Intelligent control method and system for laser power for 3D printing

Abstract

The application relates to the field and discloses a laser power intelligent control method and system for 3D printing, a 3D model information database is established, and printing original data is collected in real time; a prediction model constructed based on finite element analysis and a convolutional neural network is used for laser power precalculation, and a prediction result is obtained; a preliminary laser power setting value of each layer and each region given by the prediction result is used as a starting power input, a laser generator is driven to start printing operation of the current layer or region according to the setting value; a molten pool state and forming quality information of a printed layer are collected in real time to obtain actual forming quality parameters, the actual forming quality parameters are analyzed, laser power is dynamically adjusted according to the deviation size and direction, and 3D printing forming is realized; the application realizes accurate and intelligent control of 3D printing laser power, effectively improves printing quality and stability, and reduces the waste rate.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D printing, and in particular to a laser power intelligent control method and system for 3D printing. Background Art

[0002] As a rapid prototyping technology, 3D printing has been widely used in many fields such as aerospace, medical treatment, and automobile manufacturing. In 3D printing processes based on laser sintering or melting, laser power plays a vital role in the molding quality. At present, traditional 3D printing laser power control methods mostly use fixed power modes or simple open-loop control methods, which are difficult to adapt to complex printing needs. On the one hand, different printing materials, model structures, and printing stages have different absorption and requirements for laser energy. On the other hand, factors such as temperature changes and material property fluctuations that may occur during the printing process will also affect the effect of the laser. These situations often lead to problems such as loose interlayer bonding, large surface roughness, poor dimensional accuracy, and even internal defects in the printed products, which seriously restricts the development of 3D printing technology towards higher precision and higher quality. Summary of the Invention

[0003] The purpose of the present invention is to solve the above problems and to design a laser power intelligent control method and system for 3D printing.

[0004] A first aspect of the present invention provides a method for intelligently controlling laser power for 3D printing, the method comprising the following steps:

[0005] Obtain the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establish a 3D model information database;

[0006] Sensors arranged around the printing platform collect information about the printing environment temperature, energy distribution in the laser action area, and material composition changes in real time to obtain the original printing data.

[0007] Reading the 3D model information database, combining the original printing data, and using a prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power to obtain a prediction result;

[0008] The initial laser power setting value for each layer or area given by the prediction result is used as the initial power input, and the laser generator is driven to start the printing operation of the current layer or area according to the setting value;

[0009] The molten pool status and the molding quality information of the printed layers are collected in real time to obtain the actual molding quality parameters. The actual molding quality parameters are analyzed, and the laser power is dynamically adjusted according to the deviation size and direction to achieve 3D printing. The molding quality information includes at least the molten pool size, solidification rate and interlayer flatness.

[0010] Optionally, in a first implementation of the first aspect of the present invention, the real-time collection of printing environment temperature, energy distribution in the laser action area, and material composition change information by sensors arranged around the printing platform to obtain printing raw data includes:

[0011] A thermistor is used as a temperature sensor. By applying a constant current to the thermistor and measuring the voltage across the thermistor, the current ambient temperature is obtained according to Ohm's law, and temperature data is continuously collected.

[0012] An optical power meter based on the photoelectric effect is used as an energy sensor. The laser scanning area is divided into multiple small squares. The optical power meter quickly moves to the center of each square and briefly stops for 0.1 second. The laser power value at that point is measured and recorded. This cycle is repeated to construct an energy distribution matrix for the entire laser action area.

[0013] Use material composition monitors to monitor material composition changes in real time. For metal powder materials, X-ray fluorescence spectroscopy is used, and for organic polymer materials, infrared spectroscopy is used. The material composition change data monitored each time is recorded.

[0014] The printing environment temperature, energy distribution in the laser action area, and material composition change information are acquired in real time to obtain the original printing data.

[0015] Optionally, in a second implementation of the first aspect of the present invention, the reading of the 3D model information database, combined with the original printing data, uses a prediction model based on finite element analysis and a convolutional neural network to pre-calculate the laser power to obtain a prediction result, including:

[0016] Discretize the 3D printing model and divide it into multiple tiny units. Assign corresponding thermal conductivity properties to each unit according to the thermophysical parameters of the 3D printing material.

[0017] Using laser as a point heat source input, according to the law of conservation of energy, within the time interval Δt, the heat entering the unit, the change in the unit's own stored heat, and the heat flowing out of the unit satisfy the following relationship:

[0018]

[0019] wherein, p is density, c is specific heat capacity, V is unit volume, and AT is the change of unit temperature within At, is the heat flow from j adjacent units, and m is the number of adjacent units of the inflow heat flow, is the heat flow to k adjacent units, and n is the number of adjacent units of the outflow heat flow, and Ax is the distance in the heat conduction direction;

[0020] An energy distribution is obtained from the laser action area energy distribution in the print original data, and the heat flow input of each unit at different times is determined according to the laser scanning path, speed and energy distribution;

[0021] The heat transfer process in the model during the laser scanning process is simulated through iterative heat balance, the energy input required for different regions of the model to reach the expected melting temperature is calculated, and the energy density requirement E of each unit is obtained d :

[0022]

[0023] wherein, t is the time required to reach the melting temperature, is the heat entering the unit within the i-th time interval;

[0024] The calculated energy density requirement value is obtained to constitute the preliminary output result of the finite element analysis, the energy distribution required based on the physical heat conduction process is determined, the laser power is pre-calculated in combination with the prediction model, and the prediction result is obtained.

[0025] Optionally, in the third implementation manner of the first aspect of the present application, the energy density requirement value calculated is obtained to constitute the preliminary output result of the finite element analysis, the energy distribution required based on the physical heat conduction process is determined, the laser power is pre-calculated in combination with the prediction model, and the prediction result is obtained, including:

[0026] The historical printing data is preprocessed to obtain sample data, each sample data is arranged into a multi-dimensional array containing model structure characteristics, environmental parameters, material characteristics and corresponding laser power values required for successful printing, the prediction model is constructed by training sample data and adopting a bat algorithm optimization, wherein the model structure characteristics at least include model volume, surface area and layer thickness distribution, the environmental parameters at least include printing environment temperature and humidity, and the material characteristics at least include thermal conductivity, specific heat capacity and corresponding laser power values;

[0027] The convolution kernel is used to slide on the input data in the convolution layer of the prediction model to perform feature extraction, the data dimension is continuously reduced through multi-layer convolution and pooling to extract higher-level features, and finally these features are mapped to an output value through a fully connected layer to obtain a predicted laser power adjustment coefficient;

[0028] The energy density requirement obtained by finite element analysis is combined with the laser power adjustment coefficient output by the prediction model to calculate the final laser power prediction and obtain the prediction result.

[0029] Optionally, in a fourth implementation of the first aspect of the present invention, the training using sample data and optimizing the construction of a prediction model using a bat algorithm includes:

[0030] Initialize the parameters of the population, including at least position, speed, frequency, loudness, and population size. The first position value is a random initial solution generated using a uniform distribution.

[0031] Calculate the fitness of each initial solution for the parameters of each individual bat and find the solution with the best fitness, where the objective function is the training error rate of the convolutional neural network;

[0032] The pulse rate is the parameter set by the initialization. The bat's position and speed are updated in each iteration, and a new frequency is generated for each bat. If the random number is greater than the pulse rate, the bat will generate a new solution around the current optimal solution. If the fitness of the new solution is better than the current solution and the random number is less than the loudness, the new solution is taken.

[0033] Evaluate the fitness of the new solution. If the fitness of the new solution is better than the current best solution, update the current best solution and minimum fitness value.

[0034] When the maximum number of iterations is reached, the optimal solution and the corresponding minimum fitness value are returned. The bat algorithm is used to optimize the hyperparameter of the number of convolution kernels of the convolutional neural network to build a prediction model.

[0035] Optionally, in a fifth implementation of the first aspect of the present invention, the step of using the preliminary laser power setting value for each layer or each area given by the prediction result as the initial power input and driving the laser generator to start the printing operation of the current layer or area according to the setting value includes:

[0036] Extracting preliminary laser power setting values ​​corresponding to the current printing layer and each area within the layer from the prediction results;

[0037] The laser power is controlled by adjusting the applied voltage to control the output light intensity. The required laser power setting value P is known. pre , and the light-to-electricity conversion efficiency η of the laser generator, the required input light intensity I is obtained according to the relationship between power and light intensity pre :

[0038]

[0039] Where A is the cross-sectional area of ​​the laser beam;

[0040] Then, according to the Pockels effect formula, the corresponding applied voltage V is calculated. pre :

[0041]

[0042] Among them, V π is the half-wave voltage, I0 is the reference light intensity, and arcsin(·) converts the light intensity ratio into the corresponding voltage value;

[0043] The calculated voltage value is applied to the electro-optical modulation element of the laser generator, driving the laser generator to output laser according to the power setting value and start printing the current layer or area.

[0044] Optionally, in a sixth implementation of the first aspect of the present invention, the step of using the preliminary laser power setting value for each layer or each area given by the prediction result as the initial power input and driving the laser generator to start the printing operation of the current layer or area according to the setting value further includes:

[0045] When driving the laser generator to start printing, a synchronous clock signal is started. If the scanning speed of the printing device is v scan , with a resolution of r res , then the synchronous clock signal frequency f sync Can be set to:

[0046] f sync =v scan ·r res ;

[0047] During the printing process, sensors are used to continuously collect real-time data, including at least the actual laser output power and the ambient temperature of the current printing area. The power error is calculated by comparing the actual output power with the set power.

[0048] Determine whether the power error exceeds the preset allowable error range. If so, trigger the alarm mechanism and perform power compensation according to the error size.

[0049] Monitor the ambient temperature of the printing area. If the temperature exceeds the threshold, printing is paused and the printing process resumes after the temperature drops below the threshold.

[0050] The second aspect of the present invention provides a laser power intelligent control system for 3D printing, which includes a database establishment module, a data acquisition module, a laser power prediction module, a printing operation module and a dynamic adjustment module, wherein:

[0051] Establish a database module to obtain the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establish a 3D model information database;

[0052] The data acquisition module is used to collect the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time through sensors arranged around the printing platform to obtain the original printing data;

[0053] A laser power prediction module is used to read the 3D model information database, combine the original printing data, and use a prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power to obtain a prediction result;

[0054] A printing operation module is used to use the preliminary laser power setting value for each layer and each area given by the prediction result as the initial power input, and drive the laser generator to start the printing operation of the current layer or area according to the setting value;

[0055] The dynamic adjustment module is used to collect the molten pool status and the molding quality information of the printed layers in real time to obtain the actual molding quality parameters, analyze the actual molding quality parameters, and dynamically adjust the laser power according to the deviation size and direction to achieve 3D printing molding. The molding quality information at least includes the molten pool size, solidification rate and interlayer flatness.

[0056] Optionally, in a first implementation of the second aspect of the present invention, the data acquisition module includes a temperature measurement submodule, an energy measurement submodule, a material monitoring submodule and an acquisition submodule, wherein:

[0057] The temperature measurement submodule uses a thermistor as a temperature sensor. By applying a constant current to the thermistor and measuring the voltage across the thermistor, the current ambient temperature is obtained according to Ohm's law, and temperature data is continuously collected.

[0058] The energy measurement submodule uses an optical power meter based on the photoelectric effect as an energy sensor to divide the laser scanning area into multiple small squares. The optical power meter quickly moves to the center of each square and briefly stops for 0.1 second to measure and record the laser power value at that point. This cycle is repeated to construct an energy distribution matrix for the entire laser action area.

[0059] The material monitoring submodule is used to monitor the changes in material composition in real time using a material composition monitor. For metal powder materials, X-ray fluorescence spectroscopy is used, and for organic polymer materials, infrared spectroscopy is used. The data of each monitored material composition change is recorded.

[0060] The acquisition submodule is used to obtain the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time to obtain the original printing data.

[0061] Optionally, in a second implementation of the second aspect of the present invention, the laser power prediction module includes a discretization processing submodule, an adoption submodule, a determination submodule, a first calculation submodule and a second calculation submodule, wherein:

[0062] The discretization processing submodule is used to discretize the 3D printing model, divide the 3D printing model into multiple small units, and assign corresponding thermal conductivity properties to each unit according to the thermophysical parameters of the 3D printing material;

[0063] A submodule is used to use laser as a point heat source input. According to the law of conservation of energy, within the time interval Δt, the heat entering the unit, the change in the unit's own stored heat, and the heat flowing out of the unit satisfy the following relationship:

[0064]

[0065] Where ρ is the density, c is the specific heat, V is the unit volume, and ΔT is the change in unit temperature within Δt. is the heat flux from j adjacent cells, m is the number of adjacent cells into which the heat flux flows, is the heat flow to k adjacent cells, n is the number of adjacent cells from which the heat flow flows, and Δx is the distance in the direction of heat conduction;

[0066] The determination submodule is used to obtain the energy distribution from the energy distribution of the laser action area in the printing original data, and determine the heat flow input of each unit at different times according to the laser scanning path, speed and energy distribution;

[0067] The first calculation submodule is used to simulate the heat transfer process in the model during laser scanning through iterative thermal balance, calculate the energy input required for different areas of the model to reach the expected melting temperature, and obtain the energy density requirement E corresponding to each unit. d :

[0068]

[0069] Where t is the time required to reach the melting temperature, is the heat entering the cell during the i-th time interval;

[0070] The second calculation submodule is used to obtain the calculated energy density requirement value to constitute the preliminary output result of the finite element analysis, determine the energy distribution required based on the physical heat conduction process, and perform laser power pre-calculation in combination with the prediction model to obtain the prediction result.

[0071] In the technical solution provided by the present invention, the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models are obtained to establish a 3D model information database; the printing environment temperature, the energy distribution in the laser action area, and the material composition change information are collected in real time by sensors arranged around the printing platform to obtain the original printing data; the 3D model information database is read, and in combination with the original printing data, a prediction model based on finite element analysis and a convolutional neural network is used to pre-calculate the laser power to obtain a prediction result; the preliminary laser power setting value for each layer and each area given by the prediction result is used as the starting power input, and the laser generator is driven to start the printing operation of the current layer or area according to the setting value; the molten pool state and the molding quality information of the printed layer are collected in real time to obtain the actual molding quality parameters, the actual molding quality parameters are analyzed, and the laser power is dynamically adjusted according to the deviation size and direction to achieve 3D printing molding, wherein the molding quality information includes at least the molten pool size, solidification rate and interlayer flatness; the present invention realizes precise and intelligent control of the 3D printing laser power, effectively improves the printing quality and stability, reduces the scrap rate, and meets the high-precision and high-performance requirements of 3D printing in different fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Various other advantages and benefits will become apparent to those skilled in the art by reading the following detailed description of the preferred embodiment.The accompanying drawings are only for the purpose of illustrating the preferred embodiment and are not to be considered as limiting the present invention.

[0073] Figure 1 A schematic diagram of a first embodiment of a laser power intelligent control method for 3D printing provided by an embodiment of the present invention;

[0074] Figure 2 A schematic diagram of a second embodiment of the laser power intelligent control method for 3D printing provided by an embodiment of the present invention;

[0075] Figure 3 A schematic diagram of a third embodiment of the laser power intelligent control method for 3D printing provided by an embodiment of the present invention;

[0076] Figure 4 This is a schematic diagram of the structure of a laser power intelligent control system for 3D printing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The terms "first", "second", "third", "fourth" and the like in the description and in the claims of the present application, and above-mentioned drawings, if any, are used to distinguish between similar objects and not necessarily for describing a specific sequential or chronological order. It is to be understood that the use of the terms so-termed, data, can, where appropriate, be interchanged with one another, so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprise" or "have" and any variations thereof, are intended to cover non-exclusive inclusions, for example, a process, method, device, product or apparatus that comprises a list of steps or units must not necessarily be limited to those steps or units that are clearly listed, but can include other steps or units that are not clearly listed or inherent to such processes, methods, products or apparatuses.

[0078] For the sake of understanding, the specific flow of the embodiments of the present application is described below, please refer to Figure 1 The first embodiment of the method for intelligent control of laser power for 3D printing provided by the embodiments of the present application is shown in the figure, and the method specifically comprises the following steps:

[0079] Step 101, obtaining the thermal physical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establishing a 3D model information database;

[0080] Step 102, collecting the printing environment temperature, energy distribution of the laser action area and material composition change information in real time through the sensors arranged around the printing platform, and obtaining the printing original data;

[0081] Step 103, reading the 3D model information database, combining the printing original data, and using the prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power, and obtaining the prediction result;

[0082] In this embodiment, the 3D model information database is read, the printing original data is combined, the prediction model based on finite element analysis and convolutional neural network is used to pre-calculate the laser power, and the prediction result is obtained, including:

[0083] Discretizing the 3D printing model, dividing the 3D printing model into a plurality of micro units, and assigning each unit with corresponding heat conduction properties according to the thermal physical parameters of the 3D printing materials;

[0084] Using laser as point heat source input, according to the law of conservation of energy, the heat entering the unit, the change of the heat stored by the unit itself and the heat flowing out of the unit in the time interval Δt satisfy the following relationship:

[0085]

[0086] Where ρ is the density, c is the specific heat, V is the unit volume, and ΔT is the change in unit temperature within Δt. is the heat flux from j adjacent cells, m is the number of adjacent cells into which the heat flux flows, is the heat flow to k adjacent cells, n is the number of adjacent cells from which the heat flow flows, and Δx is the distance in the direction of heat conduction;

[0087] Obtain energy distribution from the energy distribution of the laser action area in the original printing data, and determine the heat flow input of each unit at different times based on the laser scanning path, speed and energy distribution;

[0088] The heat transfer process in the model during laser scanning is simulated by iterative thermal balance, and the energy input required to reach the expected melting temperature in different areas of the model is calculated to obtain the energy density requirement E corresponding to each unit. d :

[0089]

[0090] Where t is the time required to reach the melting temperature, is the heat entering the cell during the i-th time interval;

[0091] The calculated energy density requirement value constitutes the preliminary output result of the finite element analysis, determines the energy distribution required based on the physical heat conduction process, and combines the prediction model to pre-calculate the laser power to obtain the prediction result.

[0092] Step 104: Using the initial laser power setting value for each layer or each area given by the prediction result as the initial power input, the laser generator is driven to start the printing operation of the current layer or area according to the setting value;

[0093] In this embodiment, the preliminary laser power setting values ​​corresponding to the current printing layer and each area in the layer are extracted from the prediction results;

[0094] The laser power is controlled by adjusting the applied voltage to control the output light intensity. The required laser power setting value P is known. pre , and the light-to-electricity conversion efficiency η of the laser generator, the required input light intensity I is obtained according to the relationship between power and light intensity pre :

[0095]

[0096] Where A is the cross-sectional area of ​​the laser beam;

[0097] Then, according to the Pockels effect formula, the corresponding applied voltage V is calculated. pre :

[0098]

[0099] wherein, V π is the half-wave voltage, I0 is the reference light intensity, arcsin(·) converts the light intensity ratio into the corresponding voltage value;

[0100] The calculated voltage value is applied to the electro-optical modulation element of the laser generator, and the laser generator is driven to output laser according to the power setting value to start printing the current layer or area.

[0101] In this embodiment, a synchronization clock signal is started at the same time as driving the laser generator to start printing. If the scanning speed of the printing device is v scan , the resolution is r res , and the frequency f sync of the synchronization clock signal can be set as:

[0102] f sync =v scan ·r res ;

[0103] During printing, real-time data is continuously collected by the sensor, including at least the actual output power of the laser and the ambient temperature of the current printing area. By comparing the actual output power with the set power, the power error is calculated.

[0104] If the power error exceeds the pre-set allowable error range, an alarm mechanism is triggered, and power compensation is performed according to the error size.

[0105] The ambient temperature of the printing area is monitored. If the temperature change exceeds the temperature threshold, printing is paused until the temperature is lower than the temperature threshold before continuing printing.

[0106] Step 105, real-time collection of molten pool state and forming quality information of printed layers to obtain actual forming quality parameters, analysis of the actual forming quality parameters, dynamic adjustment of the laser power according to the deviation size and direction, and realization of 3D printing forming.

[0107] In this embodiment, the forming quality information at least includes the molten pool size, the solidification rate and the interlayer flatness.

[0108] Please refer to Figure 2 , the second embodiment of the laser power intelligent control method for 3D printing provided by the embodiment of the present application, which comprises:

[0109] Step 201, using a thermistor as a temperature sensor, measuring the voltage across the thermistor by applying a constant current to the thermistor, and obtaining the current ambient temperature according to Ohm's law, and continuously collecting temperature data.

[0110] Step 202: Using an optical power meter based on the photoelectric effect as an energy sensor, the laser scanning area is divided into a plurality of small squares. The optical power meter is quickly moved to the center of each square in turn and briefly stays there for 0.1 seconds to measure and record the laser power value at that point. This process is repeated repeatedly to construct an energy distribution matrix for the entire laser action area.

[0111] Step 203: Using a material composition monitor to monitor material composition changes in real time, using X-ray fluorescence spectroscopy for metal powder materials and infrared spectroscopy for organic polymer materials, and recording the material composition change data monitored each time;

[0112] Step 204: Acquire the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time to obtain original printing data.

[0113] See also Figure 3 , a schematic diagram of a third embodiment of a laser power intelligent control method for 3D printing provided by an embodiment of the present invention, the method comprising:

[0114] Step 301: Preprocess historical printing data to obtain sample data. Each sample data is organized into a multidimensional array containing model structure characteristics, environmental parameters, material properties, and the corresponding laser power value required for successful printing. The sample data is used for training and the Bat Algorithm is used to optimize and construct a prediction model.

[0115] In this embodiment, the model structural characteristics include at least the model volume, surface area, and layer thickness distribution; the environmental parameters include at least the printing environment temperature and humidity; and the material properties include at least the thermal conductivity, specific heat capacity, and the corresponding laser power value;

[0116] In this embodiment, the parameters of the initialized population include at least position, speed, frequency, loudness, and population size, and the first position value is a random initial solution generated using a uniform distribution; the fitness of each initial solution is calculated for the parameters of each individual bat, and the solution with the best fitness is found, where the objective function is the training error rate of the convolutional neural network; the pulse rate is the parameter set by the initialization, and the position and speed of the bat are updated in each iteration, and a new frequency is generated for each bat. If the random number is greater than the pulse rate, the bat will generate a new solution around the current optimal solution. If the fitness of the new solution is better than the current solution and the random number is less than the loudness, the new solution is taken; the fitness of the new solution is evaluated, and if the fitness of the new solution is better than the current best solution, the current best solution and the minimum fitness value are updated; when the maximum number of iterations is reached, the found optimal solution and the corresponding minimum fitness value are returned, and the bat algorithm is used to optimize the convolution kernel number hyperparameter of the convolutional neural network to construct a prediction model.

[0117] Step 302: In the convolution layer of the prediction model, the convolution kernel is slid over the input data to extract features. Through multiple layers of convolution and pooling, the data dimension is continuously reduced to extract higher-level features. Finally, these features are mapped to an output value through a fully connected layer to obtain the predicted laser power adjustment coefficient.

[0118] Step 303: Combine the energy density requirement obtained by the finite element analysis with the laser power adjustment coefficient output by the prediction model to calculate the final laser power prediction to obtain a prediction result.

[0119] See also Figure 4 , a schematic diagram of the structure of a laser power intelligent control system for 3D printing provided by an embodiment of the present invention, the system includes a database establishment module, a data acquisition module, a laser power prediction module, a printing operation module and a dynamic adjustment module, wherein,

[0120] Establish a database module to obtain the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establish a 3D model information database;

[0121] The data acquisition module is used to collect the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time through sensors arranged around the printing platform to obtain the original printing data;

[0122] The laser power prediction module is used to read the 3D model information database, combine it with the original printing data, and use the prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power and obtain the prediction result;

[0123] A printing operation module is used to use the preliminary laser power setting value for each layer and each area given by the prediction result as the initial power input, and drive the laser generator to start the printing operation of the current layer or area according to the setting value;

[0124] The dynamic adjustment module is used to collect the molten pool status and the molding quality information of the printed layers in real time to obtain the actual molding quality parameters, analyze the actual molding quality parameters, and dynamically adjust the laser power according to the deviation size and direction to achieve 3D printing molding. The molding quality information at least includes the molten pool size, solidification rate and interlayer flatness.

[0125] In this embodiment, the data acquisition module includes a temperature measurement submodule, an energy measurement submodule, a material monitoring submodule and an acquisition submodule, wherein:

[0126] The temperature measurement submodule uses a thermistor as a temperature sensor. By applying a constant current to the thermistor and measuring the voltage across the thermistor, the current ambient temperature is obtained according to Ohm's law, and temperature data is continuously collected.

[0127] The energy measurement submodule uses an optical power meter based on the photoelectric effect as an energy sensor to divide the laser scanning area into multiple small squares. The optical power meter quickly moves to the center of each square and briefly stops for 0.1 second to measure and record the laser power value at that point. This cycle is repeated to construct an energy distribution matrix for the entire laser action area.

[0128] The material monitoring submodule is used to monitor the changes in material composition in real time using a material composition monitor. For metal powder materials, X-ray fluorescence spectroscopy is used, and for organic polymer materials, infrared spectroscopy is used. The data of each monitored material composition change is recorded.

[0129] The acquisition submodule is used to obtain the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time to obtain the original printing data.

[0130] In this embodiment, the laser power prediction module includes a discretization processing submodule, an adoption submodule, a determination submodule, a first calculation submodule and a second calculation submodule, wherein:

[0131] The discretization processing submodule is used to discretize the 3D printing model, divide the 3D printing model into multiple small units, and assign corresponding thermal conductivity properties to each unit according to the thermophysical parameters of the 3D printing material;

[0132] A submodule is used to use laser as a point heat source input. According to the law of conservation of energy, within the time interval Δt, the heat entering the unit, the change in the unit's own stored heat, and the heat flowing out of the unit satisfy the following relationship:

[0133]

[0134] Where ρ is the density, c is the specific heat, V is the unit volume, and ΔT is the change in unit temperature within Δt. is the heat flux flowing in from j adjacent cells, m is the number of adjacent cells into which the heat flux flows, is the heat flux to k adjacent cells, n is the number of adjacent cells from which the heat flux flows, and Δx is the distance in the direction of heat conduction;

[0135] The determination submodule is used to obtain the energy distribution from the energy distribution of the laser action area in the printing original data, and determine the heat flow input of each unit at different times according to the laser scanning path, speed and energy distribution;

[0136] The first calculation submodule is used to simulate the heat transfer process in the model during laser scanning through iterative thermal balance, calculate the energy input required for different areas of the model to reach the expected melting temperature, and obtain the energy density requirement E corresponding to each unit.d :

[0137]

[0138] Where t is the time required to reach the melting temperature, is the heat entering the cell during the i-th time interval;

[0139] The second calculation submodule is used to obtain the calculated energy density requirement value to constitute the preliminary output result of the finite element analysis, determine the energy distribution required based on the physical heat conduction process, and perform laser power pre-calculation in combination with the prediction model to obtain the prediction result.

[0140] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A laser power intelligent control method for 3D printing, characterized in that: The laser power intelligent control method for 3D printing comprises the following steps: Obtain the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establish a 3D model information database; Sensors arranged around the printing platform collect information about the printing environment temperature, energy distribution in the laser action area, and material composition changes in real time to obtain the original printing data. Reading the 3D model information database, combining the original printing data, and using a prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power to obtain a prediction result; The initial laser power setting value for each layer or area given by the prediction result is used as the initial power input, and the laser generator is driven to start the printing operation of the current layer or area according to the setting value; Real-time acquisition of the melt pool status and the molding quality information of the printed layers to obtain the actual molding quality parameters, analyze the actual molding quality parameters, and dynamically adjust the laser power according to the deviation size and direction to achieve 3D printing. The molding quality information includes at least the melt pool size, solidification rate and interlayer flatness. The 3D model information database is read, combined with the original printing data, and a prediction model based on finite element analysis and convolutional neural network is used to pre-calculate the laser power to obtain a prediction result, including: Discretize the 3D printing model and divide it into multiple tiny units. Assign corresponding thermal conductivity properties to each unit according to the thermophysical parameters of the 3D printing material. Using laser as a point heat source input, according to the law of conservation of energy, within a time interval, the heat entering the unit, the change in the unit's own stored heat, and the heat flowing out of the unit satisfy the following relationship: ; in, is the density, is the specific heat capacity, is the unit volume, is the unit temperature at The amount of change within, It is from The heat flow into the adjacent unit, is the number of adjacent cells into which the heat flow flows, Is the flow direction The heat flow of adjacent cells, is the number of neighboring cells with outgoing heat flux, is the distance in the direction of heat conduction; Obtain energy distribution from the energy distribution of the laser action area in the original printing data, and determine the heat flow input of each unit at different times based on the laser scanning path, speed and energy distribution; By iteratively simulating the heat transfer process within the model during laser scanning, the energy input required to reach the expected melting temperature in different areas of the model is calculated, and the energy density requirement corresponding to each unit is obtained. : ； Where t is the time required to reach the melting temperature, It is in The amount of heat entering the unit during a time interval; The calculated energy density requirement value constitutes the preliminary output result of the finite element analysis, determines the energy distribution required based on the physical heat conduction process, and combines the prediction model to pre-calculate the laser power to obtain the prediction result.

2. The laser power intelligent control method for 3D printing according to claim 1, wherein: The sensors arranged around the printing platform collect information about the printing environment temperature, energy distribution in the laser action area, and material composition changes in real time to obtain the original printing data, including: A thermistor is used as a temperature sensor. By applying a constant current to the thermistor and measuring the voltage across the thermistor, the current ambient temperature is obtained according to Ohm's law, and temperature data is continuously collected. An optical power meter based on the photoelectric effect is used as an energy sensor. The laser scanning area is divided into multiple small squares. The optical power meter quickly moves to the center of each square and briefly stops for 0.1 second. The laser power value at that point is measured and recorded. This cycle is repeated to construct an energy distribution matrix for the entire laser action area. Use material composition monitors to monitor material composition changes in real time. For metal powder materials, X-ray fluorescence spectroscopy is used, and for organic polymer materials, infrared spectroscopy is used. The material composition change data monitored each time is recorded. The printing environment temperature, energy distribution in the laser action area, and material composition change information are acquired in real time to obtain the original printing data.

3. The laser power intelligent control method for 3D printing according to claim 1, wherein: The energy density requirement value obtained and calculated constitutes the preliminary output result of the finite element analysis, determines the energy distribution required based on the physical heat conduction process, and performs laser power pre-calculation in combination with the prediction model to obtain the prediction result, including: Preprocess historical printing data to obtain sample data. Each sample data is organized into a multidimensional array containing model structural characteristics, environmental parameters, material properties, and the corresponding laser power value required for successful printing. A prediction model is trained using the sample data and optimized using the Bat Algorithm. The model structural characteristics include at least the model volume, surface area, and layer thickness distribution; the environmental parameters include at least the printing environment temperature and humidity; and the material properties include at least the thermal conductivity, specific heat capacity, and the corresponding laser power value. In the convolution layer of the prediction model, the convolution kernel is used to slide on the input data to extract features. Through multiple layers of convolution and pooling, the data dimension is continuously reduced and more advanced features are extracted. Finally, these features are mapped to an output value through the fully connected layer to obtain the predicted laser power adjustment coefficient. The energy density requirement obtained by finite element analysis is combined with the laser power adjustment coefficient output by the prediction model to calculate the final laser power prediction and obtain the prediction result.

4. The laser power intelligent control method for 3D printing according to claim 3, wherein: The method of training with sample data and optimizing the prediction model using the bat algorithm includes: Initialize the parameters of the population, including at least position, speed, frequency, loudness, and population size. The first position value is a random initial solution generated using a uniform distribution. Calculate the fitness of each initial solution for the parameters of each individual bat and find the solution with the best fitness, where the objective function is the training error rate of the convolutional neural network; The pulse rate is the parameter set by the initialization. The bat's position and speed are updated in each iteration, and a new frequency is generated for each bat. If the random number is greater than the pulse rate, the bat will generate a new solution around the current optimal solution. If the fitness of the new solution is better than the current solution and the random number is less than the loudness, the new solution is taken. Evaluate the fitness of the new solution. If the fitness of the new solution is better than the current best solution, update the current best solution and minimum fitness value. When the maximum number of iterations is reached, the optimal solution and the corresponding minimum fitness value are returned. The bat algorithm is used to optimize the hyperparameter of the number of convolution kernels of the convolutional neural network to build a prediction model.

5. The laser power intelligent control method for 3D printing according to claim 1, wherein: The method of using the initial laser power setting value for each layer or each area given by the prediction result as the initial power input and driving the laser generator to start the printing operation of the current layer or area according to the setting value includes: Extracting preliminary laser power setting values ​​corresponding to the current printing layer and each area within the layer from the prediction results; The laser power is controlled by adjusting the applied voltage to control the output light intensity. The required laser power setting value is known. , and the light-to-electricity conversion efficiency of the laser generator , according to the relationship between power and light intensity, the required input light intensity is obtained : ； in, is the cross-sectional area of ​​the laser beam; Then, according to the Pockels effect formula, the corresponding applied voltage is calculated : ； in, is the half-wave voltage, is the reference light intensity, Convert the light intensity ratio into a corresponding voltage value; The calculated voltage value is applied to the electro-optical modulation element of the laser generator, driving the laser generator to output laser according to the power setting value and start printing the current layer or area.

6. The laser power intelligent control method for 3D printing according to claim 1, wherein: The method further includes: using the initial laser power setting value for each layer or each area given by the prediction result as the initial power input, and driving the laser generator to start the printing operation of the current layer or area according to the setting value; When driving the laser generator to start printing, a synchronous clock signal is started. If the scanning speed of the printing device is , with a resolution of , then the synchronous clock signal frequency Can be set to: ； During the printing process, sensors are used to continuously collect real-time data, including at least the actual laser output power and the ambient temperature of the current printing area. The power error is calculated by comparing the actual output power with the set power. Determine whether the power error exceeds the preset allowable error range. If so, trigger the alarm mechanism and perform power compensation according to the error size; Monitor the ambient temperature of the printing area. If the temperature exceeds the threshold, printing is paused and the printing process resumes after the temperature drops below the threshold.

7. Laser power intelligent control system for 3D printing, characterized in that: The laser power intelligent control system for 3D printing includes a database establishment module, a data acquisition module, a laser power prediction module, a printing operation module and a dynamic adjustment module, wherein: Establish a database module to obtain the thermophysical parameters of 3D printing materials and the structural characteristics of different types of 3D models, and establish a 3D model information database; The data acquisition module is used to collect the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time through sensors arranged around the printing platform to obtain the original printing data; A laser power prediction module is used to read the 3D model information database, combine the original printing data, and use a prediction model based on finite element analysis and convolutional neural network to pre-calculate the laser power to obtain a prediction result; A printing operation module is used to use the preliminary laser power setting value for each layer and each area given by the prediction result as the initial power input, and drive the laser generator to start the printing operation of the current layer or area according to the setting value; The dynamic adjustment module is used to collect the molten pool status and the molding quality information of the printed layers in real time to obtain the actual molding quality parameters, analyze the actual molding quality parameters, and dynamically adjust the laser power according to the deviation size and direction to achieve 3D printing molding. The molding quality information includes at least the molten pool size, solidification rate and interlayer flatness; The laser power prediction module includes a discretization processing submodule, an adoption submodule, a determination submodule, a first calculation submodule and a second calculation submodule, wherein: The discretization processing submodule is used to discretize the 3D printing model, divide the 3D printing model into multiple small units, and assign corresponding thermal conductivity properties to each unit according to the thermophysical parameters of the 3D printing material; The submodule is used to use laser as a point heat source input, according to the law of conservation of energy, at time intervals The heat entering the unit, the change in the unit's own stored heat, and the heat flowing out of the unit satisfy the following relationship: ； ； in, is the density, is the specific heat capacity, is the unit volume, is the unit temperature at The amount of change within, It is from The heat flow into the adjacent unit, is the number of adjacent cells into which the heat flow flows, Is the flow direction The heat flow of adjacent cells, is the number of neighboring cells with outgoing heat flux, is the distance in the direction of heat conduction; The determination submodule is used to obtain the energy distribution from the energy distribution of the laser action area in the printing original data, and determine the heat flow input of each unit at different times according to the laser scanning path, speed and energy distribution; The first calculation submodule is used to simulate the heat transfer process in the model during laser scanning through iterative thermal balance, calculate the energy input required for different areas of the model to reach the expected melting temperature, and obtain the energy density requirement corresponding to each unit. : ； Where t is the time required to reach the melting temperature, It is in The amount of heat entering the unit during a time interval; The second calculation submodule is used to obtain the calculated energy density requirement value to constitute the preliminary output result of the finite element analysis, determine the energy distribution required based on the physical heat conduction process, and perform laser power pre-calculation in combination with the prediction model to obtain the prediction result.

8. The laser power intelligent control system for 3D printing according to claim 7, characterized in that: The data acquisition module includes a temperature measurement submodule, an energy measurement submodule, a material monitoring submodule and an acquisition submodule, wherein: The temperature measurement submodule uses a thermistor as a temperature sensor. By applying a constant current to the thermistor and measuring the voltage across the thermistor, the current ambient temperature is obtained according to Ohm's law, and temperature data is continuously collected. The energy measurement submodule uses an optical power meter based on the photoelectric effect as an energy sensor. The laser scanning area is divided into multiple small squares. The optical power meter quickly moves to the center of each square and briefly stops for 0.1 second. It then measures and records the laser power value at that point. This cycle is repeated to construct an energy distribution matrix for the entire laser action area. The material monitoring submodule is used to monitor the changes in material composition in real time using a material composition monitor. For metal powder materials, X-ray fluorescence spectroscopy is used, and for organic polymer materials, infrared spectroscopy is used. The data of each monitored material composition change is recorded. The acquisition submodule is used to obtain the printing environment temperature, energy distribution in the laser action area, and material composition change information in real time to obtain the original printing data.

Images