Method and system for predicting height of laser metal directional energy deposition molten pool based on ANN deep learning

Abstract

The invention provides a laser metal directional energy deposition molten pool height prediction method and system based on ANN deep learning, and the method comprises the steps: firstly constructing a training data set for the height prediction training of a laser metal directional energy deposition molten pool, and enabling the input data to comprise technological parameters, process parameters and molten pool height; the process parameters comprise laser power, scanning speed and powder feeding amount, the process parameter is the width of the molten pool, then a molten pool height prediction model based on the ANN network is obtained through machine learning, the laser power, the scanning speed, the powder feeding amount and the width of the molten pool of the actual process serve as input quantities, the trained molten pool height prediction model is input, and then the height of the molten pool can be output. According to the method, the features highly related to the prediction target are selected, redundant or irrelevant features are removed, the risk of overfitting is reduced, the generalization ability of the molten pool height prediction model is improved, the width of the molten pool serves as the input feature of deep learning, and the prediction precision and the generalization ability of the model are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of laser metal directional deposition, in particular to parameter prediction of a molten pool, and specifically to a method and system for predicting the molten pool height of laser metal directional energy deposition based on ANN deep learning. Background Art

[0002] In the laser metal directed energy deposition process, the stability of the molten pool determines the stability of the material forming process to a large extent. Accurately predicting the molten pool morphology is a prerequisite for ensuring the quality of laser cladding. The molten pool height characteristic is one of the important characteristics of the molten pool during the deposition process. It is very important to predict the height of the molten pool during the deposition process.

[0003] The laser metal directional energy deposition process involves complex dynamic changes such as heat transfer, flow, deformation, melting and solidification. It is difficult to detect the process even if high-precision and high-speed imagers are used for real-time dynamic monitoring. In addition, the melt channel size and forming accuracy are also affected by many factors.

[0004] At present, the relationship model between deposition process parameters and melt channel size includes methods based on numerical simulation and artificial intelligence. The method based on numerical simulation has been relatively mature after long-term development, but it needs to solve complex partial differential equations, especially the laser energy, absorption rate, powder flow concentration and other parameters that are difficult to accurately measure must be reflected in the boundary conditions; in the prediction of intelligent algorithms such as neural networks, the aforementioned laser energy, absorption rate, powder flow concentration and other parameters that are difficult to accurately measure are not used as inputs to the model, which makes the prediction results of the intelligent algorithm less uncertain than the numerical simulation results. Summary of the invention

[0005] The purpose of the present invention is to provide a method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning. ANN network model training is performed based on the process parameters and process parameters of laser metal directional deposition to obtain a molten pool height prediction model, which greatly improves the prediction accuracy and generalization ability of the model and can realize accurate prediction of the molten pool height in the deposition morphology characteristics.

[0006] According to a first aspect of the present invention, a method for predicting the height of a molten pool of laser metal directional deposition based on ANN deep learning is proposed, comprising the following steps:

[0007] Constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set includes multiple groups of laser metal directional deposition processing data sources, each group of data sources consists of process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters include laser power, scanning speed and powder feeding amount, and the process parameter is molten pool width;

[0008] Based on the training data set, an input-output relationship model of an ANN neural network is trained to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; and

[0009] According to the ANN network-based molten pool height prediction model, the output molten pool height is predicted by taking the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process and the molten pool width obtained by the corresponding deposition processing as input quantities.

[0010] According to the second aspect of the present invention, a laser metal directional deposition molten pool height prediction system based on ANN deep learning is also proposed, which comprises:

[0011] A training data construction module for constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set comprising a plurality of groups of laser metal directional deposition processing data sources, each group of data sources comprising process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters comprising laser power, scanning speed and powder feeding amount, and the process parameter being molten pool width;

[0012] A molten pool height prediction model training module for training an input-output relationship model using an ANN neural network based on the training data set to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; and

[0013] A molten pool height prediction output module is used to predict the output molten pool height according to the molten pool height prediction model based on the ANN network, taking the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process, and the molten pool width obtained by the corresponding deposition processing as input.

[0014] According to a third aspect of the present invention, a computer system is also provided, comprising:

[0015] one or more processors; and

[0016] Memory, storing instructions that can be operated;

[0017] Among them, when the instructions are executed by one or more processors, they can enable the one or more processors to execute the process of the laser metal directional deposition molten pool height prediction method based on ANN deep learning in the aforementioned embodiment.

[0018] It should be understood that all combinations of the foregoing concepts and additional concepts described in greater detail below are considered to be part of the inventive subject matter of the present disclosure, provided that such concepts are not mutually inconsistent. Additionally, all combinations of the claimed subject matter are considered to be part of the inventive subject matter of the present disclosure.

[0019] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description taken in conjunction with the accompanying drawings. Other additional aspects of the present invention, such as the features and / or advantageous effects of exemplary embodiments, will be apparent from the following description or will be learned from the practice of specific embodiments in accordance with the teachings of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in various figures may be represented by the same reference numeral. For clarity, not every component is labeled in each figure. Now, embodiments of various aspects of the present invention will be described by way of example and with reference to the drawings.

[0021] Figure 1 is a schematic flowchart of a method for predicting the molten pool height in laser metal direct deposition based on ANN deep learning according to an embodiment of the present invention.

[0022] Figure 2 is a schematic diagram of obtaining the molten pool width according to an embodiment of the present invention.

[0023] Figure 3 is a schematic diagram of the network structure of a molten pool height prediction model based on an ANN network according to an embodiment of the present invention.

[0024] Figure 4 is a schematic diagram of the training process of a molten pool height prediction model based on an ANN network according to an embodiment of the present invention.

[0025] Figure 5 is a schematic diagram of the modules of a system for predicting the molten pool height in laser metal direct deposition based on ANN deep learning according to an embodiment of the present invention.

[0026] Figure 6 is a schematic diagram of the predicted value and relative error of a molten pool height prediction model based on an ANN network according to an embodiment of the present invention.

[0027] Figure 7 is a schematic diagram of the predicted value and relative error using a conventional network. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] To better understand the technical content of the present invention, specific embodiments are given below in conjunction with the accompanying drawings for illustration.

[0029] Various aspects of the present invention are described in this disclosure with reference to the accompanying drawings, in which many illustrative embodiments are shown. The embodiments of the present disclosure are not necessarily intended to include all aspects of the present invention. It should be understood that the various concepts and embodiments introduced above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed by the present invention are not limited to any implementation. In addition, some aspects disclosed by the present invention can be used alone or in any appropriate combination with other aspects disclosed by the present invention.

[0030] [Prediction method of molten pool height in laser metal directional deposition based on ANN deep learning]

[0031] In combination with the embodiments of the present invention, Figure 1 The example shown is a method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning, comprising the following steps:

[0032] S101: constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set comprising a plurality of groups of laser metal directional deposition processing data sources, each group of data sources comprising process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters comprising laser power, scanning speed and powder feeding amount, and the process parameter being molten pool width;

[0033] S102: training an input-output relationship model of an ANN neural network based on the training data set to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; and

[0034] S103: According to the ANN network-based molten pool height prediction model, the output molten pool height is predicted using the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process, and the molten pool width obtained by the corresponding deposition process as input.

[0035] In an optional embodiment, in the aforementioned step S101, the laser power, scanning speed and powder feeding amount in the process parameters can be determined according to the test process. The molten pool width as a process parameter is obtained by visual recognition of the molten pool image, for example, including but not limited to the applicant's self-developed method for extracting the molten pool width based on contour search (publication number CN116797645A), the method for extracting the molten pool width of laser metal directional energy deposition based on edge corrosion (publication number CN116029997A), etc.

[0036] In an optional embodiment, an input-output relationship model using an ANN network structure is trained based on the training data set to obtain a molten pool height prediction model based on an ANN network, including:

[0037] Normalize the training data set so that the training data range is between [-1, 1];

[0038] The normalized data is input into the ANN neural network for training. The ANN neural network adopts two hidden layers. The number of nodes is calculated to be 4 nodes and 5 nodes respectively. The input layer includes normalized laser power, scanning speed and powder feeding amount. The process parameter is the molten pool width, and the output layer is the molten pool height. The ANN neural network adopts the linear unit ReLU function as the activation function, and obtains the neural network structure of 4 nodes in the first layer and 5 nodes in the second layer through training.

[0039] On the basis of the obtained double-layer neural network structure, the initial network is iterated, and finally a melt pool height prediction model based on the ANN network is obtained.

[0040] In an optional embodiment, the aforementioned normalization preprocessing of the training data set includes:

[0041] The following normalization formula is used for processing:

[0042]

[0043] Among them, X n is the normalized sample, X is the original sample, X max and X min are the maximum and minimum values ​​of each data variable respectively.

[0044] In an optional embodiment, during the model network iteration process, the number of configuration iterations is set to N, N is greater than or equal to 200 times, the target error is set to 0.001, and the learning efficiency is 0.005.

[0045] {Example 1}

[0046] In order to better understand the implementation process of the method of the above embodiment of the present invention, it is further described below in conjunction with specific embodiments.

[0047] Data Source

[0048] The data source, whether it is the data source used for training or the data source used for predictive output of the trained model, includes two parts, namely: process parameters + process parameters.

[0049] The process parameters include laser power, scanning speed, and powder feeding amount.

[0050] The process parameter is the molten pool width.

[0051] In this example, the molten pool width of the process parameter is obtained by collecting the molten pool image by the coaxial vision acquisition system and through computer vision recognition, such as the molten pool width extraction method based on contour search and the laser metal directed energy deposition molten pool width extraction method based on edge corrosion to achieve the extraction of the molten pool width.

[0052] Data source acquisition

[0053] The relationship model between input and output is established by machine learning. Machine learning training is required based on sample data. Sample data is obtained through orthogonal experiments, and deposition tests are carried out. The test preliminarily determines the process range based on powder and equipment.

[0054] As an example, 316l powder was selected, and the test process parameter range was: power 300-700W, scanning speed 480-720mm / min, powder feeding amount 0.24-0.45g / s. An orthogonal test was designed, with the laser power interval of 100W, the scanning speed interval of 60mm / min, and the powder feeding amount interval of 0.05g / s. There were 30 groups in total, of which 25 groups of test data were as follows:

[0055]

[0056] Combination Figure 2 As shown in Figure 1, the molten pool width acquisition test system consists of a laser (IPG YLR-1000-WC), a processing head (RCND-26), a CNC system (Siemens 420D), a powder feeder (RC-PF-01B-2) and supporting components.

[0057] like Figure 2 As shown, the laser is connected to the processing head by an optical fiber, and the metal powder is transported to the powder tube by a powder feeder, and four-way coaxial powder feeding is used. The detection module consists of a computer, a camera module, and an optical path system.

[0058] The camera module includes a CMOS camera, a zoom lens, an attenuation plate and a filter. The camera is installed in a coaxial manner, which has the advantages of high device integration and no deformation of the collected image. A bandpass filter (pass wavelength 540±10nm) and a neutral density filter (transmittance 5%) are arranged at the front end of the camera. The bandpass filter can reduce the interference of redundant band radiation, and the neutral density filter is used to weaken the reflected light intensity to prevent the reflected light from exceeding the dynamic acquisition range of the camera.

[0059] The optical path system is divided into forward optical path and reverse optical path, which are split by a beam splitter. The forward direction can pass lasers of 900-1070nm, and the reverse direction can pass visible light of 520-720nm. The reflected light from the molten pool reaches the camera module through the reverse optical path. The computer performs graphic recognition processing on the collected images to obtain the molten pool width. After testing, the data collection error is less than 5%.

[0060] Model Training

[0061] like Figure 4 As shown, the process of ANN model training is exemplarily represented.

[0062] In order to improve the performance of the model, the data is first normalized so that the data range falls between [-1, 1].

[0063] As an example, the normalization formula used is as follows:

[0064]

[0065] Among them, X n is the normalized sample, X is the original sample, X max and X min are the maximum and minimum values ​​of each data variable respectively.

[0066] The normalized values ​​are input into the neural network model.

[0067] In the example of the present invention, the machine learning algorithm is implemented using Python 3.7, and the weights and biases of the OvNN model are initialized using the Keras internal random initialization method.

[0068] In order to avoid the gradient vanishing problem of the logistic function (Sigmoid) and the hyperbolic tangent function (Tanh), this example uses the linear unit (ReLU) function as the activation function of the OvNN network.

[0069] Due to the relatively small amount of data, 25 groups of data were used for training and learning. The neurons were determined by trial and error until the optimal performance was obtained. The final test result was a double-layer 8-node neuron final network structure.

[0070] It should be understood that in other embodiments, more sample data can be obtained through experiments to participate in training and learning.

[0071] After determining the 8-node network structure, the initial network was iterated, the number of iterations was set to 500, the target error was set to 0.001, and the learning efficiency was set to 0.005.

[0072] The model obtained after iteration is used as the final melt pool prediction model. Figure 6 The comparison of the prediction results and the relative error level shown in the figure shows that the average error is 4.03%, which has a high accuracy rate.

[0073] In combination with the above embodiments of the present invention and the specific exemplary implementation process, it can be seen that the laser metal directional deposition molten pool height prediction method based on ANN deep learning proposed in the present invention not only considers the three main process parameters (laser power, scanning speed, powder feeding amount), but also considers the molten pool width characteristics in the deposition morphology characteristics. In the model training process of the present invention and the final molten pool height prediction model, the addition of the molten pool width feature greatly increases the correlation between the input data and the output data of the ANN network.

[0074] Table 1 below is a triangle line table of correlation between parameters obtained by Pearson's method. It can be found from Table 1 that there is a correlation of 0.505 between the molten pool width and the molten pool height, and its correlation value even exceeds the correlation between the scanning speed and the molten pool height. In the deep learning process, features that are highly correlated with the predicted target are selected, and redundant or irrelevant features are removed, which helps to reduce the risk of overfitting and improve the generalization ability of the molten pool height prediction model. Therefore, in an embodiment of the present invention, using the molten pool width as an input feature for deep learning can greatly increase the prediction accuracy and generalization ability of the model.

[0075] Table 1 Feature correlation triangle table

[0076]

[0077] Combination Figure 6 , 7 As shown, all are under the condition of ANN network model. Figure 6 The figure shows the predicted value and relative error comparison of the molten pool height prediction model constructed according to the method of the present invention. Figure 7 It is the average error data between the conventional input (laser power, scanning speed, powder feeding amount) and the molten pool height output, and the average error is 17.85%, which is much higher than the method proposed in the present invention.

[0078] [Laser metal directional deposition melt pool height prediction system based on ANN deep learning]

[0079] In combination with the above embodiments, Figure 5 The example of the ANN deep learning-based laser metal directional deposition melt pool height prediction system shown includes:

[0080] A training data construction module for constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set comprising a plurality of groups of laser metal directional deposition processing data sources, each group of data sources comprising process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters comprising laser power, scanning speed and powder feeding amount, and the process parameter being molten pool width;

[0081] A molten pool height prediction model training module for training an input-output relationship model using an ANN neural network based on the training data set to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; and

[0082] A molten pool height prediction output module is used to predict the output molten pool height according to the molten pool height prediction model based on the ANN network, taking the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process, and the molten pool width obtained by the corresponding deposition processing as input.

[0083] Among them, the molten pool height prediction model training module is configured to include:

[0084] The preprocessing module is used to normalize the training data set so that the training data range is between [-1, 1];

[0085] A model training module is used to input the normalized data into an ANN neural network for training. The ANN neural network uses two hidden layers. The number of nodes is calculated to be 4 nodes and 5 nodes respectively. The input layer includes normalized laser power, scanning speed and powder feeding amount. The process parameter is the molten pool width, and the output layer is the molten pool height. The ANN neural network uses a linear unit ReLU function as an activation function, and a neural network structure with 4 nodes in the first layer and 5 nodes in the second layer is obtained through training.

[0086] The model iteration module is used to iterate the initial network based on the obtained double-layer neural network structure, and finally obtain the melt pool height prediction model based on the ANN network.

[0087] The preprocessing module is configured to perform normalization preprocessing on the training data set in the following manner:

[0088] The following normalization formula is used for processing:

[0089]

[0090] Among them, X n is the normalized sample, X is the original sample, X max and Xmin are the maximum and minimum values ​​of each data variable respectively.

[0091] The model iteration module is configured to iterate in the following configuration:

[0092] The number of iterations is set to N, N is greater than or equal to 100 times, the target error is set to 0.001, and the learning efficiency is set to 0.005.

[0093] [Computer System]

[0094] According to an embodiment of the present invention, a computer system is further provided, comprising:

[0095] one or more processors; and

[0096] Memory stores instructions that can be operated.

[0097] Among them, when the instructions are executed by one or more processors, they can enable the one or more processors to execute the process of the laser metal directional deposition molten pool height prediction method based on ANN deep learning in any of the aforementioned embodiments.

[0098] [Computer readable storage medium]

[0099] According to an embodiment of the present invention, a non-volatile computer-readable storage medium is also provided, comprising one or more programs for being executed by one or more processors of a computer system, wherein the one or more programs comprise instructions or instruction sets.

[0100] When these instructions or instruction sets are executed by one or more processors, the computer system executes the process of the laser metal directional deposition molten pool height prediction method based on ANN deep learning in any of the aforementioned embodiments.

[0101] It should be understood that the aforementioned computer system may be implemented as a personal computer system, a commercial computer system, or may be configured as a server or a cloud server.

[0102] Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. A person with ordinary knowledge in the technical field to which the present invention belongs may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be determined by the definition of the claims.

Claims

1. A method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning, It is characterized in that The following steps are involved: Constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set includes multiple groups of laser metal directional deposition processing data sources, each group of data sources consists of process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters include laser power, scanning speed and powder feeding amount, and the process parameter is molten pool width; Based on the training data set, an input-output relationship model of an ANN neural network is trained to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; as well as According to the ANN network-based molten pool height prediction model, the output molten pool height is predicted by taking the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process and the molten pool width obtained by the corresponding deposition processing as input quantities.

2. According to the ANN deep learning-based laser metal directional deposition molten pool height prediction method of claim 1, It is characterized in that The step of training the input-output relationship model using the ANN network structure based on the training data set to obtain the molten pool height prediction model based on the ANN network includes: Normalize the training data set so that the training data range is between [-1, 1]; The normalized data is input into the ANN neural network for training. The ANN neural network adopts two hidden layers. The number of nodes is calculated to be 4 nodes and 5 nodes respectively. The input layer includes the normalized laser power, scanning speed and powder feeding amount. The process parameter is the molten pool width, and the output layer is the molten pool height. The ANN neural network uses the linear unit ReLU function as an activation function, and obtains a neural network structure with 4 nodes in the first layer and 5 nodes in the second layer through training; On the basis of the obtained double-layer node neural network structure, the initial network is iterated, and finally the melt pool height prediction model based on the ANN network is obtained.

3. According to claim 2, the method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning, It is characterized in that The normalization preprocessing of the training data set includes: The following normalization formula is used for processing: Among them, X n is the normalized sample, X is the original sample, X max and X min are the maximum and minimum values ​​of each data variable respectively.

4. According to claim 2, the method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning, It is characterized in that The number of iterations is set to N, N is greater than or equal to 200 times, the target error is set to 0.001, and the learning efficiency is set to 0.

005.

5. According to claim 2, the method for predicting the molten pool height of laser metal directional deposition based on ANN deep learning, It is characterized in that The invention discloses a training data set for predicting the molten pool height of laser metal directional deposition, wherein the molten pool width as a process parameter is set to be obtained by visual recognition of the molten pool image.

6. A laser metal directional deposition molten pool height prediction system based on ANN deep learning, It is characterized in that include: A training data construction module for constructing a training data set for laser metal directional deposition molten pool height prediction training, the training data set comprising a plurality of groups of laser metal directional deposition processing data sources, each group of data sources comprising process parameters, process parameters obtained by processing according to the process parameters, and molten pool height obtained by processing according to the process parameters, the process parameters comprising laser power, scanning speed and powder feeding amount, and the process parameter being molten pool width; A molten pool height prediction model training module for training an input-output relationship model using an ANN neural network based on the training data set to obtain a molten pool height prediction model based on the ANN network, wherein the input is laser power, scanning speed, powder feeding amount and molten pool width, and the output is molten pool height; as well as A molten pool height prediction output module is used to predict the output molten pool height according to the molten pool height prediction model based on the ANN network, taking the laser power, scanning speed and powder feeding amount in the laser metal directional deposition process, and the molten pool width obtained by the corresponding deposition processing as input.

7. The laser metal directional deposition molten pool height prediction system based on ANN deep learning according to claim 6, It is characterized in that The molten pool height prediction model training module is configured to include: The preprocessing module is used to normalize the training data set so that the training data range is between [-1, 1]; The model training module is used to input the normalized data into the ANN neural network for training. The ANN neural network adopts two hidden layers, and the number of nodes is calculated to be 4 nodes and 5 nodes respectively; the input layer includes normalized laser power, scanning speed and powder feeding amount, the process parameter is the molten pool width, and the output layer is the molten pool height; wherein the ANN neural network adopts the linear unit ReLU function as the activation function, and obtains the neural network structure of 4 nodes in the first layer and 5 nodes in the second layer through training; The model iteration module is used to iterate the initial network based on the obtained double-layer neural network structure, and finally obtain the melt pool height prediction model based on the ANN network.

8. The laser metal directional deposition molten pool height prediction system based on ANN deep learning according to claim 7, It is characterized in that The preprocessing module is configured to perform normalization preprocessing on the training data set in the following manner: The following normalization formula is used for processing: Among them, X n is the normalized sample, X is the original sample, X max and X min are the maximum and minimum values ​​of each data variable respectively.

9. The laser metal directional deposition molten pool height prediction system based on ANN deep learning according to claim 7, It is characterized in that The model iteration module is configured to iterate in the following configuration: The number of iterations is set to N, N is greater than or equal to 200 times, the target error is set to 0.001, and the learning efficiency is set to 0.

005.

10. A computer system, It is characterized in that include: one or more processors; as well as Memory, storing instructions that can be operated; Wherein, when the instruction is executed by one or more processors, it can enable the one or more processors to execute the process of the laser metal directional deposition molten pool height prediction method based on ANN deep learning as described in any one of claims 1-5.

Images