convergence is met, and the current point cloud fitting result is output; or when the number of iterative fittings equals the preset fitting number threshold, the current point cloud fitting result is output. Finally, in the 3D visualization platform, 3D simulation modeling is performed based on the current point cloud fitting result to generate the first sub-stage overlay 3D model.

[0040] Furthermore, step six of this application includes: The second sub-stage coating deviation dataset includes a thickness deviation dataset, a width deviation dataset, and a shape deviation dataset. The mean values ​​of the thickness deviation dataset and the width deviation dataset are calculated to obtain the mean thickness deviation and the mean width deviation, respectively. To satisfy the mean thickness deviation, the mean width deviation, and the mean shape deviation dataset, a first-correction analysis is performed on the predetermined powder feed rate based on the predetermined laser power and the predetermined scanning speed to obtain a first-corrected powder feed rate. A second-correction analysis is then performed on the first-corrected powder feed rate based on the real-time powder quantity in the powder cylinder to output the optimized powder feed rate.

[0041] Specifically, the second sub-stage coating deviation dataset includes a thickness deviation dataset, a width deviation dataset, and a shape deviation dataset. The mean values ​​of the thickness deviation dataset and the width deviation dataset are then calculated to obtain the mean thickness deviation and mean width deviation. Next, to satisfy the mean thickness deviation, the mean width deviation, and the shape deviation dataset, a first-correction analysis is performed on the predetermined powder feed rate based on the predetermined laser power and the predetermined scanning speed to obtain a first-correction powder feed rate. Furthermore, the real-time powder quantity in the powder cylinder is acquired, where the real-time powder quantity is the powder quantity at the start of the second sub-stage processing node. Then, a second-correction analysis is performed on the first-correction powder feed rate based on the real-time powder quantity. There is a correlation between the powder quantity in the powder cylinder and the powder feed rate; as the powder quantity in the powder cylinder decreases, the ability of the negative pressure device to absorb powder gradually decreases, leading to a reduction in the conveying volume. The optimized powder feed rate is obtained based on the results of the second-correction analysis.

[0042] By performing a first-correction analysis on the predetermined powder feeding flow rate based on the coating deviation dataset, a first-correction powder feeding flow rate is obtained. Then, a second-correction analysis is performed on the first-correction powder feeding flow rate based on the real-time powder quantity in the powder cylinder to obtain the optimized powder feeding flow rate. This can avoid the interference caused by the real-time powder quantity in the powder cylinder on the powder feeding flow rate while meeting the processing deviation of the previous stage, thereby further improving the accuracy and rationality of the optimized powder feeding flow rate.

[0043] Furthermore, this application also includes the following steps: The melting process log is invoked, and a sample dataset is obtained based on the melting process log. The sample data includes sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, sample scanning speed, and sample powder feed rate. The sample dataset is divided into Q equal parts, and Q random selections are made without replacement to construct the first sample set, and so on, until the Qth sample set. Using the sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, and sample scanning speed as inputs, and the sample powder feed rate as supervision, the BP neural network is trained using the Q sample sets to obtain multiple convergence correction units. The multiple convergence correction units are then fused to construct a convergence correction channel, where the output of the convergence correction channel is the mode of the multiple convergence correction units. The mean thickness deviation, mean width deviation, shape deviation dataset, predetermined laser power, and predetermined scanning speed are input into the convergence correction channel for a first correction analysis, and the first-corrected powder feed rate is output.

[0044] Specifically, firstly, the melt processing log is invoked. This log is a crucial recording tool used to document key parameters, equipment status, and anomalies during the processing. Next, data extraction is performed based on the melt processing log to obtain a sample dataset. This sample data includes sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, sample scanning speed, and sample powder flow rate. Then, the sample dataset is divided into Q equal parts, where Q is an integer greater than 5. The specific value of Q can be set according to actual conditions. From these Q sample datasets, the first sample set is constructed by randomly selecting Q times without replacement. The same method is used to obtain the second sample set, and so on, until the Qth sample set is obtained, resulting in Q sample sets.

[0045] Next, using the sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, and sample scanning speed as input data, and the sample powder delivery flow rate as supervised data, the BP neural network is trained and cross-validated using Q sample sets. For example, batch gradient descent and other optimization algorithms can be used for supervised training of the BP neural network to improve the training speed and accuracy of the model. Then, when the network output results tend to stabilize, the trained network is cross-validated using other sample sets to evaluate the model's generalization ability, resulting in multiple convergence correction units that meet the expected convergence constraints. Further, a convergence correction channel is constructed by fusing these multiple convergence correction units, where the output of the convergence correction channel is the mode of the multiple convergence correction units. This method can improve the accuracy and reliability of correcting the powder delivery flow rate output.

[0046] Finally, the average thickness deviation, the average width deviation, the shape deviation dataset, the predetermined laser power, and the predetermined scanning speed are input into the convergence correction channel for a first correction analysis, and the first correction powder feed rate is output.

[0047] Furthermore, this application also includes the following steps: Based on the principle of single variable analysis, a set of sample powder cartridge powder ratios and a set of sample powder delivery flow rates are collected, wherein the sample powder cartridge powder ratios and sample powder delivery flow rates are in one-to-one correspondence. Correlation analysis is performed on the sample powder cartridge powder ratios and sample powder delivery flow rates to determine the correlation influence curve. The real-time powder cartridge powder ratio is determined based on the real-time powder cartridge powder quantity, and the real-time powder cartridge powder ratio is matched with the correlation influence curve to determine the correlation influence coefficient. A secondary correction analysis is performed on the first-correction powder delivery flow rate based on the correlation influence coefficient to obtain the optimized powder delivery flow rate.

[0048] Specifically, based on the principle of single variable analysis, that is, without considering other variable factors, only considering the correlation between the powder ratio of the sample powder cylinder and the powder delivery flow rate, data is collected to obtain the set of sample powder cylinder powder ratios and the set of sample powder delivery flow rates. For example, data can be extracted by querying the powder delivery processing logs of the same type as the negative pressure device. Here, the sample powder cylinder powder ratio and the sample powder delivery flow rate are in one-to-one correspondence. The powder cylinder powder ratio refers to the weight ratio of the powder in the current powder cylinder to the powder in the powder cylinder when it is fully loaded.

[0049] Next, a correlation analysis is performed based on the sample powder cylinder powder ratio set and the sample powder delivery flow rate set. As the amount of powder in the powder cylinder decreases, the ability of the negative pressure device to pick up powder will gradually decrease, resulting in a reduction in delivery volume. For example, a two-dimensional coordinate system can be constructed, with the sample powder cylinder powder ratio as the X-axis and the sample powder delivery flow rate as the Y-axis. The sample powder cylinder powder ratio set and the sample powder delivery flow rate set are distributed in the two-dimensional coordinate system, and multiple distribution points are fitted and connected in series to generate a correlation curve.

[0050] Then, the real-time powder ratio of the powder cylinder is determined based on the real-time powder quantity of the powder cylinder, which is the weight ratio of the current powder quantity of the powder cylinder to the powder quantity of the powder cylinder when fully loaded. Next, the real-time powder ratio of the powder cylinder is matched with the correlation influence curve to obtain the correlation influence coefficient. Finally, the primary correction powder delivery flow rate is analyzed for secondary correction based on the correlation influence coefficient to obtain the optimized powder delivery flow rate. For example, when the real-time powder ratio of the powder cylinder is small, the primary correction powder delivery flow rate is increased according to the correlation influence coefficient.

[0051] Furthermore, step eight of this application includes: The optimization control analysis of the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage is executed sequentially to generate the optimization control strategy for the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage, where N is set based on the predetermined number of repetitions; the feeding control of the subsequent stages is executed sequentially according to the optimization control strategy for the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage.

[0052] Specifically, after executing the feeding control within the second sub-stage scanning time window according to the second sub-stage optimized control strategy, the same method used to obtain the second sub-stage optimized control strategy is used to sequentially execute the third sub-stage, fourth sub-stage, and so on up to the Nth sub-stage optimized control analysis, generating the third sub-stage optimized control strategy, the fourth sub-stage optimized control strategy, and so on up to the Nth sub-stage optimized control strategy, where N is the predetermined number of repetitions. Finally, the feeding control of subsequent stages is executed according to the third sub-stage optimized control strategy, the fourth sub-stage optimized control strategy, and so on up to the Nth sub-stage optimized control strategy.

[0053] In summary, the precision control method for feeding and unloading of CNC machine tools provided in this application has the following technical effects: 1. By acquiring images and point clouds using dual-source monitoring equipment during the repeated melting process of the workpiece, and then performing coating fitting analysis based on the image and point cloud datasets, a first-stage coating 3D model is generated. Simultaneously, a second-stage standard coating 3D model is constructed based on the processing indicators of the second stage. The mapping deviation between the first-stage coating 3D model and the second-stage standard coating 3D model is compared to obtain a second-stage coating deviation dataset. Then, aiming to meet the second-stage coating deviation dataset, the predetermined powder feeding flow rate is corrected and analyzed in conjunction with the real-time powder cylinder quantity to obtain an optimized powder feeding flow rate. Based on the optimized powder feeding flow rate and the second-stage scanning time window, a second-stage optimized control strategy is generated. Finally, the feeding control within the second-stage scanning time window is executed according to the second-stage optimized control strategy, and the same method is used to perform feeding optimization compensation in subsequent stages until the workpiece completes the melting process. This improves the precision and accuracy of stage cladding deviation analysis during the cladding process, enabling timely and effective compensation for stage cladding deviations, reducing or avoiding the accumulation of deviations during the cladding process, and ultimately improving the overall cladding quality of the workpiece.

[0054] 2. A first-corrected powder feeding flow rate is obtained by performing a correction analysis on the predetermined powder feeding flow rate based on the coating deviation dataset; then, a second-corrected powder feeding flow rate is obtained by performing a correction analysis on the first-corrected powder feeding flow rate based on the real-time powder quantity in the powder cylinder, thereby obtaining an optimized powder feeding flow rate. This can avoid the interference caused by the real-time powder quantity in the powder cylinder on the powder feeding flow rate while meeting the processing deviation of the previous stage, thus further improving the accuracy and rationality of the optimized powder feeding flow rate.

[0055] Example 2 Based on the refined feeding and unloading control method for CNC machine tools described in the foregoing embodiments, and using the same inventive concept, this application also provides a refined feeding and unloading control system for CNC machine tools. Please refer to the appendix. Figure 3 The system includes: The melt processing index acquisition module 11 is used to acquire the melt processing index of the target workpiece, wherein the melt processing index includes a preset coating thickness, a preset coating width, and a preset coating shape.

[0056] The melting processing module 12 is used to perform melting processing of the target workpiece according to a preset melting processing scheme, wherein the preset melting processing scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder feeding flow rate, and a predetermined number of repetitions.

[0057] The second-stage melting index selection module 13 is used to divide the melting processing index according to the predetermined number of repetitions, determine multiple stage melting indices, and select the second-stage melting index, wherein the second-stage melting index includes the second coating thickness, the second coating width, and the second coating shape.

[0058] The data acquisition module 14 is used to acquire data on the melting process of the target workpiece using a dual-source monitoring device within a preset period to obtain a dual-source monitoring dataset, wherein the dual-source monitoring dataset includes an image dataset and a point cloud dataset.

[0059] The overlay deviation dataset determination module 15 is used to perform overlay fitting analysis based on the image dataset and the point cloud dataset, generate a first sub-stage overlay 3D model, construct a second-stage standard overlay 3D model based on the second overlay thickness, the second overlay width and the second overlay shape, and compare the mesh deviation between the first sub-stage overlay 3D model and the second-stage standard overlay 3D model to determine the second sub-stage overlay deviation dataset.

[0060] The calibration analysis module 16 is used to monitor and acquire the real-time powder quantity of the powder cylinder in order to meet the second sub-stage coating deviation dataset. It combines the real-time powder quantity of the powder cylinder to perform calibration analysis on the predetermined powder delivery flow rate to obtain the optimized powder delivery flow rate.

[0061] The scanning time prediction module 17 is used to predict the scanning time of the second sub-stage based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, and to determine the scanning time window of the second sub-stage.

[0062] The feeding control module 18 is used to generate a second sub-stage optimized control strategy based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and to execute the feeding control within the second sub-stage scanning time window according to the second sub-stage optimized control strategy.

[0063] Furthermore, the overlay deviation dataset determination module 15 in the system is also used for: A first data acquisition node is set based on the preset cycle, wherein the first data acquisition node is the end node of the first preset cycle; under the first data acquisition node, the dual-source monitoring device is used to acquire data on the melting process of the target workpiece to obtain an image dataset and a point cloud dataset, wherein the image dataset consists of multiple images from different angles; the image dataset is fused based on an image fusion strategy to generate a fused overlay image, and the fused overlay image is input into a preset contour extraction channel to obtain an overlay contour, wherein the preset contour extraction channel is constructed based on a convolutional neural network; using the overlay contour as a constraint, an overlay fitting analysis is performed on the point cloud dataset to obtain the three-dimensional overlay model of the first sub-stage.

[0064] Furthermore, the overlay deviation dataset determination module 15 in the system is also used for: Using the overlay contour as a constraint, the point cloud dataset is randomly fitted to obtain a first fitting result, and a first fitting convergence of the first fitting result is calculated, wherein the first fitting convergence is the ratio of the number of point cloud data falling within the overlay contour to the total number of point cloud data in the point cloud dataset; it is determined whether the first fitting convergence meets a preset convergence, wherein the preset convergence is set based on the image acquisition accuracy and point cloud acquisition accuracy; if not, the point cloud dataset is iteratively randomly fitted again using the overlay contour as a constraint until the preset convergence or a preset fitting number threshold is met, and the current point cloud fitting result is output; in the 3D visualization platform, 3D simulation modeling is performed based on the current point cloud fitting result to generate the first sub-stage overlay 3D model.

[0065] Furthermore, the correction analysis module 16 in the system is also used for: The second sub-stage coating deviation dataset includes a thickness deviation dataset, a width deviation dataset, and a shape deviation dataset. The mean values ​​of the thickness deviation dataset and the width deviation dataset are calculated to obtain the mean thickness deviation and the mean width deviation, respectively. To satisfy the mean thickness deviation, the mean width deviation, and the mean shape deviation dataset, a first-correction analysis is performed on the predetermined powder feed rate based on the predetermined laser power and the predetermined scanning speed to obtain a first-corrected powder feed rate. A second-correction analysis is then performed on the first-corrected powder feed rate based on the real-time powder quantity in the powder cylinder to output the optimized powder feed rate.

[0066] Furthermore, the correction analysis module 16 in the system is also used for: The melting process log is invoked, and a sample dataset is obtained based on the melting process log. The sample data includes sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, sample scanning speed, and sample powder feed rate. The sample dataset is divided into Q equal parts, and Q random selections are made without replacement to construct the first sample set, and so on, until the Qth sample set. Using the sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, and sample scanning speed as inputs, and the sample powder feed rate as supervision, the BP neural network is trained using the Q sample sets to obtain multiple convergence correction units. The multiple convergence correction units are then fused to construct a convergence correction channel, where the output of the convergence correction channel is the mode of the multiple convergence correction units. The mean thickness deviation, mean width deviation, shape deviation dataset, predetermined laser power, and predetermined scanning speed are input into the convergence correction channel for a first correction analysis, and the first-corrected powder feed rate is output.

[0067] Furthermore, the correction analysis module 16 in the system is also used for: Based on the principle of single variable analysis, a set of sample powder cartridge powder ratios and a set of sample powder delivery flow rates are collected, wherein the sample powder cartridge powder ratios and sample powder delivery flow rates are in one-to-one correspondence. Correlation analysis is performed on the sample powder cartridge powder ratios and sample powder delivery flow rates to determine the correlation influence curve. The real-time powder cartridge powder ratio is determined based on the real-time powder cartridge powder quantity, and the real-time powder cartridge powder ratio is matched with the correlation influence curve to determine the correlation influence coefficient. A secondary correction analysis is performed on the first-correction powder delivery flow rate based on the correlation influence coefficient to obtain the optimized powder delivery flow rate.

[0068] Furthermore, the feeding control module 18 in the system is also used for: The optimization control analysis of the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage is executed sequentially to generate the optimization control strategy for the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage, where N is set based on the predetermined number of repetitions; the feeding control of the subsequent stages is executed sequentially according to the optimization control strategy for the third sub-stage, the fourth sub-stage, and so on up to the Nth sub-stage.

[0069] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. The CNC machine tool feeding and unloading precision control method and specific examples in Embodiment 1 described above are also applicable to the CNC machine tool feeding and unloading precision control system described in this embodiment. Through the foregoing detailed description of the CNC machine tool feeding and unloading precision control method, those skilled in the art can clearly understand the CNC machine tool feeding and unloading precision control system described in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As for the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be referred to in the method section.

[0070] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0071] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for precise control of material feeding and unloading in a CNC machine tool, characterized in that, The method includes: Obtain the melt processing parameters of the target workpiece, wherein the melt processing parameters include a preset coating thickness, a preset coating width, and a preset coating shape; The target workpiece is melt-processed according to a preset melting process scheme, wherein the preset melting process scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder feed rate, and a predetermined number of repetitions; The melting process index is divided according to the predetermined number of repetitions to determine multiple stages of melting indexes, and a second stage melting index is selected, wherein the second stage melting index includes a second coating thickness, a second coating width, and a second coating shape; Within a preset period, data is collected on the melting process of the target workpiece using a dual-source monitoring device to obtain a dual-source monitoring dataset, wherein the dual-source monitoring dataset includes an image dataset and a point cloud dataset. Based on the image dataset and the point cloud dataset, a coating fitting analysis is performed to generate a first sub-stage coating 3D model. A second-stage standard coating 3D model is constructed based on the second coating thickness, the second coating width, and the second coating shape. The mesh deviation of the first sub-stage coating 3D model and the second-stage standard coating 3D model is compared to determine the second sub-stage coating deviation dataset. The real-time powder quantity of the powder cylinder is monitored and obtained in order to meet the coating deviation dataset of the second sub-stage. The predetermined powder delivery flow rate is then corrected and analyzed in conjunction with the real-time powder quantity of the powder cylinder to obtain the optimized powder delivery flow rate. Based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, the scanning time of the second sub-stage is predicted, and the scanning time window of the second sub-stage is determined. The second sub-stage optimization control strategy is generated based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and the feeding control within the second sub-stage scanning time window is executed according to the second sub-stage optimization control strategy.

2. The method according to claim 1, characterized in that, The feeding control within the second sub-stage scanning time window is executed according to the second sub-stage optimized control strategy, and then the following is also included: The optimization control analysis is executed sequentially for the third sub-stage, the fourth sub-stage, and up to the Nth sub-stage, generating optimization control strategies for the third sub-stage, the fourth sub-stage, and up to the Nth sub-stage, where N is set based on the predetermined number of repetitions. The feeding control of subsequent stages is executed sequentially according to the third sub-stage optimization control strategy, the fourth sub-stage optimization control strategy, and so on up to the Nth sub-stage optimization control strategy.

3. The method according to claim 1, characterized in that, Based on the image dataset and the point cloud dataset, overlay fitting analysis is performed to generate a first-stage overlay 3D model, including: A first data acquisition node is set based on the preset period, wherein the first data acquisition node is the end node of the first preset period; At the first data acquisition node, the dual-source monitoring device is used to collect data on the melting process of the target workpiece to obtain an image dataset and a point cloud dataset, wherein the image dataset consists of multiple images from different angles. The image dataset is fused based on an image fusion strategy to generate a fused overlay image. The fused overlay image is then input into a preset contour extraction channel to obtain the overlay contour. The preset contour extraction channel is constructed based on a convolutional neural network. Using the overlay contour as a constraint, an overlay fitting analysis is performed on the point cloud dataset to obtain the three-dimensional overlay model of the first sub-stage.

4. The method according to claim 3, characterized in that, Using the overlay contour as a constraint, an overlay fitting analysis is performed on the point cloud dataset to obtain the first sub-stage overlay 3D model, including: Using the overlay contour as a constraint, the point cloud dataset is randomly fitted to obtain a first fitting result, and the first fitting convergence of the first fitting result is calculated, wherein the first fitting convergence is the ratio of the number of point cloud data falling within the overlay contour to the total number of point cloud data in the point cloud dataset. Determine whether the first fitting convergence meets the preset convergence, wherein the preset convergence is set based on the image acquisition accuracy and the point cloud acquisition accuracy; If not, the point cloud dataset is iteratively and randomly fitted again using the overlay contour as a constraint until the preset convergence or preset fitting number threshold is met, and the current point cloud fitting result is output. Within the 3D visualization platform, 3D simulation modeling is performed based on the current point cloud fitting results to generate the first sub-stage overlay 3D model.

5. The method according to claim 1, characterized in that, To meet the requirements of the second sub-stage coating deviation dataset, the predetermined powder delivery flow rate is corrected and analyzed in conjunction with the real-time powder cylinder powder quantity to obtain an optimized powder delivery flow rate, including: The second sub-stage coating deviation dataset includes a thickness deviation dataset, a width deviation dataset, and a shape deviation dataset. The mean values ​​of the thickness deviation dataset and the width deviation dataset are calculated respectively to obtain the mean thickness deviation and the mean width deviation. To satisfy the average thickness deviation, the average width deviation, and the shape deviation dataset, a correction analysis is performed on the predetermined powder feed rate based on the predetermined laser power and the predetermined scanning speed to obtain a corrected powder feed rate. The powder delivery flow rate is then subjected to a secondary correction analysis based on the real-time powder quantity in the powder cylinder, and the optimized powder delivery flow rate is output.

6. The method according to claim 5, characterized in that, To obtain a corrected toner delivery flow rate, including: The melt processing log is invoked, and a sample dataset is obtained based on the melt processing log. The sample data includes sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, sample scanning speed, and sample powder flow rate. The sample dataset is divided into Q equal parts, and Q random selections are made without replacement to construct the first sample set, and so on, until the Qth sample set is obtained. Using the sample thickness deviation, sample width deviation, sample shape deviation, sample laser power, and sample scanning speed as inputs, and the sample powder delivery flow rate as supervision, a BP neural network is trained using Q sample sets to obtain multiple convergence correction units. A convergence correction channel is constructed by fusing the multiple convergence correction units, wherein the output of the convergence correction channel is the mode of the multiple convergence correction units. The average thickness deviation, the average width deviation, the shape deviation dataset, the predetermined laser power, and the predetermined scanning speed are input into the convergence correction channel for a first correction analysis, and the first correction powder feed rate is output.

7. The method according to claim 5, characterized in that, A secondary correction analysis is performed on the primary correction powder delivery flow rate based on the real-time powder quantity in the powder cylinder, including: Based on the principle of single variable analysis, the sample powder cylinder powder ratio set and the sample powder delivery flow rate set are collected and obtained, wherein the sample powder cylinder powder ratio and the sample powder delivery flow rate are in one-to-one correspondence. Based on the sample powder ratio set and the sample powder delivery flow rate set, a correlation influence analysis was performed to determine the correlation influence curve; Based on the real-time powder quantity in the powder cartridge, the real-time powder ratio in the powder cartridge is determined, and the real-time powder ratio in the powder cartridge is matched with the correlation influence curve to determine the correlation influence coefficient. The optimized powder delivery flow rate is obtained by performing a secondary correction analysis on the first-correction powder delivery flow rate based on the correlation influence coefficient.

8. A precision control system for feeding and unloading materials in a CNC machine tool, characterized in that, The system comprises: steps for implementing the method according to any one of claims 1 to 7, wherein the system includes: The melt processing index acquisition module is used to acquire the melt processing index of the target workpiece, wherein the melt processing index includes a preset coating thickness, a preset coating width, and a preset coating shape; The melting processing module is used to perform melting processing of the target workpiece according to a preset melting processing scheme, wherein the preset melting processing scheme includes a predetermined laser power, a predetermined scanning speed, a predetermined powder feeding flow rate, and a predetermined number of repetitions; The second-stage melting index selection module is used to divide the melting processing index according to the predetermined number of repetitions, determine multiple stage melting indices, and select the second-stage melting index, wherein the second-stage melting index includes the second coating thickness, the second coating width, and the second coating shape. The data acquisition module is used to acquire data on the melting process of the target workpiece using a dual-source monitoring device within a preset period, and obtain a dual-source monitoring dataset, wherein the dual-source monitoring dataset includes an image dataset and a point cloud dataset. The overlay deviation dataset determination module is used to perform overlay fitting analysis based on the image dataset and the point cloud dataset, generate a first sub-stage overlay 3D model, construct a second-stage standard overlay 3D model based on the second overlay thickness, the second overlay width and the second overlay shape, and compare the mesh deviation between the first sub-stage overlay 3D model and the second-stage standard overlay 3D model to determine the second sub-stage overlay deviation dataset. The calibration analysis module is used to monitor and acquire the real-time powder quantity of the powder cylinder in order to meet the second sub-stage coating deviation dataset. It combines the real-time powder quantity of the powder cylinder to perform calibration analysis on the predetermined powder delivery flow rate to obtain the optimized powder delivery flow rate. The scanning time prediction module is used to predict the scanning time of the second sub-stage based on the predetermined scanning speed and the mapped workpiece area of ​​the first sub-stage, and to determine the scanning time window of the second sub-stage. The feeding control module is used to generate a second sub-stage optimized control strategy based on the optimized powder feeding flow rate and the second sub-stage scanning time window, and to execute the feeding control within the second sub-stage scanning time window according to the second sub-stage optimized control strategy.

Images