Method for high-power laser fuse pool synchronous intelligent shaping

Abstract

The application provides a kind of high-power laser fuse pool synchronous intelligent shaping method of deposition, it is related to additive manufacturing technical field, comprising: S1 according to the parameters of the piece to be printed, using three-dimensional software to model;S2 determine the material and forming process parameters of the piece to be printed;S3 utilize high-power laser to the material of current layer is deposited with fuse, adopts CCD camera to melt pool topography and surrounding topography real-time three-dimensional contour scanning obtains the topography, depth, width and temperature information of melt pool;S4 according to the optimal heating path of small power laser obtained by identification algorithm, and real-time control the working power, working position and working time of small power laser;S5 using small power laser according to heating path to melt pool is remelted and shaped;S6 repeat steps S3-S5, complete the additive manufacturing process of each layer, until the printed piece is obtained.The application can reduce splashes, control the defects of printed piece and stabilize melt pool.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of additive manufacturing, and particularly relates to a method for synchronous intelligent shaping of a high-power laser wire deposition molten pool. BACKGROUND

[0002] Additive manufacturing technology is a manufacturing method for directly manufacturing parts based on the discrete-accumulation principle and driven by three-dimensional data of parts, and a heat source commonly used in the additive manufacturing process of metal materials is an electric arc, an electron beam or a laser. Different from MIG electric arc wire, which causes problems such as coarse internal grains of a printed part and unstable molten pool, in the high-power laser wire high-efficiency additive manufacturing process, metal spatter and manufacturing defects are often caused by the high-efficiency wire deposition process, thereby affecting the quality of the printed part. Patent No. CN 115008017A discloses a MIG electric arc double-wire low-heat-input additive manufacturing method assisted by a scanning laser for shaping a molten pool, which uses MIG electric arc as a main heat source to form a molten pool, and uses an auxiliary cold wire and a scanning laser to shape the molten pool, so as to improve coarse columnar crystals of a formed part and reduce the performance difference of an additive part. However, the molten pool shaping technology is not intelligent enough. SUMMARY

[0003] Therefore, the application provides a method for synchronous intelligent shaping of a high-power laser wire deposition molten pool, which uses a CCD detection method and an identification algorithm to monitor and feedback the molten pool in real time, and uses a low-power laser to shape the molten pool, so as to reduce spatter caused by high-efficiency wire deposition, control defects of a printed part and stabilize the molten pool.

[0004] The technical scheme of the application is as follows:

[0005] The application provides a method for synchronous intelligent shaping of a high-power laser wire deposition molten pool, which includes the following steps:

[0006] S1. According to parameters of a to-be-printed part, a three-dimensional software is used to model, and the model is imported into a computer software to slice and generate an additive manufacturing path of each layer;

[0007] S2. According to forming requirements of the to-be-printed part, materials and forming process parameters of the to-be-printed part are determined;

[0008] S3. A high-power laser is used to perform wire deposition on materials of a current layer, and a CCD camera is used to perform real-time three-dimensional contour scanning on a molten pool appearance and a surrounding appearance to obtain information such as the appearance, depth, width and temperature of the molten pool;

[0009] S4. An optimal heating path of a low-power laser is obtained according to an identification algorithm, and working power, working position and working time of the low-power laser are adjusted in real time;

[0010] S5 uses a low-power laser to remelt and reshape the molten pool according to the heating path;

[0011] S6 repeats steps S3-S5 to complete the additive manufacturing process of each layer until the printed part is obtained.

[0012] On the basis of the above technical solution, preferably, step S1 comprises:

[0013] S11 obtains the parameters of the part to be printed, including size, shape and material;

[0014] S12 uses three-dimensional software to establish a three-dimensional model according to the parameters of the part to be printed;

[0015] S13 imports the model into computer software and performs slicing operation on the model to decompose the model into a series of planar layers;

[0016] S14 for each planar layer, the computer software generates an additive manufacturing path according to the parameters of the part to be printed and the characteristics of the material, each additive manufacturing path specifies the way the laser moves on each layer to achieve the required shape and structure.

[0017] On the basis of the above technical solution, preferably, step S2 comprises:

[0018] S21 obtains the forming requirements of the part to be printed, including size, shape, structure and surface quality;

[0019] S22 selects the material of the part to be printed according to the forming requirements of the part to be printed;

[0020] S23 determines the forming process parameters of the part to be printed according to the forming requirements of the part to be printed and the material of the part to be printed, including preheating temperature, high-power laser scanning speed, wire feeding speed and scanning spacing;

[0021] S24 tests and adjusts the forming process parameters of the part to be printed to verify the feasibility of the forming process parameters, and after verification, the final forming process parameters are obtained.

[0022] On the basis of the above technical solution, preferably, step S3 comprises:

[0023] S31 installs and fixes the CCD camera, aligns the CCD camera to the proofing area, and fixes it on the side wall of the printer cavity;

[0024] S32 during the process of laser fusing deposition on the current layer of material, the CCD camera real-time fast collects images of the molten pool and its periphery, and uses image processing algorithm to extract the molten pool topography, depth, width and temperature information;

[0025] S33 imports the molten pool shape and the surrounding shape collected by the CCD camera into the three-dimensional software, uses the three-dimensional software to model the three-dimensional contour data of the molten pool shape and the surrounding shape, and obtains the three-dimensional graph of the molten pool shape, depth, width and temperature;

[0026] S34 extracts the molten pool shape, depth, width and temperature features in the three-dimensional graph and performs recognition by using a recognition algorithm, and obtains the information of the molten pool shape, depth, width and temperature according to the recognition result.

[0027] On the basis of the above technical solution, preferably, the step S3 comprises:

[0028] S31 installs and fixes the CCD camera, aligns the CCD camera to the proofing area, and fixes the CCD camera on the side wall of the printer cavity;

[0029] S32 collects images of the molten pool and the surrounding area in real time by using the CCD camera during the process of laser deposition of the current layer, and extracts the molten pool shape and the surrounding shape by using an image processing algorithm;

[0030] S33 imports the molten pool shape and the surrounding shape collected by the CCD camera into the three-dimensional software, uses the three-dimensional software to model the three-dimensional contour data of the molten pool shape and the surrounding shape, and obtains the three-dimensional graph of the molten pool shape and the surrounding shape;

[0031] S34 extracts the molten pool features in the three-dimensional graph and performs recognition by using a recognition algorithm, and obtains the position information and the contour information of the molten pool according to the recognition result.

[0032] On the basis of the above technical solution, preferably, the step S34 comprises:

[0033] constructs a target recognition model, obtains an image sample set in the printing cavity, classifies and trains the target recognition model by using the image sample set, modifies the configuration parameters of the target recognition model after the training is completed, makes the target recognition model learn to recognize the molten pool, and obtains the trained target recognition model;

[0034] loads the three-dimensional graph into the trained target recognition model, obtains a feature graph and a prediction result by forward propagation calculation, and the feature graph contains the molten pool shape, depth, width and temperature feature information;

[0035] performs post-processing according to the prediction result, filters according to a confidence threshold, removes the prediction results lower than the confidence threshold, uses a non-maximum suppression algorithm to merge the overlapping boundary boxes, and obtains the final recognition result;

[0036] outputs the recognition result, and obtains the molten pool shape, depth, width and temperature information.

[0037] Based on the technical scheme, preferably, step S4 comprises:

[0038] S41 generates a heating path of the low-power laser based on the profile information and surface quality of the molten pool;

[0039] S42 optimizes the working power, working position and working time of the low-power laser in real time according to the detected characteristics of the molten pool.

[0040] Based on the technical scheme, preferably, step S41 comprises:

[0041] Intelligently generating a heating path of the low-power laser according to the profile information and surface quality of the molten pool;

[0042] For places with splashes on the surface or uneven molten pool profile, they are identified as a target point;

[0043] The area with the most target points is taken as the starting point for the low-power laser to start working.

[0044] Based on the technical scheme, preferably, step S42 comprises:

[0045] According to the working position of the low-power laser itself, and the width and profile information of the molten pool, the working power and working time of the low-power laser are automatically optimized.

[0046] Based on the technical scheme, preferably, step S5 comprises:

[0047] The low-power laser is used to shape the molten pool, and a CCD camera is used to scan the bottom of the molten pool and identify the profile of the molten pool and calculate the aspect ratio. When the profile of the molten pool has no concave-convex points and the aspect ratio of the bottom reaches the stopping condition, the low-power laser is stopped, and the remelting shaping is completed.

[0048] The method of the present application has the following beneficial effects compared with the prior art:

[0049] (1) The present application uses three-dimensional modeling software and slicing software, uses a CCD camera to perform real-time three-dimensional profile scanning on the molten pool morphology and the surrounding morphology, and through an identification algorithm, the optimal heating path of the low-power laser can be obtained. According to the optimal path obtained by the identification algorithm, the working power, working position and working time of the low-power laser can be controlled in real time. The low-power laser is used to remelt and shape the molten pool according to the heating path. The present application combines three-dimensional software, image recognition algorithm and laser technology to realize an efficient, accurate and real-time controlled additive manufacturing process, thereby obtaining a high-quality printed part.

[0050] (2) The present application can realize comprehensive analysis and identification of the molten pool shape, depth, width and temperature by combining the CCD camera with three-dimensional software and recognition algorithms, and can obtain more accurate and detailed molten pool information, helping to optimize the laser wire deposition process. At the same time, the shape, depth, width and temperature information of the molten pool can be displayed by three-dimensional graphics, which can provide intuitive and comprehensive visual presentation, helping to better understand and analyze the characteristics of the molten pool, so as to make relevant decisions and adjustments;

[0051] (3) The present application can realize the identification of the molten pool shape, depth, width and temperature characteristics by using recognition algorithms, and the recognition algorithms have high target detection capability, which can simultaneously predict the position and category of multiple targets in one forward propagation process. This makes the identification of the molten pool shape, depth, width and temperature characteristics can be completed in a short time and has high accuracy. The use of recognition algorithms for molten pool feature identification can realize automatic processing and reduce the need for manual intervention. This makes the analysis and identification process of molten pool characteristics more convenient and efficient. By identifying the molten pool shape, depth, width and temperature characteristics, a comprehensive analysis result of the molten pool can be obtained. Real-time monitoring of molten pool characteristics using recognition algorithms can timely detect and identify abnormal conditions of the molten pool. This helps to take timely measures to avoid potential quality problems and production risks;

[0052] (4) The present application can intelligently generate a heating path for a small-power laser based on the contour information and surface quality of the molten pool, which can shape and repair the molten pool in a targeted manner, reduce heating time and energy consumption, realize automatic shaping and repair of the molten pool, reduce energy consumption and material waste during the heating process, and can help improve processing efficiency, reduce production cost, and reduce the need for manual intervention, thereby improving the automation level of production;

[0053] (5) The present application can optimize and control the working position of the small-power laser in real time according to the characteristics of the molten pool, which can ensure that the laser can accurately cover the entire area of the molten pool, thereby improving the processing precision. The working power and working time of the small-power laser can be optimized and controlled in real time according to the characteristics of the molten pool, which can ensure that the laser reaches the required heating effect in the shortest time, thereby improving the processing efficiency. Real-time optimization and control of the working power and working time of the small-power laser can avoid excessive heating and energy waste, thereby reducing energy consumption;

[0054] (6) The present application can realize accurate shaping effect and stop condition judgment by using small-power laser to shape the molten pool and combining CCD camera scanning and contour recognition, which has the effects of improving shaping precision, reducing energy consumption, improving production efficiency and reducing material loss. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only aim to some of the embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort.

[0056] Figure 1 The method flowchart of the embodiments of the present application is shown in the following.

[0057] Figure 2 The printing process description diagram of the embodiments of the present application is shown in the following. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without any creative effort belong to the protection scope of the present application.

[0059] As shown in the following, Figure 1 and Figure 2 The present application provides a method for large-power laser fused deposition molten pool synchronous intelligent shaping, which comprises the following steps:

[0060] S1, according to the parameters of the to-be-printed part, using three-dimensional software to model, and importing the model into the computer software to slice and generate the additive manufacturing path of each layer;

[0061] S2, according to the forming requirements of the to-be-printed part, determining the material and forming process parameters of the to-be-printed part;

[0062] S3, using large-power laser to perform fused deposition on the material of the current layer, and using CCD camera to perform real-time three-dimensional contour scanning on the molten pool morphology and the surrounding morphology to obtain the morphology, depth, width and temperature information of the molten pool;

[0063] S4, obtaining the optimal heating path of the small-power laser according to the recognition algorithm, and real-time regulating and controlling the working power, working position and working time of the small-power laser;

[0064] S5, using the small-power laser to perform remelting shaping on the molten pool according to the heating path;

[0065] S6, repeating steps S3-S5 to complete the additive manufacturing process of each layer until the to-be-printed part is obtained.

[0066] Specifically, in an embodiment of the present application, step S1 comprises:

[0067] S11 Obtain the parameters of the object to be printed, including size, shape, and material.

[0068] First, the parameters of the object to be printed, such as size, shape, material, etc., need to be obtained. These parameters will be used in the modeling and slicing process.

[0069] S12 Use three-dimensional software to establish a three-dimensional model based on the parameters of the object to be printed.

[0070] Using professional three-dimensional modeling software, a precise three-dimensional model is created based on the parameters of the object to be printed. The modeling process may involve drawing basic geometric shapes, combining complex components, adding details, etc. The modeling software usually provides various tools and functions to help users create accurate models. The three-dimensional software used in this embodiment is Solidworks software.

[0071] S13 Import the model into computer software and perform slicing operations on the model to decompose the model into a series of flat layers.

[0072] After completing the three-dimensional modeling, the model is imported into the computer-aided design / manufacturing software CAD. The CAD software can read and edit the three-dimensional model and prepare for subsequent slicing and path generation.

[0073] In the CAD software, the three-dimensional model is sliced. Slicing is the process of decomposing a three-dimensional model into a series of flat layers. Each layer represents a printing operation that the printer needs to perform on that layer.

[0074] S14 For each flat layer, the computer software generates additive manufacturing paths based on the parameters of the object to be printed and the characteristics of the material. Each additive manufacturing path specifies the way the laser moves on each layer to achieve the desired shape and structure. The generation of the path takes into account factors such as the shape of the printed object, support structure, filling density, etc., to ensure the quality and stability of the printed object.

[0075] Specifically, step S1 can also include additive manufacturing path optimization, as follows:

[0076] The shortest path algorithm is used to generate the shortest path based on the position of the laser and the target position, and the additive manufacturing path is optimized in terms of path length.

[0077] For example, the shortest path algorithm is Dijkstra's algorithm or A* algorithm to find the shortest path. These algorithms can consider the shortest path between the current position of the print head and the target position, and generate a path that passes through the fewest points.

[0078] By analyzing the shape and continuity of the object to be printed, the best printing order is obtained, and the additive manufacturing path is optimized in terms of printing order.

[0079] In some cases, different parts of the printed object can be printed in different orders to reduce the number and time of movements of the print head. Path optimization can be achieved by analyzing the geometry and connection relationship of the printed object to determine an optimal printing order. For example, the connection part can be printed first, and then the main body part is printed to reduce the movement distance of the print head.

[0080] Through the above steps, an efficient and accurate additive manufacturing process can be achieved. Using three-dimensional modeling software for modeling operations can accurately convert the shape and size of the printed object into a three-dimensional model, ensuring the accuracy and precision of the printed object. By slicing the three-dimensional model, the model is decomposed into a series of planar layers, which can achieve more precise control and adjustment to meet design requirements. The computer software generates additive manufacturing paths based on the parameters of the printed object and the material characteristics, taking into account factors such as shape, support structure and filling density to achieve the best printing effect and quality. By using computer software for slicing and path generation, an automated and efficient additive manufacturing process can be achieved, improving production efficiency and reducing costs.

[0081] Specifically, in an embodiment of the present application, step S2 comprises:

[0082] S21 obtains the forming requirements of the printed object, including size, shape, structure and surface quality.

[0083] In the additive manufacturing process, the forming requirements of the printed object are first obtained. These requirements include the size (length, width, height) of the printed object, the shape (such as cube, cylinder, complex curved surface, etc.), the structure (such as whether it needs an internal cavity or support structure) and the surface quality requirements (such as smoothness, roughness, etc.). These requirements will provide a basis for subsequent material selection and determination of forming process parameters.

[0084] S22 selects the material of the printed object according to the forming requirements of the printed object.

[0085] Different materials have different characteristics, such as plastic has a lower melting point and plasticity, metal has a higher melting point and strength, ceramic has a higher hardness and high temperature resistance, etc. According to the requirements of the printed object, the appropriate material can ensure that the printed object has the required performance and quality. In this embodiment, metal is selected as the material of the printed object.

[0086] S23 determines the forming process parameters of the printed object according to the forming requirements of the printed object and the material of the printed object, including preheating temperature, high-power laser scanning speed, wire feeding speed and scanning interval.

[0087] According to the forming requirements of the to-be-printed piece and the selected material, appropriate forming process parameters are determined. These parameters include preheating temperature, high-power laser scanning speed, wire feeding speed, and scanning spacing. These parameters will directly affect the quality and performance of the printed piece. In this embodiment, titanium alloy is selected, the laser power is 1000-10000W, the scanning speed is 500-1200mm / s, the scanning spacing is 0.06-0.08mm, and the wire feeding speed is 1.5-2m / min. The reshaping laser power is 200-3000W, and the scanning speed is 500-1200mm / s.

[0088] S24 tests and adjusts the forming process parameters of the to-be-printed piece to verify the feasibility of the forming process parameters. After verification, the final forming process parameters are obtained.

[0089] After determining the material and forming process parameters, testing and adjustment are usually needed to verify their feasibility and effectiveness in actual printing. This can include printing sample pieces, conducting physical performance tests, and fine-tuning parameters based on test results.

[0090] Through the above steps, appropriate materials can be selected according to the forming requirements of the to-be-printed piece, and the best forming process parameters can be determined. Selecting appropriate materials and forming process parameters according to the forming requirements of the to-be-printed piece can ensure that the printed piece meets the requirements of size, shape, structure, and surface quality. Selecting appropriate materials according to the forming requirements of the to-be-printed piece can make the printed piece have the required performance characteristics, such as strength, hardness, and high-temperature resistance. By testing and adjusting the forming process parameters, the forming quality and performance of the printed piece can be optimized, and the printing efficiency and stability can be improved. By determining the final forming process parameters, quality control of the printed piece can be achieved, ensuring that each printed piece has the same dimensional accuracy, surface quality, and structural integrity.

[0091] Specifically, in a specific embodiment of the present application, step S3 includes:

[0092] S31 installs and fixes the CCD camera, aligns the CCD camera to the proofing area, and fixes it on the side wall of the printer cavity; this can ensure that the camera can accurately capture the information of the molten pool morphology and temperature.

[0093] The position and angle of the camera need to be adjusted according to the actual situation to ensure that the information of the molten pool morphology and the surrounding morphology can be accurately captured. Once the camera is installed and fixed, it will remain in a fixed position for real-time monitoring during the entire laser wire deposition process.

[0094] S32 During the laser melting deposition of the current layer of material, the CCD camera captures images of the molten pool and its surroundings in real-time. Image processing algorithms are used to extract the molten pool topography, depth, width, and temperature information.

[0095] During the laser melting deposition of the current layer of material, the CCD camera captures images of the molten pool and its surroundings in real-time. These images are typically captured in the form of a video stream and are updated rapidly at a high frame rate. The captured images are transmitted to a computer and processed using image processing algorithms. These algorithms can extract information about the molten pool topography and the surrounding topography, such as the shape, depth, width, and temperature of the molten pool. This step includes:

[0096] The CCD camera captures images in real-time during the laser melting deposition process by continuously taking rapid successive shots. Typically, the camera's capture frequency is matched to the laser scanning speed to ensure that the molten pool morphology changes for each layer are captured.

[0097] The captured images may be affected by noise, uneven lighting, and other factors, and need to be pre-processed. The pre-processing steps can include image denoising, grayscale equalization, background removal, and other methods to improve the accuracy and stability of subsequent image processing.

[0098] The image processing algorithms are used to extract the molten pool topography information. Edge detection algorithms such as the Canny algorithm can be used to detect the boundaries of the molten pool.

[0099] In addition to topography information, the depth and width of the molten pool can also be extracted from the images. This can be achieved by analyzing the grayscale changes of the molten pool. Typically, the depth and width of the molten pool have a certain relationship with their grayscale values, and can be extracted by grayscale threshold segmentation or grayscale gradient analysis.

[0100] The temperature of the molten pool is an important parameter in the laser melting deposition process. The temperature information can be extracted from the grayscale values in the image and the relationship between the grayscale values and the temperature of the molten pool. This can be achieved by establishing a grayscale-temperature calibration curve or using a thermal imager to assist.

[0101] S33 The molten pool topography and surrounding topography captured by the CCD camera are imported into the three-dimensional software, and the three-dimensional profile data modeling of the molten pool topography and surrounding topography is performed using the three-dimensional software, obtaining the three-dimensional graph of the molten pool topography, depth, width, and temperature.

[0102] First, the molten pool images captured by the CCD camera are imported into the three-dimensional software. This can be done by importing the image file into the software or directly obtaining the image data through the interface between the software and the camera.

[0103] Next, the molten pool image is processed and analyzed using 3D software. First, the image needs preprocessing, such as denoising and grayscale equalization, to improve the accuracy of subsequent processing. Then, the molten pool morphology can be modeled using tools or algorithms in the software. This can be achieved by extracting the molten pool contour from the image and converting it into 3D surface or point cloud data.

[0104] During the modeling process, the 3D model of the molten pool can be further refined based on the topographic information surrounding the pool. For example, the shape and size of the model can be adjusted according to the edge features and shape of the molten pool. Simultaneously, tools or algorithms within the software can be used to measure and model the depth and width of the molten pool.

[0105] Additionally, temperature data can be correlated with the 3D model of the molten pool based on its temperature information. This can be achieved by adding temperature attributes or color mapping to the model. This allows the temperature distribution of the molten pool to be visually displayed in a 3D diagram.

[0106] Ultimately, the 3D maps of the molten pool's morphology, depth, width, and temperature generated by 3D software can provide more intuitive and comprehensive information.

[0107] S34 uses a recognition algorithm to extract and identify the features of the molten pool in the 3D image, and obtains the location and contour information of the molten pool based on the recognition results.

[0108] In this embodiment, laser fused filament deposition (LMD) is a process for 3D printing that melts metal powder with a laser and deposits it onto the surface of a workpiece. In this step, the laser beam is focused at a specific location on the workpiece surface, melting the metal powder and depositing it into layers. The laser power and scanning speed control the formation and deposition rate of the molten pool. A CCD camera is a high-speed, high-resolution image acquisition device that can capture images of the workpiece surface in real time. During the LMD process, the CCD camera is used to perform 3D contour scanning of the molten pool morphology and its surrounding area. The camera can acquire clear images using appropriate light sources and filters.

[0109] Specifically, step S34 includes:

[0110] Construct a target recognition model, obtain an image sample set inside the printing cavity, use the image sample set to classify and train the target recognition model, modify the configuration parameters of the target recognition model after training, so that the target recognition model learns to recognize the melt pool, and obtain a trained target recognition model.

[0111] A specific example of obtaining a trained target recognition model by modifying configuration parameters is as follows:

[0112] (1) Since there is no information about the molten pool, metal wire, large / small power laser, etc. in the Yolo training model, the photos of the objects in the printing cavity are classified and learned using a convolutional neural network. Then, the Makefile in Yolo is modified, and then recompiled to enable Yolo to identify the molten pool alone;

[0113] (2) Use Yolo and Python to identify the objects in the three-dimensional graph. The three-dimensional graph in solidworks is automatically imported into the YOLO / darknet folder. The downloaded source code is decompressed and moved to the object-detection-yolo-opencv folder.

[0114] (3) Download two pre-trained model parameters in advance, yolov3-tiny.weights and yolov3.weights, and place them in the object-detection-yolo-opencv folder.

[0115] (4) According to the instructions in readme.md, identify the molten pool in the three-dimensional profile graph in YOLO / darknet.

[0116] Load the three-dimensional graph into the trained object recognition model, and calculate the feature map and prediction result through forward propagation. The feature map contains the molten pool morphology, depth, width, and temperature feature information, and the prediction result contains the molten pool position and bounding box information.

[0117] According to the prediction result, post-processing is performed. According to the confidence threshold, remove the prediction results below the confidence threshold, and use the non-maximum suppression algorithm to merge the overlapping bounding boxes to get the final recognition result.

[0118] Control the CCD image lens to move with the molten pool, and try to reduce the number of objects that need to be identified in the actual scene by Yolo algorithm. Select a higher confidence threshold to improve accuracy.

[0119] Output the recognition result, and get the position information and contour information of the molten pool.

[0120] The recognition algorithm of the embodiment has high-efficiency target detection capability, which can simultaneously predict the positions and categories of multiple targets in one forward propagation process. This enables the recognition of the morphology, depth, width, and temperature characteristics of the molten pool to be completed in a relatively short time with high accuracy. The recognition of the molten pool characteristics using the recognition algorithm can realize automatic processing and reduce the need for manual intervention. This makes the analysis and recognition process of the molten pool characteristics more convenient and efficient. By recognizing the morphology, depth, width, and temperature characteristics of the molten pool, a comprehensive analysis result of the molten pool can be obtained. Real-time monitoring of the molten pool characteristics using the recognition algorithm can timely discover and identify abnormal conditions of the molten pool. This helps to take timely measures to avoid potential quality problems and production risks.

[0121] Specifically, in an embodiment of the present application, step S4 comprises:

[0122] S41 generates a heating path of the low-power laser based on the contour information and surface quality of the molten pool.

[0123] Step S41 comprises:

[0124] Intelligently generating a heating path of the low-power laser based on the contour information and surface quality of the molten pool;

[0125] Identifying a place with splashes on the surface or an uneven place of the molten pool contour as a target point;

[0126] Taking a region with the most target points as the starting point for the low-power laser to start working for optimal shaping.

[0127] An embodiment is used to illustrate step S41:

[0128] Extracting contour information of the molten pool: Through image processing algorithms, the contour information of the molten pool can be extracted from the molten pool image. This can be achieved through edge detection algorithms (such as Canny edge detection) or contour extraction algorithms (such as the findContours function in the OpenCV library). The extracted contour information can represent the shape and boundary of the molten pool.

[0129] Analyzing surface quality: Based on the contour information of the molten pool, the quality of the molten pool surface can be analyzed. This can be achieved by calculating the geometric features of the contour (such as area, perimeter, shape, etc.) or surface quality evaluation indicators (such as roughness, flatness, etc.). Analyzing the surface quality can help determine the flatness and quality of the molten pool.

[0130] Identifying target points: Based on the surface quality and unevenness of the molten pool contour, places with splashes on the surface or uneven places of the molten pool contour can be identified as target points. These target points may represent areas that need to be shaped and repaired.

[0131] Dense area selection: Based on the distribution of target points, the most dense area of target points can be found. This can be achieved through clustering algorithms (such as K-means clustering) or density estimation algorithms (such as DBSCAN). Finding the most dense area of target points can help determine the priority of shaping and repairing.

[0132] Optimal shaping path generation: The most dense area of target points is used as the starting point for the small power laser to start working. Based on the distance and connection between the starting point and other target points, the optimal heating path can be generated. This can be achieved through path planning algorithms (such as the shortest path algorithm, genetic algorithm, etc.).

[0133] S42 optimizes the working power, working position and working time of the small power laser in real time according to the detected molten pool characteristics.

[0134] According to the working position of the small power laser itself and the width and contour information of the molten pool, the working power and working time of the small power laser are automatically optimized.

[0135] A specific example is used to illustrate step S42:

[0136] Molten pool feature detection: First, the characteristics of the molten pool need to be detected. This can be achieved through image processing algorithms (such as edge detection, contour extraction, etc.) or sensors (such as infrared sensors, laser range finders, etc.) to obtain the width, contour information and other characteristics of the molten pool.

[0137] Small power laser working position optimization: According to the width and contour information of the molten pool, the working position of the small power laser can be optimized in real time. This can be achieved by controlling the scanning range and scanning speed of the laser. According to the width and contour information of the molten pool, the working position of the laser can be adjusted to the most suitable position to ensure that the laser can cover the entire molten pool area, thereby achieving more accurate heating effect.

[0138] Small power laser working power optimization: According to the characteristic information of the molten pool, the working power of the small power laser can be optimized in real time. This can be achieved by adjusting the output power of the laser. According to the width, contour information and material characteristics of the molten pool, the most suitable working power can be determined to ensure that the molten pool can reach the required temperature and melting state.

[0139] Small power laser working time optimization: According to the characteristic information of the molten pool, the working time of the small power laser can be optimized in real time. This can be achieved by controlling the heating time of the laser. According to the width, contour information and material characteristics of the molten pool, the most suitable working time can be determined to ensure that the molten pool can reach the required melting and shape repair effect.

[0140] The embodiment can optimize the working position of the low-power laser in real time according to the characteristics of the molten pool, ensure that the laser can accurately cover the entire area of the molten pool, improve the machining precision, optimize the working power and working time of the low-power laser in real time according to the characteristics of the molten pool, ensure that the laser can reach the required heating effect in the shortest time, improve the machining efficiency, and optimize the working power and working time of the low-power laser in real time, so as to avoid excessive heating and energy waste and reduce energy consumption.

[0141] Specifically, in an embodiment of the present application, step S5 comprises:

[0142] The low-power laser is used to reshape the molten pool, and a CCD camera is used to scan the bottom of the molten pool and identify the profile of the molten pool and calculate the aspect ratio. When the profile of the molten pool has no concave-convex points and the aspect ratio of the bottom reaches the stopping condition, the low-power laser is stopped, and the remelting shaping is completed.

[0143] In the remelting shaping process, the low-power laser is aligned with the molten pool, and the working power and working time of the laser are controlled so that it can heat and melt the material on the surface of the molten pool. The low-power laser can provide local heating and melting, so that the shape of the bottom of the molten pool is adjusted and flattened. At the same time, a CCD camera is installed below the molten pool, which scans the image of the bottom of the molten pool in real time through image acquisition and processing technology. Through image processing algorithms, the profile information of the bottom of the molten pool is extracted by processing the images collected by the CCD camera. Then, the aspect ratio of the bottom of the molten pool is calculated according to the profile information. The aspect ratio can reflect the shape and flatness of the bottom of the molten pool. If the profile of the bottom of the molten pool has no concave-convex points and the aspect ratio reaches the preset stopping condition, the working of the low-power laser is stopped, that is, the required shaping effect is achieved, and the low-power laser can be stopped. The advantage of this is that the shape and flatness of the bottom of the molten pool can be ensured to meet the requirements through real-time monitoring and adjustment. At the same time, using the low-power laser for shaping can avoid excessive heating and excessive shape change, and ensure the accuracy and stability of the shaping. In the embodiment, the aspect ratio of the bottom of the molten pool that reaches the stopping condition is 2.

[0144] After the single-layer printing process is completed, scanning is performed along the direction of the single-direction strip according to the printing path planning, and the material is printed layer by layer.

[0145] The above process is repeated to complete the entire printed part. After printing is completed, the shielding gas, the multi-laser machining robot system, and the additive manufacturing control system are turned off. After the substrate is completely cooled, the substrate and the additive printed part are taken out.

[0146] In the embodiment, when printing, some complex geometrical shapes can need support structures to keep stable, and long strip linear supports can be used, and after printing is completed, the long strip linear support structures are mechanically removed, and polishing, sand blasting and sintering and other processes are used to ensure the performance and surface quality of the printed part.

[0147] The above merely provides the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of high power laser fusion deposition melt pool synchronization intelligent shaping, characterized in that, Comprise: S1 according to the parameters of the to-be-printed piece, using three-dimensional software modeling, and importing the model into the computer software for slicing, generating each layer of additive manufacturing path; S2 according to the forming requirements of the to-be-printed piece to determine the material and forming process parameters of the to-be-printed piece; S3 using high-power laser on the current layer of material for fused deposition, using CCD camera for real-time three-dimensional contour scanning of the molten pool morphology and the surrounding morphology to obtain the morphology, depth, width and temperature information of the molten pool; Step S3 includes: S31 install and fix the CCD camera, align the CCD camera to the proofing area, and fix it on the side wall of the printer cavity; S32 in the process of laser fused deposition on the current layer of material, the CCD camera real-time fast acquisition image of the molten pool and its surrounding, and using image processing algorithm to extract the molten pool morphology, depth, width and temperature information; S33 the molten pool morphology and surrounding morphology collected by the CCD camera are imported into the three-dimensional software, and the three-dimensional contour data modeling of the molten pool morphology and surrounding morphology is carried out by using the three-dimensional software, to obtain the three-dimensional graph of the molten pool morphology, depth, width and temperature; S34 using recognition algorithm to extract the molten pool features in the three-dimensional graph and identify, according to the recognition result to obtain the position information and contour information of the molten pool; S4 according to the recognition algorithm to obtain the heating path of the optimal low-power laser, and real-time control the working power, working position and working time of the low-power laser; Step S4 includes: S41 based on the contour information and surface quality of the molten pool to generate the heating path of the low-power laser, wherein, According to the contour information of the molten pool, the quality of the molten pool surface is analyzed, the flatness and quality of the molten pool are determined; for the places with splashes on the surface or uneven molten pool contour, it is identified as a target point; the area with the most target points is taken as the starting point of the optimal shaping of the low-power laser, the priority of the shaping and repairing is determined, and the optimal heating path is generated by path planning algorithm according to the distance and connection relationship between the starting point and other target points; S5 using low-power laser according to the heating path to remelt and shape the molten pool; S6 repeat steps S3-S5, complete the additive manufacturing process of each layer, until the printed piece is obtained.

2. A method of high power laser fuse pool synchronous intelligent shaping as claimed in claim 1 wherein, Step S1 includes: S11 obtain the parameters of the to-be-printed piece, including size, shape and material; S12 using three-dimensional software to establish three-dimensional model according to the parameters of the to-be-printed piece; S13 import the model into the computer software, and slice the model to decompose the model into a series of plane layers; S14 for each plane layer, the computer software generates additive manufacturing path according to the parameters of the to-be-printed piece and the characteristics of the material, each additive manufacturing path specifies the way of laser movement on each layer to realize the required shape and structure.

3. A method of high power laser fusion deposition molten pool synchronization intelligent shaping as claimed in claim 1 wherein, Step S2 includes: S21 obtain the forming requirements of the to-be-printed piece, including size, shape, structure and surface quality; S22 according to the forming requirements of the to-be-printed piece, select the material of the to-be-printed piece; S23 determining forming process parameters of the part to be printed according to forming requirements of the part to be printed and material of the part to be printed, including preheating temperature, high-power laser scanning speed, wire feeding speed, scanning interval; S24 testing and adjusting the forming process parameters of the part to be printed to verify the feasibility of the forming process parameters, and obtaining final forming process parameters after verification.

4. The method of claim 1, wherein the method is a method of high power laser fuse deposition melt pool synchronization intelligent shaping, characterized in that, Step S34 includes: constructing a target recognition model, obtaining an image sample set in the printing cavity, classifying and training the target recognition model using the image sample set, modifying the configuration parameters of the target recognition model after training is completed, and making the target recognition model learn to recognize the molten pool to obtain a trained target recognition model; loading the three-dimensional graph into the trained target recognition model, calculating a feature map and a prediction result through forward propagation, and the feature map containing molten pool topography, depth, width, and temperature characteristic information; post-processing according to the prediction result, screening according to a confidence threshold, removing prediction results below the confidence threshold, using a non-maximum suppression algorithm to merge overlapping bounding boxes, and obtaining a final recognition result; outputting the recognition result to obtain position information and contour information of the molten pool.

5. A method of high power laser fusion deposition melt pool synchronization intelligent shaping as claimed in claim 1 wherein, Step S4 further includes: S42 optimizing and controlling the working power, working position, and working time of the low-power laser in real time according to the detected molten pool characteristics.

6. A method of high power laser fuse deposition melt pool synchronization intelligent shaping as claimed in claim 5 wherein, Step S42 includes: automatically optimizing the working power and working time of the low-power laser according to the working position of the low-power laser itself and the width and contour information of the molten pool.

7. A method of high power laser fuse pool synchronous intelligent shaping as claimed in claim 1 wherein, Step S5 includes: using the low-power laser to reshape the molten pool, using a CCD camera to scan the bottom of the molten pool and recognize the molten pool contour and calculate the aspect ratio, and stopping the low-power laser when the molten pool contour has no concave-convex points and the aspect ratio of the bottom reaches a stop condition, thereby completing the remelting and reshaping.

Images