MATHEMATICAL METHODS FOR TRACK ANALYSIS IN ADDITIVE MANUFACTURING AND A SYSTEM FOR METHODS POOL DYNAMICS.
Patent Information
- Authority / Receiving Office
- TR · TR
- Patent Type
- Patents
- Current Assignee / Owner
- GAZI UNIVERISTESI
- Filing Date
- 2024-11-13
- Publication Date
- 2026-06-22
Smart Images
Figure 00000012_0000 
Figure 00000013_0000 
Figure 00000013_0001
Abstract
Description
1 TARIFF MATHEMATICAL METHODS FOR TRACE ANALYSIS IN ADDITIVE MANUFACTURING AND A SYSTEM FOR THE DYNAMICS OF A MELTING POOL 5 Technical Field of the Invention This invention relates to Additive Manufacturing (AM), involving in-situ melt pool, layer thickness, and Development of line spacing measuring devices, real-time quality monitoring and adaptive 10 It focuses on control systems. This invention is particularly relevant to additive manufacturing and laser powder bed melting (LPBF) processes. and production can be expanded to related fields such as laser coating and welding. It aims to increase precision and control over quality. State of the Art Additive manufacturing (AM) or 3D printing technology is fundamentally changing production processes. It transforms complex geometries and customized products. Understanding the dynamics of the melt pool and trace analysis enables the production of E1 20. This is critically important in terms of optimizing processes. The melt pool is a key component in electrical engineering that directly affects material properties and part quality. It is a key element. Proper control of the solution pool dynamics leads to better results. It reduces defects and improves mechanical properties by ensuring layer adhesion. Mathematical modeling plays a crucial role in the analysis of trace geometry in electrical engineering. 25 Current techniques generally estimate basic parameters such as width, height, and depth. Traditional approaches rely on simplified geometric models to achieve this. It usually assumes idealized shapes, for example, perfect parabolas, but The trace formation created by the influence of changing process parameters in the real world It cannot reflect their complexity. 30 Trace analysis is a discipline that examines the traces of the energy source during the printing process. This In the analysis, mathematical models consider parameters such as speed, power, and material feed rate. It evaluates the effects on track geometry and surface quality. Various studies have shown that mathematical methods can predict melt pool dynamics. This has demonstrated its applicability. For example, researchers have shown that different materials exhibit thermal properties of 35... Using Finite Element Analysis (FEA) to model their behavior optimally They have obtained important information about the processing conditions. 2 Numerical methods such as Finite Element Method (FEM) and Computational Fluid Dynamics (CFD) methods are increasingly being used to model melt pool behavior These simulations are used to provide information about temperature distribution and solidification rates. It provides valuable information. However, the effectiveness of these methods is questioned due to high computational demands and each Time is limited due to precise boundary conditions that are practically impossible to achieve. 5 Some researchers are exploring the relationships between process parameters and trace characteristics. They have begun to apply statistical modeling techniques for this. These models are valuable. Although it can provide correlations, it is generally the fundamental physics that governs melt pool dynamics. It neglects this, which results in less comprehensive insights. One of the biggest challenges encountered in melt pool dynamics is in-situ 10 The problem is the lack of measurement capabilities. Most existing systems measure the melt during production. It relies on post-transaction evaluations, not capturing the dynamic behavior of the pool. High-speed imaging and thermal imaging to facilitate real-time monitoring. Advances in sensor technologies, such as sensors, are necessary. In the known state of the art, trace analysis, process control and optimization in EI are 15. A sophisticated system needs to be developed. This system will optimize the electrical engineering process. should improve, enhance part quality, and understand the fundamental physics of trace formation in EI. It should provide deeper insights. For this, the latest mathematical-analytical modeling, Utilizing sensor technologies and control engineering principles It is required. 20 In conclusion, the drawbacks described above and the existing solutions are relevant to the issue. Due to these shortcomings, it has become necessary to make an improvement in the relevant technical field. Purposes of the Invention This invention enables real-time monitoring, prediction, and optimization of additive manufacturing processes. It is related to developing an advanced trace analysis and control system through this method. Thus, the existing Limitations of quality control methods necessitate the introduction of sophisticated mathematical models and advanced technologies into the system. This will be overcome by integrating sensor technology and machine learning algorithms. Another aim of the invention is to provide a comprehensive 30-degree range of contact angles, including sharp and wide-angle contacts. The goal is to provide a mathematical framework. This framework explains the complex relationships between trace parameters, enabling dimensionless wetting. By providing a detailed characterization of shape factors and track profiles, the system can be predicted. It aims to improve its capabilities. Thus, powder layer thickness, laser energy 3 Interactions between absorption and trace dimensions using hyperbolic modeling. By examining it closely, it provides a deeper understanding. Another objective of the invention is to combine a mathematical basis with the latest sensor technology and machinery. The goal is to develop an integrated control system that combines learning with advanced technology. This will enable high-level control. Real-time monitoring using high-resolution cameras and advanced imaging systems. 5 Formation data is collected, while thermal sensors detect temperature gradients and material phases. It makes it possible to track their movements. Another aim of the invention is to simulate and account for realistic additive manufacturing conditions. The goal is to incorporate control algorithm predictions and control decisions into the production process. It closely reflects the actual behavior of the materials during the process. 10 Another objective of the invention is a convolutional neural network, which is the deep learning component of the system. Input parameters and real-time sensor data via (convolutional neural network) The goal is to provide the ability to predict and track trail characteristics based on data. Thus Proactive adjustments can be made to process parameters. Another objective of the invention is to incorporate surface energy balance and phase 15 into the system's prediction models. volumetric factors affecting track geometry by including change dynamics factors The goal is to enable changes and surface tension effects. Another aim of the invention is to enable engineers to specify desired part properties and materials. an advanced trace with an intuitive user interface where you can specify your specifications The goal is to provide an analysis and control system. The control algorithm monitors laser power throughout production, 20 Autonomous optimization of process parameters such as scanning speed and powder feed rate does. Another aim of the invention is to improve part quality, reduce defects, and use suitable materials. and to expand the geometric range. Thus, real-time sensing and adaptive Thanks to machine learning, 25 critical areas in invention, traceability and quality assurance It responds to challenges and caters to various sectors, including aviation and medical device manufacturing. It offers a transformative solution. Explanations of the Figures Figure 1 shows the trace analysis and control system. Figure 2: Powder layer and trace characteristics: sharp wetting angle θ < 90° Figure 3: Powder layer and trace characteristics: wide wetting angle θ > 90° Figure 4: Traces used to measure the relationship between normalized enthalpy and depth. Figure 5: Fitted circles, wetting angles, and parabolas for sharp and wide angles. 35 4 Reference Numbers: 1: Acoustic sensors 2: Photonic sensors 3: X-ray measurement system 4: Laser energy absorption sensors 5 5: Equipment room 6: Denudation area 7: Track width 8: Track reinforcement area 9: Trace penetration area 10 10: Printing plate 11: Angle of contact or wetting 12: Track depth 13: Track height 14: Dust layer thickness 15 15: Gas flow 16: Scanning direction 17: Spot size 18: Image analysis 19: Real-time analysis 20 20: Algorithms 21: Computer equipment Description of the Invention This invention enables real-time monitoring of additive manufacturing processes, 25 with an advanced track analysis and control system that improves through prediction and optimization It is related to advanced mathematical models, sensor technology, and machine learning. By integrating its algorithms (20), this system replaces traditional quality control methods It provides precision and reliability by overcoming its limitations. This invention covers track width (7), track height (13), track depth (12), beam spacing and wetting angles a comprehensive explanation of the complex relationships between key trace parameters such as It is a mathematical framework. This framework encompasses sharp and wide contact angles and traces It introduces new dimensionless wetting shape factors that detail their profiles. These shape factors It is expressed as follows: The area of a track segment can be calculated using these shape factors with the following equation: computable: These equations relate the dust layer thickness (14) to the track dimensions and laser energy absorption 10 It establishes the relationship between them and predicts how laser parameters affect track profiles. For example, the contact angle theta (θ) is expressed as follows: Furthermore, the absorption capacity of the material depends on the properties of the material and the laser. It depends on the parameters and is scaled with normalized enthalpy and track depth (12). This equation, second order between dust layer thickness (14) and track width (7) (W) manages the relationship: The following equation represents the stationary hatch ( ) which is highly reliable and accurate: This method relies on constant state height and is not dependent on depth; because 25 This model may be inaccurate due to insufficient parabolic fitting over depth. This model is mostly... This scenario predicts a line spacing of 0.71 times the width, which ensures full penetration. To ensure a fully dense structure, the maximum track spacing is 71% of the width. It indicates that it should be. 6 Mathematical models form the basis of a system's predictive capabilities, thus It provides precise control of additive manufacturing processes and improves overall efficiency and quality. increases. The control system is based on this mathematical foundation, combined with state-of-the-art sensor technology and machinery. It integrates with learning. High-resolution cameras and advanced imaging 5 The systems collect real-time data on trail formation using thermal sensors. It monitors temperature gradients and material phase transitions. An advanced deep learning network, This multimodal data stream processes edge detection and boundary point analysis, whether liquid or solid. It accurately identifies and tracks geometric directions in phases and wetting properties. The present invention simulates and accounts for realistic additive manufacturing conditions. Control 10 The algorithm involves the adsorption of molten material components onto the substrate surface and different It addresses material interactions such as mass transfer effects between phases. This holistic approach... The approach involves making predictions and control decisions based on the materials during the production process. It allows him to closely reflect his true behavior. The present invention addresses volumetric changes and surface stress affecting the track geometry. 15 They incorporate these factors into their predictive models to anticipate and mitigate their effects. The control algorithm, production process, laser power, scanning speed, and powder feeding speed. It autonomously determines and adjusts optimal process parameters throughout. This advanced track analysis and control system represents a significant advantage in additive manufacturing technology. It represents progress. Dramatically improving part quality, reducing errors, and 20 Expanding the range of possible materials and geometries for additive manufacturing. It promises advanced mathematical-analytical modeling, real-time detection, and By combining adaptive machine learning, this invention enables traceability control and quality assurance. It addresses their challenges and is suitable for many industries, from aviation to medical device manufacturing. It offers a transformative solution. 25 The present invention provides advanced trace analysis and control for additive manufacturing, welding, and coating. It is a system that incorporates material preprocessing, process analysis, trace formation, and deep learning. It includes computer hardware (21) with capabilities. Computer hardware (21) includes all It is a database where process data is recorded. Material pre-processing equipment uses powders of different sizes and properties, 30 It mimics the behavior of the material in dense configurations. Process analysis equipment, acoustic sensors for material density and viscosity (1), Photonic sensors for chemical composition analysis, X-rays for crystal structure and elemental composition. beam measurement systems (3) and laser energy absorption for normalized enthalpy measurement 7 Includes sensors for real-time material characterization such as sensors (4). Track formation equipment creates tracks on a substrate under controlled conditions, and a temperature-controlled instrument chamber (5), high-resolution to capture the trace geometry imaging devices and dust layer thickness (14), liquid layer thickness and It contains sensors that measure the solidified trace dimensions. The instrument room (5) contains the additive manufacturing process 5 It is a controlled environment in which it takes place. Computer equipment (21) analyzes trace features using deep learning methods. It detects edges, trains convolutional neural networks with boundary points, and makes predictions. It selects target boundary points based on errors and determines track characteristics. It performs derivative calculations, including wetting angles. 10 Analyze trace features using deep learning methods with computer equipment (21). It obtains an edge map by performing edge detection on trace images, and makes a prediction. to create the model, use a specific number of boundary points to construct a convolutional neural network. It trains the target boundary based on the differences between the predicted and actual coordinates. It selects the points, calculates the derivatives at the target boundary points, and determines the trace characteristics. This is 15 Features include contact angles for powder, liquid and solid material interfaces, The volume of solidified material trace, powder consumption, and powder area are included. The present invention has been trained with various datasets on track formation under various conditions. Deep learning utilizes a convolutional neural network. This network processes input parameters and real-world situations. It predicts track characteristics based on time-sensitive sensor data and the process is 20 It enables proactive adjustments to the parameters. The system also monitors the observed parameters. an adaptive feedback loop that continuously improves prediction models based on results. The feed control mechanism includes a loop, which improves accuracy and responsiveness over time. increases its speed. The present invention incorporates a feedback control mechanism. Feedback control 25 The mechanism continuously monitors scar formation parameters and measures scar characteristics. compares the results with the desired outcomes and optimizes the process for trace formation. It adjusts the parameters. These parameters include laser power, scanning speed, and powder feed. These include velocity, substrate temperature and inert gas flow rate (15). The feedback control mechanism method includes the following steps: • Continuously monitoring scar formation parameters, • To compare trace features measured with deep learning with the desired results, • Adjusting process parameters in real time, including laser power, 8 These include scanning speed, powder feed rate, substrate temperature and inert gas flow rate (15). The system also continuously updates prediction models based on observed results. It includes an adaptive feedback loop that improves accuracy over time. and increases reaction speed. 5 Process monitoring equipment, CCD camera, captures the process with high-resolution images. It enables capture and the laser system (16) can be controlled to scan the powder in the direction of melting. (16) and provides energy with spot size (17). The operational process and analysis method of the analysis and control system includes the following steps: • Release the thickness of the powder layer (14) on the building board (10), 10 • A melting pool is created by scanning the powder bed and performing laser melting with a laser. creation, • The creation of a scar, • Real-time analysis with data collected via sensors (19), • Performing deep learning analysis using deep learning algorithms, 15 • Adaptive control is achieved using laser results, • Real-time process parameters and quality measurements in the user interface. showing visualizations, • All process data is recorded in the database. Creating the melting pool by scanning the powder bed and performing laser melting, 20 basic parameters such as laser scanning direction (16), laser spot size (17) and gas flow (15) These include elements that are essential for maintaining the melting pool. The creation of the trace is carried out through the operational processes of the analysis and control system. This is one of the steps in the analysis method and includes the following procedural steps: • Measurement of track width (7) with track formation equipment, 25 • Analysis of the trace enhancement area (8) in the powder bed by deep learning, • Evaluation of the trace penetration area (9) into previous layers with deep learning, • Calculation of contact or wetting angle (11) by deep learning, • Measuring track depth (12) and track height (13), Real-time analysis (19), operational process and analysis of the analysis and control system 30 This method is one of the steps and includes the following procedural steps: • Monitoring of the peeling zone (6) around the melting pool with deep learning, • Sensors continuously collect data about material properties and process conditions, • Captures real-time high-resolution images with its CCD camera and 9 processing. The peeling zone (6) is where dust particles are located due to vapor-driven flow. It is the area around the melting pool that it altered. Performing deep learning analysis with deep learning algorithms (20), analysis 5 and is one of the operational process and analysis method steps of the control system. This includes the following steps: • High-resolution images of the tracks formed on the substrate are captured by a CCD camera. obtaining, • Preprocessing images with deep learning, noise reduction, and Gaussian scale. smoothing application, • Edge detection using deep learning methods and generating an edge map. being done, • Training a convolutional neural network with deep learning, using a specific number of boundary points, • Selection of target boundary points based on prediction errors using deep learning, 15 • Track features are determined by using derivatives at target boundary points with deep learning. determination, • Laser energy absorption, powder layer thickness (14), liquid layer thickness and solidified trace measuring the dimensions, • Calculation of solidified material trace volume and powder consumption using deep learning 20 and determining the relationship between trail and dust areas.
Claims
REQUESTS 1. Advanced additive manufacturing, welding, and coating processes as described below. It is a trace analysis and control system; its features include: • The behavior of powders of different sizes and properties in dense configurations 5 Material preprocessing equipment for simulation, • Material density and viscosity with acoustic sensors (1), photonic sensors (2) chemical composition analysis, crystal structure and elemental composition by X-ray measurement systems (3) and real normalized enthalpy measurement with laser energy absorption sensors (4) Process analysis equipment including sensors for time-based material characterization, 10 • A temperature-controlled system that creates tracks on a substrate under controlled conditions. device room (5), high-resolution imaging devices for capturing track geometry and to measure the powder layer thickness (14), liquid layer thickness and solidified track size Tracking equipment containing sensors, • Using deep learning methods to analyze trail features, edge detection 15 It performs, trains convolutional neural networks with boundary points, and is based on prediction errors. derivatives for selecting target boundary points and determining track characteristics including wetting angles It includes computer equipment (21) that calculates.
2. Advanced trace analysis and control for additive manufacturing, welding and coating according to Claim 1 20 It is a computer system whose characteristic feature is that it has a database in which all transaction data is recorded. It includes equipment (21).
3. The operational process and analysis method of the analysis and control system, and its characteristics are: • Depositing a layer of powder of a certain thickness (14) on the building plate (10), 25 • Creating a melt pool by scanning the powder bed and laser, • The creation of a scar, • Real-time analysis (19) of data collected through sensors, • Performing deep learning analysis with deep learning algorithms (20), • Adaptive control is achieved using laser results, 30 • Real-time display of process parameters and quality metrics in the user interface. demonstrating its visualization, • Saving all transaction data to the database includes the transaction steps.
4. Operational process and analysis method according to claim 3, characterized by; trace formation 35 the process step to be carried out; • Measuring the track width (7) with track-forming equipment, • Deep learning analysis of the trace reinforcement area (8) on the powder bed, • Evaluation of the trace penetration area (9) to previous layers by deep learning, • Calculation of contact or wetting angle (11) by deep learning, 40 • The procedure includes measuring track depth (12) and track height (13). 11 5. Operational process and analysis method according to claim 3, characterized by its use of sensors. The process step is to perform real-time analysis (19) with the collected data; • Deep learning monitoring of the peeling zone (6) around the melting pool, • Sensors continuously collect data on material properties and process conditions, 5 • Real-time capture of high-resolution images with a CCD camera. and includes the processing steps.
6. According to claim 3, it is an operational process and analysis method, characterized by its deep learning feature. The process step of performing deep learning analysis with algorithms (20): 10 • High-resolution images of the traces created on the substrate using a CCD camera. obtaining images, • To reduce noise and apply Gaussian smoothing, the depth of the images learning and preprocessing, • Edge detection using deep learning method to obtain edge map 15 to be done, • Deep learning convolutional neural network with a predetermined number of boundary points training, • Selection of target boundary points based on prediction errors using deep learning, • Track features based on derivatives at target boundary points can be obtained using deep learning. determination, • Laser energy absorption, powder layer thickness (14), liquid layer thickness and Measuring the dimensions of solidified traces, • Calculation of solidified material trace volume and powder consumption using deep learning. and includes procedural steps for determining the relationship between trace and dust areas. 25