Orthopedic 3D printing method and device based on AI

Abstract

This invention relates to the field of orthopedic 3D printing technology, specifically to an AI-based orthopedic 3D printing method and apparatus, comprising the following steps: based on patient physiological data, using outlier detection algorithms and missing value processing methods, cleaning and standardizing the raw data, including removing outliers and filling in missing data, and performing data normalization to generate a physiological dataset. In this invention, the use of support vector machines demonstrates high efficiency in pattern recognition and classification, enabling more accurate analysis of bone density and blood flow characteristics data, which is crucial for customized printing solutions. The application of deep learning networks allows for a deeper analysis of the performance of 3D printing equipment and material properties, optimizing printing strategies and improving printing quality and efficiency. The use of CFD simulation and finite element analysis models provides more accurate predictions for the printing process, reducing trial-and-error costs.

Description

Technical Field

[0001] This invention relates to the field of orthopedic 3D printing technology, and more particularly to an AI-based orthopedic 3D printing method and apparatus. Background Technology

[0002] Orthopedic 3D printing technology combines modern 3D printing technology with orthopedic medical needs to create orthopedic implants, assistive devices, or bone models. This technology allows doctors to precisely design and manufacture implants and supports based on a patient's specific bone structure and medical requirements. The application of 3D printing in orthopedics can improve surgical precision and success rates while reducing patient recovery time.

[0003] AI-based orthopedic 3D printing refers to a technology that utilizes artificial intelligence to optimize the orthopedic 3D printing process. Its aim is to further improve the effectiveness and efficiency of 3D printing in the orthopedic field through AI's data analysis and learning capabilities. AI can process large amounts of medical imaging data, helping to identify and analyze patient needs, thereby achieving more precise printing solutions. The goal of this method is to achieve higher surgical success rates, faster patient recovery times, and better implant biocompatibility.

[0004] Traditional methods have shortcomings in terms of print quality, efficiency, and process optimization. Regarding print quality and efficiency, the lack of equipment performance and material property analysis leads to inconsistent accuracy and quality of printed products. In terms of print process management, traditional methods lack analytical and optimization tools, resulting in inefficient steps and impacting overall production efficiency. These shortcomings, in practice, increase production costs and extend print delivery times. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an AI-based orthopedic 3D printing method and apparatus.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an AI-based orthopedic 3D printing method, comprising the following steps:

[0007] S1: Based on patient physiological data, outlier detection algorithms and missing value handling methods are used to clean and standardize the raw data, including removing outliers and filling in missing data, and performing data normalization to generate a physiological dataset.

[0008] S2: Based on the physiological dataset, support vector machine is used to perform pattern recognition and classification on the data, analyze the patient's bone density and blood flow characteristics data, and adjust the 3D printing parameters to generate printing parameter configuration;

[0009] S3: Based on the aforementioned printing parameter configuration, a deep learning network is used to analyze the performance and material properties of the 3D printing equipment, evaluate the printing effect under multiple configurations, and generate an optimized printing strategy.

[0010] S4: Based on the optimized printing strategy, CFD simulation and finite element analysis models are used to simulate and test the printing process, simulate material behavior and structural performance, and generate simulation test results;

[0011] S5: Based on the simulation test results, event log analysis is used to analyze multiple stages in the 3D printing process, identify inefficient links in the process, including design, preparation, printing and post-processing, and generate an improved printing process.

[0012] S6: Based on the improved printing process, a convolutional neural network is used to analyze historical printing data and real-time production environment data, identify and learn data patterns and trends, and generate a process prediction and scheduling scheme.

[0013] S7: Based on the process prediction and scheduling scheme, the shortest path algorithm is used to schedule the printing tasks according to the dependencies and priorities between tasks, and generate 3D printed products.

[0014] As a further aspect of the present invention, the physiological dataset includes bone density data and blood flow characteristic data; the printing parameter configuration includes adjusted printing speed, material type, and layer thickness parameters; the optimized printing strategy specifically refers to matching printing path selection, support structure setting, and temperature and speed settings for various materials; the simulation test results include interlayer adhesion, support structure stability, and structural durability; and the process prediction and scheduling scheme includes optimization strategies for production demand prediction, potential bottleneck identification, task order, and resource allocation.

[0015] As a further aspect of the present invention, based on patient physiological data, an outlier detection algorithm and a missing value handling method are used to clean and standardize the original data, including removing outliers and filling in missing data, and performing data normalization to generate a physiological dataset. The specific steps are as follows:

[0016] S101: Based on patient physiological data, data cleaning is performed using data deduplication and standardization methods, and the data is initially organized, including identifying and removing inconsistent data points to generate preliminary organized data;

[0017] S102: Based on the preliminary data, box plot analysis is used to identify and process outliers, including calculating data quartiles and identifying outliers that exceed the range, and generating processed data.

[0018] S103: Based on the processed data, the missing values ​​are processed using the K-nearest neighbor algorithm, including finding similar data points and filling in the missing information to generate missing data;

[0019] S104: Based on the missing data, use max-min normalization to unify the measurement standard of differential variables and adjust the data range to generate a physiological dataset.

[0020] As a further aspect of the present invention, based on the physiological dataset, a support vector machine is used to perform pattern recognition and classification on the data, analyze the patient's bone density and blood flow characteristics data, and adjust the 3D printing parameters to generate the printing parameter configuration. The specific steps are as follows:

[0021] S201: Based on the physiological dataset, a linear support vector machine is used to perform preliminary pattern recognition and classification of the data, identify key features, and generate feature recognition data;

[0022] S202: Based on the feature recognition data, a support vector machine with radial basis function kernel is used to perform data classification analysis, identify and classify multi-category patient data, and generate patient classification data;

[0023] S203: Based on the patient classification data, the decision tree analysis method is used to adjust the 3D printing parameters, match the parameters to the patient data, and generate a preliminary parameter configuration;

[0024] S204: Based on the initial parameter configuration, a genetic algorithm is used to adjust and optimize the printing parameters, evaluate the applicability of the printing parameters, and generate a printing parameter configuration.

[0025] As a further aspect of the present invention, based on the aforementioned printing parameter configuration, a deep learning network is used to analyze the performance and material properties of the 3D printing equipment, evaluate the printing effect under multiple configurations, and generate an optimized printing strategy. The specific steps are as follows:

[0026] S301: Based on the aforementioned printing parameter configuration, a feedforward neural network is used to analyze the output performance of the 3D printing equipment, including printing speed and resolution, to perform quantitative analysis of the equipment performance indicators, and to make preliminary adjustments to the printing parameters based on the analysis results, thereby generating equipment performance analysis results.

[0027] S302: Based on the performance analysis results of the aforementioned equipment, a multilayer sensor is used to analyze the characteristics of the printing material, including melting point, hardness, and elasticity, and to evaluate the material properties, identify the matching printing material type, and generate characteristic analysis results.

[0028] S303: Based on the aforementioned characteristic analysis results, a deep convolutional neural network is used to simulate the expected effects of various printing configurations, evaluate the impact of multiple configuration combinations on print quality, and generate a print effect prediction model.

[0029] S304: Based on the aforementioned printing effect prediction model, a genetic optimization algorithm is used to adjust the printing parameters according to equipment performance limitations and material characteristics, thereby generating an optimized printing strategy.

[0030] As a further aspect of the present invention, based on the optimized printing strategy, CFD simulation and finite element analysis models are used to simulate and test the printing process, simulate material behavior and structural performance, and generate simulation test results. The specific steps are as follows:

[0031] S401: Based on the optimized printing strategy, CFD simulation is used to simulate the flow characteristics of materials during the printing process, and the impact of material flow on printing quality is analyzed to generate a material flow simulation analysis.

[0032] S402: Based on the material flow simulation analysis, the finite element analysis model is used to simulate the stress distribution and deformation of the printed object under stress, and to analyze the mechanical stability of the printed structure, generating a structural stress simulation analysis.

[0033] S403: Based on the structural stress simulation analysis, the gradient descent method is used to identify and predict quality problems encountered during the printing process, including poor interlayer bonding and insufficient support, and to generate problem prediction results.

[0034] S404: Based on the predicted results of the problem, the 3D printing process is evaluated using comprehensive simulation and optimization analysis, and simulation test results are generated.

[0035] As a further aspect of the present invention, based on the simulation test results, event log analysis is used to analyze multiple stages in the 3D printing process, identify inefficient links in the process, including design, preparation, printing, and post-processing, and generate an improved printing process. The specific steps are as follows:

[0036] S501: Based on the simulation test results, principal component analysis is used to perform data dimensionality reduction and feature extraction. Through decision tree classification, the multi-stage design, preparation, printing and post-processing are analyzed to generate process efficiency analysis results.

[0037] S502: Based on the process efficiency analysis results, a genetic algorithm is used to re-plan and restructure the original process, and the adjusted process is iteratively optimized to generate an optimized process scheme.

[0038] S503: Based on the optimized process scheme, Monte Carlo simulation is used to conduct multiple rounds of simulation and testing on the scheme, and adjustments are made according to the simulation results to generate an iteratively improved process;

[0039] S504: Based on the iterative improvement process, the Six Sigma method is used to perform quality control and evaluation on multiple links in the process, and the process is optimized according to the evaluation results to generate an improved printing process.

[0040] As a further aspect of the present invention, based on the improved printing process, the steps of using a convolutional neural network to analyze historical printing data and real-time production environment data, identify and learn data patterns and trends, and generate a process prediction and scheduling scheme are as follows:

[0041] S601: Based on the improved printing process, a convolutional neural network is used to learn and analyze historical printing data and real-time production environment data. The analysis results are used to identify key data patterns and trends and generate data pattern analysis results.

[0042] S602: Based on the data pattern analysis results, a time series forecasting model is used to predict and analyze production demand and potential bottlenecks, and generate a production demand forecast.

[0043] S603: Based on the production demand forecast, linear programming is used to plan and optimize the task sequence and resource allocation, and resources are allocated according to the planning results to generate a resource allocation plan;

[0044] S604: Based on the resource allocation scheme, a dynamic programming algorithm is used to optimize the process scheduling strategy and generate a process prediction scheduling scheme.

[0045] As a further aspect of the present invention, based on the aforementioned process prediction and scheduling scheme, and employing a shortest path algorithm, the printing tasks are scheduled according to the dependencies and priorities between tasks. The specific steps for generating 3D printed products are as follows:

[0046] S701: Based on the process prediction and scheduling scheme, the Floyd algorithm is used to analyze the 3D printing task network, including evaluating the path and path time cost, and generating a full network path cost analysis.

[0047] S702: Based on the full network path cost analysis, the Johnson algorithm is used to optimize the path of printing tasks with multiple priorities and dependencies, match constraints, including resource limitations and task urgency, and generate task execution adjustment paths;

[0048] S703: Based on the task execution adjustment path, a greedy algorithm is used to adjust the printing task execution plan, including the task start time, required resources and expected completion time, to generate a task execution fine-tuning plan;

[0049] S704: Based on the task execution fine-tuning plan, perform 3D printing task execution, including material preparation, printing operation and post-processing, to generate 3D printed products.

[0050] The AI-based orthopedic 3D printing device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the AI-based orthopedic 3D printing method described above.

[0051] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0052] This invention utilizes support vector machines (SVMs) to demonstrate high efficiency in pattern recognition and classification, enabling more accurate analysis of bone density and blood flow characteristics data, which is crucial for customized printing solutions. The application of deep learning networks allows for deeper analysis of 3D printing equipment performance and material properties, optimizing printing strategies and improving printing quality and efficiency. The use of CFD simulation and finite element analysis models provides more accurate predictions for the printing process, reducing trial-and-error costs. Event log analysis significantly impacts the optimization of the printing process, making the entire process more efficient. The application of convolutional neural networks and shortest path algorithms makes the entire printing process more intelligent and automated, greatly improving production flexibility and response speed. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0054] Figure 2 This is a detailed flowchart of step S1 of the present invention;

[0055] Figure 3 This is a detailed flowchart of step S2 of the present invention;

[0056] Figure 4 This is a detailed flowchart of step S3 of the present invention;

[0057] Figure 5 This is a detailed flowchart of step S4 of the present invention;

[0058] Figure 6 This is a detailed flowchart of step S5 of the present invention;

[0059] Figure 7 This is a detailed flowchart of step S6 of the present invention;

[0060] Figure 8 This is a detailed flowchart of step S7 of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0062] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0063] Example 1

[0064] Please see Figure 1 This invention provides a technical solution: an AI-based orthopedic 3D printing method, comprising the following steps:

[0065] S1: Based on patient physiological data, outlier detection algorithms and missing value handling methods are used to clean and standardize the raw data, including removing outliers and filling in missing data, and performing data normalization to generate a physiological dataset.

[0066] S2: Based on the physiological dataset, support vector machine is used to perform pattern recognition and classification on the data, analyze the patient's bone density and blood flow characteristics data, and adjust the 3D printing parameters to generate printing parameter configuration;

[0067] S3: Based on printing parameter configuration, a deep learning network is used to analyze the performance of 3D printing equipment and material properties, evaluate the printing effect under multiple configurations, and generate an optimized printing strategy.

[0068] S4: Based on the optimized printing strategy, CFD simulation and finite element analysis models are used to simulate and test the printing process, simulate material behavior and structural performance, and generate simulation test results;

[0069] S5: Based on simulation test results, event log analysis is used to analyze multiple stages in the 3D printing process, identify inefficient links in the process, including design, preparation, printing and post-processing, and generate an improved printing process.

[0070] S6: Based on the improved printing process, it uses a convolutional neural network to analyze historical printing data and real-time production environment data, identify and learn data patterns and trends, and generate a process prediction and scheduling scheme.

[0071] S7: Based on the process prediction scheduling scheme, the shortest path algorithm is used to schedule printing tasks according to the dependencies and priorities between tasks, and generate 3D printed products.

[0072] The physiological dataset includes bone density data and blood flow characteristic data. The printing parameter configuration includes adjusted printing speed, material type, and layer thickness parameters. The optimized printing strategy specifically refers to the matching printing path selection, support structure settings, and temperature and speed settings for various materials. The simulation test results include interlayer adhesion, support structure stability, and structural durability. The process prediction and scheduling scheme includes production demand prediction, potential bottleneck identification, and optimization strategies for task sequencing and resource allocation.

[0073] In step S1, outlier detection algorithms and missing value handling methods are used to clean and standardize patient physiological data. The Z-Score method is used to identify and remove outliers by calculating the standard deviation of data points from the mean. The K-nearest neighbor algorithm is applied to handle missing values ​​by finding the K nearest neighbors of the missing data points and calculating their average to fill in the missing values. Finally, data normalization is performed by using the Min-Max normalization method to standardize all data to the range of 0 to 1. This operation generates a normalized physiological dataset, including bone density and blood flow characteristics data, providing an accurate and cleaned data foundation for subsequent steps.

[0074] In step S2, a normalized physiological dataset is subjected to pattern recognition and classification using a support vector machine. The dataset is trained using a linear kernel support vector machine, and the model improves classification accuracy by maximizing the margin between categories. This step analyzes the patient's bone density and blood flow characteristics, and adjusts 3D printing parameters such as printing speed, material type, and layer thickness based on the analysis results to generate a customized printing parameter configuration for each patient.

[0075] In step S3, the performance and material properties of the 3D printing equipment are analyzed using a deep learning network. A multilayer perceptron (MLP) network is used to evaluate different performance parameters of the printing equipment, such as printing speed and accuracy. A convolutional neural network (CNN) is used to analyze the performance of different materials under specific parameters, such as interlayer adhesion and the stability of the support structure. This step comprehensively considers the equipment performance and material properties, evaluates the printing effect under various configurations, and generates an optimized printing strategy. This strategy specifies the best printing path selection, support structure settings, and temperature and speed settings for various materials.

[0076] In step S4, the printing process is simulated and tested using CFD simulation and finite element analysis models. Computational fluid dynamics (CFD) software is used to simulate the flow and solidification behavior of materials during printing, and to evaluate the impact of different printing speeds and material temperatures on print quality. Then, finite element analysis (FEA) models are applied to test the structural performance of the printed object, such as analyzing the stress distribution and durability of the printed object. Simulation tests help identify potential problems during the printing process, such as poor interlayer adhesion and unstable support structures, and generate simulation test results to provide a basis for further improving the printing strategy.

[0077] In step S5, multiple stages of the 3D printing process are analyzed through event log analysis. Process mining technology is used to analyze the event logs in the printing process to identify the time consumption and efficiency bottlenecks in each stage of design, preparation, printing and post-processing. Based on the log data, inefficient links and possible improvement points in the process are reflected. Based on the analysis results, an improved printing process is generated. This process optimizes each stage of the entire 3D printing process and improves overall efficiency.

[0078] In step S6, historical printing data and real-time production environment data are analyzed using a convolutional neural network. The CNN is used to identify patterns and trends in the printing data, such as common problems and efficiency bottlenecks in printing tasks. By combining historical data and real-time feedback, future production needs and potential bottlenecks, such as material shortages or equipment maintenance needs, can be predicted. The analysis results are used to generate a process prediction and scheduling scheme, which includes production demand prediction, potential bottleneck identification, and optimization strategies for task order and resource allocation to improve production efficiency.

[0079] In step S7, the printing tasks are scheduled using the shortest path algorithm. The Floyd algorithm is used to analyze the entire printing task process to determine the dependencies and priorities between tasks. The algorithm provides the optimal solution for the execution order of printing tasks by calculating the shortest path between all point pairs. This method ensures that tasks are scheduled in the most efficient order, reducing waiting time and resource waste. Based on this scheduling scheme, 3D printed products are generated.

[0080] Please see Figure 2 Based on patient physiological data, outlier detection algorithms and missing value handling methods are used to clean and standardize the raw data, including removing outliers and filling in missing data, and performing data normalization. The specific steps for generating the physiological dataset are as follows:

[0081] S101: Based on patient physiological data, data cleaning is performed using data deduplication and standardization methods, and the data is initially organized, including identifying and removing inconsistent data points to generate preliminary organized data;

[0082] S102: Based on the preliminary data processing, box plot analysis is used to identify and process outliers, including calculating the data quartiles and identifying outliers that are out of range, and generating processed data.

[0083] S103: Based on the processed data, the K-nearest neighbor algorithm is used to process the missing values, including finding similar data points and filling in the missing information to generate missing data;

[0084] S104: Based on the missing data, the maximum and minimum normalization is used to unify the measurement standard of differential variables and adjust the data range to generate a physiological dataset.

[0085] In sub-step S101, the patient's physiological data is processed using data deduplication and standardization methods. Specific operations include reading the physiological dataset, which may contain multi-dimensional information such as the patient's age, gender, height, weight, and bone density. The data first undergoes deduplication, where algorithms identify and remove duplicate records to ensure the dataset's uniqueness and accuracy. Subsequently, data standardization is performed, involving adjusting various indicators to a unified measurement standard, such as converting height from inches to centimeters or weight from pounds to kilograms. The purpose of this step is to eliminate the influence of different dimensions between data items, making subsequent analysis and processing more accurate and efficient. After this step is completed, preliminary processed data is generated.

[0086] In sub-step S102, box plot analysis is used to detect and process outliers in the initially processed data. Box plots detect outliers by calculating the quartiles of the data and identifying outliers. First, the lower quartile (Q1), median (Q2), and upper quartile (Q3) of the data are calculated. Then, the interquartile range (IQR = Q3 - Q1) is calculated. Points in the data that are lower than Q1 - 1.5IQR or higher than Q3 + 1.5IQR are considered outliers. Outliers may be caused by data entry errors, measurement errors, or other abnormal factors. By identifying and processing these outliers, the quality of the dataset can be improved, ensuring the accuracy of subsequent steps. The processed data will be smoother and reflect the patient's true physiological state.

[0087] In sub-step S103, based on the processed data, the K-nearest neighbor algorithm is used to process missing values. By finding the "neighbors" of data points, the missing values ​​are predicted or filled in. Specifically, the value of K is first determined, that is, how many nearest neighbors should be considered for each data point. Then, for each missing value in the dataset, the algorithm calculates its distance from other non-missing data points and finds the K nearest data points. The feature values ​​of the neighbor points are used to estimate or fill in the missing values. For example, the mean or median of these neighbor points can be used as the estimate of the missing value. In this way, the K-nearest neighbor algorithm can effectively fill in the missing information in the dataset and improve the integrity and usability of the data.

[0088] In substep S104, based on the missing data completion, the max-min normalization method is used for data normalization. Max-min normalization is a commonly used data preprocessing method that unifies the measurement standards of different variables by adjusting the data range. Specifically, for each feature in the dataset, this method scales the feature value to the range of 0 to 1 by subtracting the minimum value of the feature and then dividing by the difference between the maximum and minimum values. This normalization process makes different features at the same order of magnitude, which helps improve the performance of subsequent algorithms such as support vector machines, and the generated physiological dataset will have a uniform scale and format.

[0089] Please see Figure 3 Based on a physiological dataset, a support vector machine is used to perform pattern recognition and classification on the data, analyze patient bone density and blood flow characteristics data, and adjust 3D printing parameters to generate printing parameter configurations. The specific steps are as follows:

[0090] S201: Based on the physiological dataset, a linear support vector machine is used to perform preliminary pattern recognition and classification of the data, identify key features, and generate feature recognition data;

[0091] S202: Based on feature recognition data, support vector machines with radial basis function kernels are used to perform data classification analysis, identify and classify multi-category patient data, and generate patient classification data;

[0092] S203: Based on patient classification data, the decision tree analysis method is used to adjust the 3D printing parameters, match the parameters to the patient data, and generate a preliminary parameter configuration;

[0093] S204: Based on the initial parameter configuration, a genetic algorithm is used to adjust and optimize the printing parameters, evaluate the applicability of the printing parameters, and generate the printing parameter configuration.

[0094] In substep S201, a linear support vector machine (SVM) is used to perform preliminary pattern recognition and classification on the prepared physiological dataset. This process involves mapping each sample in the physiological dataset to a multidimensional space, where each dimension represents a physiological feature, such as age, gender, height, weight, bone density, etc. Specifically, the linear SVM optimizes an objective function to maximize the margin between data points of different categories. During this process, the linear SVM evaluates which features are important for classification, thereby identifying key features. After completing this step, feature recognition data is generated, providing the key information needed for patient classification.

[0095] In substep S202, a radial basis function (RBF) kernel support vector machine is used to perform in-depth classification analysis on the feature recognition data. The RBF kernel support vector machine finds the optimal classification hyperplane by mapping the data to a higher-dimensional space. Specifically, the RBF kernel SVM first determines the parameters of the kernel function, such as the kernel width, and then maps the data to a high-dimensional space using kernel tricks. In this space, it searches for the optimal separating hyperplane. The process involves complex mathematical calculations, including constructing Lagrange multipliers and solving quadratic programming problems. Through this processing, the RBF kernel SVM can effectively identify and classify multi-class patient data, generating patient classification data.

[0096] In sub-step S203, based on patient classification data, decision tree analysis is used to adjust 3D printing parameters. Specifically, the algorithm first selects an optimal feature as the root node, and then divides the dataset into subsets based on this feature. This process is repeated for each subset until all features are used in the construction of the decision tree or a preset stopping condition is reached. In the context of 3D printing parameter adjustment, decision tree analysis will deduce the most suitable printing parameters, such as layer thickness and printing speed, based on the different characteristics and classifications of patients. In this way, the decision tree can generate a set of personalized preliminary parameter configurations for each type of patient, ensuring that the 3D printed product can meet the individual needs of patients to the greatest extent.

[0097] In substep S204, based on the initial parameter configuration, a genetic algorithm is used to further optimize the printing parameters. The algorithm first generates an initial population, with each individual representing a possible printing parameter configuration scheme. Then, by evaluating the fitness of these individuals (i.e., the quality of the parameter configuration), the algorithm selects a group of excellent individuals for crossover and mutation to generate a new generation of population. This process is repeated until a preset number of iterations or fitness threshold is reached. Through this operation, the genetic algorithm can explore and discover more effective printing parameter configurations and generate printing parameter configurations.

[0098] Suppose a patient's physiological dataset includes age 30 years, male, height 175cm, weight 70kg, and bone mineral density 2.5g / cm³. 3 In S201, the linear SVM identifies age and bone mineral density as key features based on this data. In S202, the RBF kernel SVM further classifies the patient into a specific bone mineral density category. In S203, the decision tree derives suitable printing parameters for this category of patient based on the classification results, such as a layer thickness of 0.2 mm and a printing speed of 50 mm / s. In S204, the genetic algorithm finds a more optimized parameter configuration through multiple generations of iteration, such as a layer thickness of 0.15 mm and a printing speed of 55 mm / s, to improve printing quality and efficiency.

[0099] Please see Figure 4 Based on printing parameter configurations, a deep learning network is used to analyze the performance and material properties of 3D printing equipment, evaluate the printing effects under multiple configurations, and generate an optimized printing strategy. The specific steps are as follows:

[0100] S301: Based on the configuration of printing parameters, a feedforward neural network is used to analyze the output performance of 3D printing equipment, including printing speed and resolution, to perform quantitative analysis of equipment performance indicators, and to make preliminary adjustments to printing parameters based on the analysis results, thereby generating equipment performance analysis results.

[0101] S302: Based on the equipment performance analysis results, a multilayer perceptron is used to analyze the characteristics of the printing material, including melting point, hardness and elasticity, and to evaluate the material properties, identify the matching printing material type, and generate characteristic analysis results.

[0102] S303: Based on the characteristic analysis results, a deep convolutional neural network is used to simulate the expected effects of various printing configurations, evaluate the impact of multiple configuration combinations on print quality, and generate a print effect prediction model.

[0103] S304: Based on the print effect prediction model, the genetic optimization algorithm is used to adjust the printing parameters according to the equipment performance limitations and material characteristics, and generate an optimized printing strategy.

[0104] In sub-step S301, the output performance of the 3D printing equipment is analyzed using a feedforward neural network. The data is formatted into a set of numerical values ​​including equipment performance indicators such as printing speed and resolution. The feedforward neural network processes this data through its multi-layer structure, where each layer performs a certain transformation on the input data, and finally outputs a quantitative evaluation of the equipment performance. In specific implementation, the input layer of the network receives the raw data, which is then processed by a series of weighted summations and activation functions in hidden layers. The output layer provides a comprehensive score for indicators such as printing speed and resolution. During this process, the network parameters are optimized using a backpropagation algorithm to reduce prediction errors. The generated equipment performance analysis results are a set of quantitative indicators that reflect the actual performance of the printing equipment.

[0105] In substep S302, the properties of the printing material are analyzed by a multilayer perceptron. The data is formatted into a set of numerical values ​​for material properties such as melting point, hardness, and elasticity. The multilayer perceptron analyzes these data through its deep network structure. Each layer of the network extracts and transforms features from the data. Specifically, the input layer of the network receives the original material property data. Through nonlinear transformation of the hidden layers, the output layer generates a comprehensive evaluation of each material property. During this process, the network adjusts its parameters using the gradient descent method to improve the accuracy of the prediction. The generated property analysis results describe in detail the physical and chemical properties of different materials.

[0106] In sub-step S303, a deep convolutional neural network simulates the expected effects of various printing configurations. The data is formatted into combinations of printing configurations and their corresponding printing effect indicators. The deep convolutional neural network learns from this data to predict the printing effect under different configurations. Specifically, the network uses its convolutional and pooling layers to extract key features from the configuration data and outputs the prediction of the printing effect through a fully connected layer. In this process, the network improves the prediction accuracy through error backpropagation and weight optimization. The generated printing effect prediction model can predict the results under different printing configurations, such as printing quality and efficiency.

[0107] In substep S304, the printing parameters are adjusted using a genetic optimization algorithm. Specifically, the algorithm initializes a set of candidate printing parameter solutions based on equipment performance limitations and material properties. Through selection, crossover, and mutation operations, the algorithm iteratively generates new solutions, evaluates the fitness of each solution, and finally converges to a set of optimal printing parameters. These parameters achieve optimal printing results while meeting the limitations of equipment performance and material properties. The generated optimized printing strategy details the adjusted printing parameters, such as printing speed, material selection, and layer thickness.

[0108] Suppose we have a set of 3D printing equipment performance data, including a printing speed range of 20-100 mm / s and a resolution of 0.1-0.4 mm. Through feedforward neural network analysis, the optimal performance score for this equipment is a printing speed of 60 mm / s and a resolution of 0.2 mm. Next, for a specific printing material with a melting point of 230℃, a hardness of 75 Shore D, and an elasticity of 1200 MPa, multilayer perceptron analysis shows that this material is best suited for a medium-speed and high-precision printing configuration. Furthermore, a deep convolutional neural network predicts that under these parameters, the expected print quality score is 90 / 100. Finally, a genetic optimization algorithm adjusts the printing parameters, determining a final printing speed of 58 mm / s and a layer thickness of 0.22 mm to optimize the printing process. Through these operations, a detailed 3D printing strategy is generated to guide the actual printing operation.

[0109] Please see Figure 5 Based on the optimized printing strategy, CFD simulation and finite element analysis models are used to simulate and test the printing process, simulating material behavior and structural performance. The specific steps for generating simulation test results are as follows:

[0110] S401: Based on an optimized printing strategy, CFD simulation is used to simulate the flow characteristics of materials during the printing process and analyze the impact of material flow on print quality, generating a material flow simulation analysis.

[0111] S402: Based on material flow simulation analysis, using the finite element analysis model, the stress distribution and deformation of the printed object under stress are simulated, and the mechanical stability of the printed structure is analyzed to generate structural stress simulation analysis.

[0112] S403: Based on structural stress simulation analysis, using the gradient descent method, identify and predict quality problems encountered during the printing process, including poor interlayer bonding and insufficient support, and generate problem prediction results.

[0113] S404: Based on the problem prediction results, the 3D printing process is evaluated using comprehensive simulation and optimization analysis, and simulation test results are generated.

[0114] In substep S401, computational fluid dynamics (CFD) simulation is used to simulate the material flow characteristics under the optimized printing strategy. The data format used in this process typically includes the hydrodynamic properties of the material, such as viscosity, density, and flow rate. The CFD simulation first establishes a physical model based on this data. This model can simulate the flow of the material during the printing process, including setting boundary conditions, initial conditions, and relevant fluid dynamics equations, such as the Navier-Stokes equations. By numerically solving these equations, the CFD simulation can generate the flow patterns of the material at different printing stages. By analyzing these patterns, the impact of material flow on printing quality can be evaluated, such as the unevenness of the printed layer that may be caused by uneven flow, thus generating a material flow simulation analysis.

[0115] In substep S402, the stress distribution and deformation of the printed object under stress are simulated using a finite element analysis (FEA) model. The data format typically includes the mechanical properties of the printing material, such as the elastic modulus and Poisson's ratio, as well as the geometric information of the printed object. FEA first divides the printed object into a series of small, manageable units, such as a mesh, based on the geometric information. Then, based on the material properties and stress conditions, physical equations (such as Hooke's Law) and mathematical methods (such as solving differential equations) are applied to calculate the stress and strain of each unit under stress. The analysis can predict the stress concentration and deformation that the printed object may encounter in actual use, generating a structural stress simulation analysis.

[0116] In the S403 sub-step, based on structural stress simulation analysis, the gradient descent method is used to identify and predict potential quality problems encountered during the printing process. The algorithm first defines a loss function that measures the severity of printing quality problems (such as poor interlayer bonding and insufficient support). Then, the algorithm calculates the gradient of the loss function in the parameter space and gradually adjusts the parameters along the direction of gradient descent to approach the point of minimum loss. During this process, the algorithm can identify key factors that lead to printing quality problems, such as printing speed and material temperature, and finally generate problem prediction results.

[0117] In sub-step S404, based on the problem prediction results, comprehensive simulation and optimization analysis are used to evaluate the entire 3D printing process. In this process, the results of the aforementioned CFD simulation, FEA model and gradient descent method are comprehensively utilized to form a comprehensive evaluation framework. Under this framework, not only the flow characteristics of materials and the structural stability of the printed object are considered, but also various quality problems that may occur during the printing process are considered. Through comprehensive evaluation, all aspects of the printing process can be analyzed, potential defects and improvement points can be found, and simulation test results can be generated.

[0118] Assuming a 3D-printed orthopedic implant is being manufactured, in S401, CFD simulation shows that under conditions of a printing speed of 60 mm / s and an extrusion temperature of 200°C, the material flow is stable, but there is slight uneven flow at the corners. In S402, the FEA model predicts that under normal use, stress concentration will occur near the contact point of the implant. In S403, gradient descent analysis identifies that reducing the printing speed to 55 mm / s can reduce the risk of poor interlayer bonding. Finally, in S404, based on the combined simulation and optimization analysis, it is suggested that support structures be added at the corners to improve printing quality. Through these steps, the generated simulation test results provide comprehensive optimization suggestions for the printing process, including adjusting the printing speed and improving the design, to ensure the quality and performance of the implant.

[0119] Please see Figure 6 Based on simulation test results, event log analysis was used to analyze multiple stages in the 3D printing process, identify inefficient steps in the process, including design, preparation, printing, and post-processing, and generate an improved printing process. The specific steps are as follows:

[0120] S501: Based on simulation test results, principal component analysis is used to perform data dimensionality reduction and feature extraction. Through decision tree classification, the multi-stage design, preparation, printing and post-processing are analyzed to generate process efficiency analysis results.

[0121] S502: Based on the process efficiency analysis results, a genetic algorithm is used to re-plan and restructure the original process, and iteratively optimize the adjusted process to generate an optimized process solution.

[0122] S503: Based on the optimized process scheme, Monte Carlo simulation is used to conduct multiple rounds of simulation and testing on the scheme, and adjustments are made according to the simulation results to generate an iteratively improved process;

[0123] S504: Based on iterative process improvement, using the Six Sigma method, quality control and evaluation are performed on multiple steps in the process, and the process is optimized based on the evaluation results to generate an improved printing process.

[0124] In the S501 sub-step, Principal Component Analysis (PCA) is used to perform data dimensionality reduction and feature extraction on the simulation test results. Specifically, the complex simulation data is transformed into a simpler form while retaining the most important information. First, the simulation data (such as interlayer adhesion and support structure stability) is standardized to eliminate the influence of different scales. Then, the covariance matrix of the data is calculated to find the direction of the greatest data variation, which is the principal component. Then, several principal components are selected as features based on the cumulative contribution rate. Finally, the extracted principal components are analyzed using a decision tree classification algorithm to determine the efficiency and problem points of different stages (design, preparation, printing, and post-processing), generating process efficiency analysis results.

[0125] In the S502 sub-step, a genetic algorithm is used to replan and restructure the process based on the process efficiency analysis results. The specific operations include initializing a series of candidate solutions based on the current process structure, each solution representing a possible process arrangement. The fitness of each solution is evaluated based on the process efficiency analysis results. Solutions with high fitness are selected for crossover and mutation operations to generate new process arrangement candidate solutions. This process is continuously iterated until the optimal process structure is found. The generated optimized process scheme is optimized through multiple iterations to provide a more efficient and reasonable 3D printing process arrangement.

[0126] In the S503 sub-step, the optimized process scheme is simulated and tested multiple times using Monte Carlo simulation. The specific operations include simulating different production scenarios using random sampling methods, evaluating the performance of the optimized process scheme under various possible conditions, and adjusting process parameters such as resource allocation and task priority based on the simulation results after each simulation. The process is repeated, and each iteration is adjusted based on the previous simulation results. The final iteratively improved process is continuously optimized in multiple simulations to ensure that the process scheme can adapt to various production changes.

[0127] In the S504 sub-step, Six Sigma methodology is used to perform quality control and evaluation on the iteratively improved process. Specifically, this involves applying the DMAIC (Define, Measure, Analyze, Improve, Control) methodology to key aspects of the process. First, the goals and key performance indicators for process improvement are defined. Then, the current performance level of the process is measured. Next, the data is analyzed to identify the root causes of performance problems. Based on these analyses, improvement measures are implemented and the implementation of the new process is controlled. During this process, Six Sigma tools such as control charts and fault tree analysis are used to monitor process performance and identify potential quality issues. The resulting improved printing process undergoes rigorous quality control and evaluation.

[0128] Assuming a set of data is obtained from simulation testing, including printing speed of 20-60 mm / s, interlayer adhesion strength of 30-70 MPa, and support structure stability score of 60-90 points, key features are extracted through principal component analysis. After decision tree classification, it is found that printing speed is the key factor affecting efficiency. Genetic algorithm optimizes the process and finds that the efficiency is highest at a printing speed of 50 mm / s. Monte Carlo simulation tests are conducted to test the process performance at different speeds, and 50 mm / s is finally determined to be the optimal speed. Finally, the Six Sigma method confirms that the process is stable and meets the quality standards at this speed, forming the final printing process.

[0129] Please see Figure 7 Based on the improved printing process, a convolutional neural network is used to analyze historical printing data and real-time production environment data, identify and learn data patterns and trends, and generate a process prediction and scheduling scheme. The specific steps are as follows:

[0130] S601: Based on an improved printing process, it uses a convolutional neural network to learn and analyze historical printing data and real-time production environment data. Through the analysis results, it identifies key patterns and trends in the data and generates data pattern analysis results.

[0131] S602: Based on the data pattern analysis results, a time series forecasting model is used to predict and analyze production demand and potential bottlenecks, and generate production demand forecasts.

[0132] S603: Based on production demand forecasting, linear programming is used to plan and optimize task sequence and resource allocation, and resources are allocated according to the planning results to generate a resource allocation plan;

[0133] S604: Based on the resource allocation scheme, a dynamic programming algorithm is used to optimize the process scheduling strategy and generate a process predictive scheduling scheme.

[0134] In the S601 sub-step, a convolutional neural network (CNN) is used to learn and analyze historical printing data and real-time production environment data in the improved printing process. The data includes historical printing speed, material usage, fault records, etc., as well as real-time monitored equipment status and output quality. The CNN effectively extracts spatial features from the data through its convolutional layers, pooling layers further reduce the data dimensionality, and fully connected layers integrate this information to make predictions and classifications. In this process, the network learns potential patterns and trends in the data through repeated training, such as material consumption patterns or early signs of equipment failure, and generates data pattern analysis results.

[0135] In sub-step S602, a time series forecasting model is used to predict and analyze production demand and potential bottlenecks. The data format includes time series data such as historical production demand, material consumption, and equipment uptime. A time series forecasting model such as Long Short-Term Memory (LSTM) is used to analyze this data. The model can capture the time dependence and periodic changes in the data. LSTM effectively memorizes long-term dependency information through its gating mechanism, predicts production demand and potential bottlenecks in the future, and the generated production demand forecast provides a quantitative basis for future resource allocation and production planning.

[0136] In sub-step S603, based on production demand forecasting, linear programming is used to plan and optimize task sequence and resource allocation. The linear programming model includes production tasks, resource constraints (such as equipment, personnel, and materials), and objective functions (such as maximizing production efficiency and minimizing costs). The model calculates the optimal resource allocation and task arrangement based on the predicted production demand and existing resources. The linear programming is solved using algorithms such as the simplex method or interior point method to find the optimal solution that satisfies all constraints. The generated resource allocation scheme specifies in detail the execution time and required resources for each task.

[0137] In the S604 sub-step, based on the resource allocation scheme, the dynamic programming algorithm is used to optimize the process scheduling strategy. The algorithm analyzes the production status and resource availability at different time points and gradually constructs the optimal production path from start to finish. In this process, the dynamic programming algorithm determines the optimal decision for each stage and how to adjust task arrangement and resource allocation to adapt to production changes. The generated process prediction scheduling scheme provides a detailed and flexible scheduling plan for the entire 3D printing process.

[0138] Suppose historical printing data shows that the failure rate of a specific 3D printer model increases significantly after running continuously for 8 hours. CNN analyzes this data to identify the correlation pattern between running time and failure rate. Then, based on the generated data from the past three months, LSTM predicts that production demand will increase in the coming week. The linear programming model, based on these predictions and equipment constraints, formulates a weekly production task schedule to ensure that the running time of each printer does not exceed 7 hours. Finally, the dynamic programming algorithm adjusts the task schedule according to the real-time production situation. For example, when a printer fails, the task is immediately reassigned to other equipment, ensuring the smooth operation of the production process, reducing failures and delays, and improving overall production efficiency.

[0139] Please see Figure 8 Based on the process prediction and scheduling scheme, and using the shortest path algorithm, the printing tasks are scheduled according to the dependencies and priorities between tasks. The specific steps for generating 3D printed products are as follows:

[0140] S701: Based on the process prediction and scheduling scheme, the Floyd algorithm is used to analyze the 3D printing task network, including evaluating the path and path time cost, and generating a full network path cost analysis.

[0141] S702: Based on the whole network path cost analysis, the Johnson algorithm is used to optimize the path of printing tasks with multiple priorities and dependencies, match constraints, including resource limits and task urgency, and generate task execution adjustment paths;

[0142] S703: Based on the task execution adjustment path, a greedy algorithm is used to adjust the printing task execution plan, including the task start time, required resources and expected completion time, and generate a task execution fine-tuning plan;

[0143] S704: Based on the task execution fine-tuning plan, perform 3D printing tasks, including material preparation, printing operation and post-processing, to generate 3D printed products.

[0144] In substep S701, the Floyd algorithm is used to perform a full network path cost analysis on the 3D printing task network. First, the data is formatted into a network containing multiple nodes and edges. Each node represents a printing task, and the edges represent the dependencies between tasks. The Floyd algorithm traverses each pair of nodes, calculates the shortest path, and evaluates the time cost of these paths. In practice, the algorithm initializes a distance matrix to store the shortest distance between all node pairs and iteratively updates this matrix until the shortest path for all nodes is found. This process involves not only directly connected nodes but also indirect paths through intermediate nodes. The resulting full network path cost analysis provides a detailed view of the 3D printing task network, showing the optimal connection methods between tasks.

[0145] In substep S702, the Johnson algorithm is used to optimize the paths of print tasks with multiple priorities and dependencies. Based on the results of the whole network path cost analysis, the Johnson algorithm focuses on tasks with specific priorities and dependencies. The algorithm first identifies these tasks and evaluates their impact on the entire printing process. Then, it adjusts the path for each task to meet specific resource constraints and task urgency requirements. During this process, the algorithm iterates continuously to search for the optimal task execution order. While taking into account the dependencies between tasks, the generated task execution adjustment path provides a detailed scheduling scheme for tasks with complex dependencies and priority relationships, ensuring the efficiency and orderliness of the entire printing process.

[0146] In sub-step S703, a greedy algorithm is used to adjust the execution plan of the printing tasks. Based on the results of the task execution adjustment path, the algorithm fine-tunes the execution plan of each printing task, including determining the start time, required resources, and expected completion time of the task. At each step, the greedy algorithm selects what appears to be the optimal choice at the moment in order to achieve the overall optimal solution. In practice, the algorithm assesses the urgency and resource requirements of each task, prioritizes urgent tasks, and allocates resources reasonably. In this process, the greedy algorithm ensures efficient use of resources and timely completion of tasks. The generated task execution fine-tuning plan provides an execution timetable and resource allocation plan for the 3D printing process, ensuring the smoothness and efficiency of the entire process.

[0147] In the S704 sub-step, 3D printing tasks are executed based on a fine-tuned plan. The process involves specific material preparation, printing operations, and post-processing. Each task is executed according to the plan, ensuring that it is carried out in the optimized order and schedule. In practice, the operator or automated system prepares the corresponding printing materials, adjusts the printer settings, and monitors the printing process to ensure that each task is carried out as planned. After printing is completed, necessary post-processing is performed, such as support material removal and surface treatment. The generated 3D printed products are manufactured according to the optimized process, ensuring product quality and compliance with predetermined specifications.

[0148] The AI-based orthopedic 3D printing device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the AI-based orthopedic 3D printing method described above.

[0149] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An AI-based orthopedic 3D printing method, characterized in that, Includes the following steps: Based on patient physiological data, outlier detection algorithms and missing value handling methods are used to clean and standardize the raw data, including removing outliers and filling in missing data, and performing data normalization to generate a physiological dataset. Based on the physiological dataset, a support vector machine is used to perform pattern recognition and classification on the data, analyze the patient's bone density and blood flow characteristics data, and adjust the 3D printing parameters to generate printing parameter configurations. Based on the aforementioned printing parameter configuration, a deep learning network is used to analyze the performance and material properties of the 3D printing equipment, evaluate the printing effect under multiple configurations, and generate an optimized printing strategy. The specific steps for generating the optimized printing strategy are as follows: Based on the aforementioned printing parameter configuration, a feedforward neural network is used to analyze the output performance of the 3D printing equipment, including printing speed and resolution. The equipment performance indicators are quantitatively analyzed, and the printing parameters are initially adjusted based on the analysis results to generate equipment performance analysis results. Based on the performance analysis results of the aforementioned equipment, a multilayer perceptron is used to analyze the characteristics of the printing material, including melting point, hardness, and elasticity, and to evaluate the material properties, identify matching printing material types, and generate characteristic analysis results. Based on the aforementioned characteristic analysis results, a deep convolutional neural network is used to simulate the expected effects of various printing configurations, evaluate the impact of multiple configuration combinations on print quality, and generate a print effect prediction model. Based on the aforementioned printing effect prediction model, a genetic optimization algorithm is used to adjust the printing parameters according to equipment performance limitations and material characteristics, thereby generating an optimized printing strategy. Based on the optimized printing strategy, CFD simulation and finite element analysis models are used to simulate and test the printing process, simulate material behavior and structural performance, and generate simulation test results. The specific steps for generating the simulation test results are as follows: Based on the optimized printing strategy, CFD simulation is used to simulate the flow characteristics of materials during the printing process and analyze the impact of material flow on printing quality, generating a material flow simulation analysis. Based on the material flow simulation analysis, the finite element analysis model is used to simulate the stress distribution and deformation of the printed object under stress, and to analyze the mechanical stability of the printed structure, generating a structural stress simulation analysis. Based on the structural stress simulation analysis, the gradient descent method is used to identify and predict quality problems encountered during the printing process, including poor interlayer bonding and insufficient support, and to generate problem prediction results. Based on the predicted results of the problem, the 3D printing process is evaluated using comprehensive simulation and optimization analysis, and simulation test results are generated. Based on the simulation test results, event log analysis was used to analyze multiple stages in the 3D printing process, identify inefficient links in the process, including design, preparation, printing and post-processing, and generate an improved printing process. Based on the improved printing process, a convolutional neural network is used to analyze historical printing data and real-time production environment data, identify and learn data patterns and trends, and generate a process prediction and scheduling scheme. Based on the aforementioned process prediction and scheduling scheme, the shortest path algorithm is used to schedule printing tasks according to the dependencies and priorities between tasks, thereby generating 3D printed products.

2. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, The physiological dataset includes bone density data and blood flow characteristic data. The printing parameter configuration includes adjusted printing speed, material type, and layer thickness parameters. The optimized printing strategy specifically refers to the matching printing path selection, support structure setting, and temperature and speed settings for various materials. The simulation test results include interlayer adhesion, support structure stability, and structural durability. The process prediction and scheduling scheme includes optimization strategies for production demand prediction, potential bottleneck identification, task sequencing, and resource allocation.

3. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, Based on patient physiological data, outlier detection algorithms and missing value handling methods were used to clean and standardize the raw data, including removing outliers and filling in missing data, and then performing data normalization. The specific steps for generating the physiological dataset are as follows: Based on patient physiological data, data cleaning and standardization methods were used to clean the data and perform preliminary data processing, including identifying and removing inconsistent data points to generate preliminary processed data. Based on the preliminary data, box plot analysis is used to identify and process outliers, including calculating the quartiles and identifying outliers that are out of range, and generating processed data. Based on the processed data, the K-nearest neighbor algorithm is used to process the missing values, including finding similar data points and filling in the missing information to generate missing data. Based on the missing data, the maxima-mina normalization was used to unify the measurement standard of the differential variables, and the data range was adjusted to generate a physiological dataset.

4. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, Based on the aforementioned physiological dataset, a support vector machine is used to perform pattern recognition and classification on the data, analyze the patient's bone density and blood flow characteristics data, and adjust the 3D printing parameters to generate the printing parameter configuration. The specific steps are as follows: Based on the aforementioned physiological dataset, a linear support vector machine is used to perform preliminary pattern recognition and classification of the data, identify key features, and generate feature recognition data. Based on the feature recognition data, a support vector machine with radial basis function kernel is used to perform data classification analysis, identify and classify multi-category patient data, and generate patient classification data. Based on the patient classification data, the decision tree analysis method is used to adjust the 3D printing parameters, match the parameters to the patient data, and generate a preliminary parameter configuration. Based on the initial parameter configuration, a genetic algorithm is used to adjust and optimize the printing parameters, evaluate the applicability of the printing parameters, and generate a printing parameter configuration.

5. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, Based on the simulation test results, event log analysis was used to analyze multiple stages in the 3D printing process, identify inefficient steps in the process, including design, preparation, printing, and post-processing, and generate an improved printing process. The specific steps are as follows: Based on the simulation test results, principal component analysis is used to perform data dimensionality reduction and feature extraction. Decision tree classification is then used to analyze the multi-stage design, preparation, printing, and post-processing processes, generating process efficiency analysis results. Based on the process efficiency analysis results, a genetic algorithm is used to re-plan and restructure the original process, and the adjusted process is iteratively optimized to generate an optimized process scheme. Based on the optimized process scheme, Monte Carlo simulation is used to conduct multiple rounds of simulation and testing on the scheme, and adjustments are made according to the simulation results to generate an iteratively improved process. Based on the aforementioned iterative improvement process, the Six Sigma method is used to perform quality control and evaluation on multiple stages of the process, and the process is optimized based on the evaluation results to generate an improved printing process.

6. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, Based on the improved printing process, the specific steps for generating a process prediction and scheduling scheme by using a convolutional neural network to analyze historical printing data and real-time production environment data, identifying and learning data patterns and trends, are as follows: Based on the improved printing process, a convolutional neural network is used to learn and analyze historical printing data and real-time production environment data. The analysis results identify key patterns and trends in the data and generate data pattern analysis results. Based on the data pattern analysis results, a time series forecasting model is used to predict and analyze production demand and potential bottlenecks, and generate production demand forecasts. Based on the production demand forecast, linear programming is used to plan and optimize the task sequence and resource allocation, and resources are allocated according to the planning results to generate a resource allocation plan. Based on the resource allocation scheme, a dynamic programming algorithm is used to optimize the process scheduling strategy and generate a process prediction scheduling scheme.

7. The AI-based orthopedic 3D printing method according to claim 1, characterized in that, Based on the aforementioned process prediction and scheduling scheme, and employing the shortest path algorithm, the printing tasks are scheduled according to the dependencies and priorities between tasks. The specific steps for generating 3D printed products are as follows: Based on the aforementioned process prediction and scheduling scheme, the Floyd algorithm is used to analyze the 3D printing task network, including evaluating the path and path time cost, and generating a full network path cost analysis. Based on the network-wide path cost analysis, the Johnson algorithm is used to optimize the path of printing tasks with multiple priorities and dependencies, match constraints, including resource limitations and task urgency, and generate task execution adjustment paths. Based on the task execution adjustment path, a greedy algorithm is used to adjust the printing task execution plan, including the task start time, required resources and expected completion time, to generate a task execution fine-tuning plan; Based on the task execution fine-tuning plan, the 3D printing task is executed, including material preparation, printing operation and post-processing, to generate 3D printed products.

8. An AI-based orthopedic 3D printing device, comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the AI-based orthopedic 3D printing method according to any one of claims 1 to 7.

Images