CN115945760A - Collapse compensation method for arc blowout part of single-channel multilayer welding seam manufactured by electric arc additive manufacturing - Google Patents

Abstract

The invention discloses a collapse compensation method for an arc blowout position of a single-channel multilayer welding seam in electric arc additive manufacturing, and relates to the field of electric arc additive manufacturing, wherein the method comprises the following steps: determining corresponding welding process parameters according to the welding equipment, the welding material and the welding method; designing a single-pass multilayer weld joint experiment according to corresponding welding process parameters; constructing a deep neural network which takes process parameters as input, a difference value between the middle residual height of the welding seam and the residual height at the arc quenching position and the collapse compensation length as output, and training and evaluating the neural network by using data obtained by experiments to obtain a collapse compensation length prediction model at the arc quenching position of the welding seam; parameters in the welding seam forming process are collected and input into a welding seam compensation prediction model after training is completed, and corresponding height difference and compensation length are output, so that accurate prediction of collapse compensation length of an arc blowout part of the electric arc additive manufacturing single-channel multi-layer welding seam is achieved. The method effectively overcomes the defect of collapse at the arc quenching position, and the high flatness of the component is improved through test verification.

Description

Arc additive manufacturing single-channel multilayer weld seam arc-quenching part collapse compensation method

Technical Field

The invention relates to the field of electric arc additive manufacturing, in particular to a collapse compensation method for an arc quenching position of a single-channel multi-layer welding seam in electric arc additive manufacturing.

Background

With the rapid iterative development of the manufacturing industry, the requirements on metal components are higher and higher, and the existing traditional manufacturing technology is difficult to meet the requirements of the current industrial development, so that the additive manufacturing technology is proposed and widely applied to various fields of the manufacturing industry. The additive manufacturing technology is to stack materials layer by layer from the bottom under the drive of a motion execution structure according to a planned running track, and is a layer-by-layer accumulated superposition manufacturing technology from scratch. It integrates many emerging technologies including machining, digital information, computer and materials science. Compared with the traditional manufacturing technology, the device has the advantages of simple device, wide used materials, unlimited molding size, high-efficiency and quick production and the like. Meanwhile, the method is not limited by the traditional manufacturing technology, and the manufacture of products with small quantity and complex structure can be rapidly finished in a short time.

At present, a plurality of technical problems exist in the field of electric arc additive manufacturing at home and abroad, and need to be deeply researched, for example, as the accumulated welding line geometric parameters are changed due to the influence of factors such as temperature and heat dissipation conditions during continuous accumulation, the problems of complicated path programming, stress concentration of a formed part, poor forming precision, increased surface roughness and the like occur. The common problem is that when a single-pass multilayer accumulation experiment is carried out, obvious collapse easily occurs at the arc quenching part of the welding line. The main reason is that at the arc blowout position, due to the influence of the movement inertia of the welding gun and the acting force of the protective gas, the liquid metal in a high-temperature state cannot be immediately solidified and can continuously flow for a short distance along the movement direction of the welding gun, and meanwhile, the electric arc is extinguished without being filled with cladding metal, so that the forming height at the tail end of the welding bead is reduced, namely, the arc blowout position has the collapse phenomenon. If do not take corresponding measure to this, not only can be unfavorable for to experiment forming piece appearance and accuracy control, can lead to moreover that the multilayer welding bead piles up the experiment in local welding wire dry extension and surpasss suitable scope gradually to appear that protective gas can't normally go on cladding metal's guard action, arc starting splash, welder hit rifle because pile up metal slope scheduling problem.

In order to solve the problem of collapse of an arc quenching part in a single-channel multi-layer welding bead forming piece, xiong Jun of Harbin university of industry provides a parallel reciprocating cladding mode in the research of the influence of GAM additive manufacturing technology on the single-channel multi-layer forming appearance; the He Jianbin of Xinjiang university provides an alternative path in the process research of 5356 aluminum alloy arc additive manufacturing to solve the problem of large height difference between an arc extinguishing end and an arc striking end; the Nanjing university of science and engineering Yin Fan provides a fixed distance compensation method in the research of multilayer stacking forming process and size control to ensure forming; in the morning of Xiao Yu of Yanshan university, a double-layer parallel reciprocating welding mode is obtained in the welding research of a thin-wall to solve the problem of collapse; in summary, it can be summarized that, in the solutions proposed by a large number of typical scholars, most scholars adopt a parallel reciprocating stacking mode or an improved method based on parallel reciprocating, such as a double-layer parallel reciprocating welding mode, for the problem that the arc quenching end collapses in the single-pass multi-layer weld bead forming process, but the existing solutions improved by parallel reciprocating or parallel reciprocating cannot accurately fill up the defect, and the phenomenon that the middle is high and the ends are low gradually occurs along with the increase of the number of stacked layers often occurs.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a collapse compensation method for an arc blowout position of a single-channel multilayer welding seam in electric arc additive manufacturing, which is based on the establishment of a neural network to predict process parameters, and carries out collapse compensation in a mode that the arc blowout position is not put out of arc and then is put out of arc after being welded back for a certain distance during welding of each layer, so that the problem of collapse of the arc blowout position in the single-layer multichannel forming process can be effectively solved, the high smoothness of finished parts in electric arc additive manufacturing is ensured, and the problem that the new appearance characteristics of the parts are influenced due to the collapse of the arc blowout position is solved.

In order to achieve the purpose, the invention can adopt the following technical scheme:

a collapse compensation method for an arc quenching position of an electric arc additive manufacturing single-channel multilayer weld comprises the following steps:

designing a corresponding welding process according to a welding material, a structural form and a welding method;

designing corresponding process experiments according to the welding process, and implementing the process experiments in batches to further obtain process experiment data; the process experimental data comprise welding current, welding voltage, welding speed, welding seam forming size and collapse compensation distance at the arc quenching position of the welding seam;

establishing a collapse compensation distance prediction data set at the arc quenching position of the welding seam according to a data combination mode of the welding current, the welding voltage, the welding speed, the welding seam forming size and the collapse compensation distance at the arc quenching position of the welding seam;

constructing a deep neural network which takes the welding current, the welding voltage and the welding speed as input and takes the welding seam forming size and the collapse compensation distance at the arc blowout position of the welding seam as output, and training and evaluating the deep neural network by utilizing an experimental data set to divide a training set and a verification set to obtain a collapse compensation distance prediction model at the arc blowout position of the welding seam;

collecting the welding current, the welding voltage and the welding speed obtained in the welding line forming process, inputting the process parameters into the welding line arc quenching part collapse compensation distance prediction model, and outputting the corresponding welding line forming size and the welding line arc quenching part collapse compensation distance, thereby accurately predicting the welding line arc quenching part collapse compensation distance.

The method for compensating collapse at arc-quenching of the arc additive manufacturing single-pass multi-layer weld comprises the following steps of: any one or any combination of the residual height in the middle of the welding seam, the residual height at the lowest part of the welding seam and the compensation residual height difference.

The method for compensating collapse at the arc-quenching position of the electric arc additive manufacturing single-channel multilayer weld joint further comprises the following specific steps of designing a corresponding process experiment according to the welding process: on the basis of corresponding materials and equipment, a process parameter range capable of ensuring good weld forming is selected, a multivariable multi-level process experiment is designed by adopting an orthogonal experiment method, and the process experiment times and the training difficulty in building a neural network are reduced to the maximum extent.

The method for compensating collapse at the arc-quenching position of the single-channel multilayer welding line in the electric arc additive manufacturing process further comprises the step that the deep neural network is a multi-input multi-output network and comprises an input layer, a hidden layer and an output layer, wherein the neuron number of the output layer is determined by process parameters, the output layer comprises the welding line forming size and the collapse compensation distance at the arc-quenching position of the welding line, the number of layers of the hidden layer and the neuron number are determined jointly by the non-linearity degree, the forming prediction accuracy, the network model weight and the calculation efficiency of the threshold value in the multi-input multi-output network formed by the process parameters and the welding line forming result, and the model structure and the network parameters are optimized and adjusted according to prediction errors.

The method for compensating collapse at the arc-quenching position of the single-channel multilayer weld seam manufactured by the electric arc additive manufacturing method further comprises the following steps of: and optimizing the model structure and the network parameters according to the size of the data set, the network structure, the model calculation speed and/or the prediction result precision.

The method for compensating collapse at the arc-quenching position of the arc additive manufacturing single-pass multilayer weld joint further comprises the following specific steps of: the size of a network mechanism, network training time, model calculation efficiency and forming prediction precision are comprehensively considered, so that the calculation cost and efficiency of the welding seam forming prediction model are evaluated, and the accuracy and stability of the model are guaranteed, wherein the model accuracy is determined by linear regression analysis of prediction data and actual data.

The method for compensating collapse at the arc quenching position of the electric arc additive manufacturing single-channel multilayer weld joint is characterized in that the structural form of the weld joint comprises the following steps: the welding method comprises any one or any combination of non-consumable electrode gas shielded welding TIG, consumable electrode gas shielded welding MIG \ MAG, cold metal transition welding CMT or plasma arc welding PAW.

According to the method for compensating collapse at the arc-quenching position of the electric arc additive manufacturing single-pass multi-layer welding line, further, the compensation distance for collapse at the arc-quenching position of the welding line is the compensation distance with the optimal flatness of the welding line after corresponding compensation.

Compared with the prior art, the invention has the beneficial effects that: the collapse compensation method is based on the technical parameter prediction of the neural network construction, and during welding of each layer, collapse compensation is carried out in a mode that arc blowout is not carried out at an arc blowout position, arc blowout is carried out after a certain distance of arc blowout is carried out in a backward welding mode. The method can effectively solve the problem of collapse of the arc quenching position in the single-layer multi-channel forming process, thereby ensuring the high smoothness of the finished part in the electric arc additive manufacturing process and further solving the problem that the new feature of the part is influenced due to collapse of the arc quenching position. In addition, compared with the existing solutions such as parallel reciprocation and the like, the method provided by the invention has the advantages of low cost, convenience in operation, precision in collapse compensation and the like.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a flow chart of the building steps of a model for predicting collapse compensation distance at the arc blowout position of a weld joint in the embodiment of the invention;

FIG. 2 is a logic flow diagram illustrating the operation of a model for predicting the collapse compensation distance at arc quenching of a weld according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of the ER308L stainless steel welding wire MIG weld prediction neural network structure according to an embodiment of the invention;

FIG. 4 is a photograph of an experimental profile of a MIG weld of an ER308L stainless steel welding wire according to an embodiment of the invention;

FIG. 5 is a schematic diagram of prediction error results of the MIG weld forming of the ER308L stainless steel welding wire according to the embodiment of the invention;

FIG. 6 is a comparison of results of an ER308L stainless steel wire MIG weld application of an embodiment of the present invention wherein (a) is a single pass multi-layer weld structure made in a conventional stacked manner; (b) The single-pass multilayer welding seam component manufactured by the method is adopted.

Detailed Description

The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.

The embodiment is as follows:

it should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

The word "exemplary" is used hereinafter to mean "serving as an example, embodiment, or illustration. Any embodiment described as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

Referring to fig. 1 to 5, the invention provides a method for compensating collapse at an arc-quenching position of a single-pass multilayer weld in electric arc additive manufacturing, which mainly comprises the steps of preparing a welding plate and welding equipment, designing and implementing a process experiment, preprocessing process parameters and experimental data, collecting and establishing a weld forming experiment data set, building a neural network, deploying a weld forming result parameter prediction model and the like, and specifically comprises the following steps:

step 101: and designing a corresponding welding process according to the welding material, the structural form and the welding method.

Specifically, a welding plate and welding equipment are prepared, and according to selected arc welding equipment, welding materials, structural forms, welding modes and technological parameter design schemes, suitable experimental materials such as the plate, welding wires and welding shielding gas and the welding equipment such as a welding robot, a welding power supply and a water cooling machine are prepared.

In some embodiments, the structural form of the welding seam may include a thick-wall single-pass multilayer stacking structure, a thin-wall single-pass multilayer stacking structure, and the like, and different materials and different thicknesses of base metal plates may be provided for support according to different types. The welding method can comprise any one or any combination of non-consumable gas shielded welding TIG, consumable gas shielded welding MIG \ MAG, cold metal transition welding CMT and plasma arc welding PAW according to different types of electric arcs.

Step 102: designing corresponding process experiments according to the welding process, and implementing the process experiments in batches to further obtain process experiment data; the process experimental data comprise welding current, welding voltage, welding speed, welding seam forming size and collapse compensation distance of the arc quenching position of the welding seam.

Specifically, the process experiment design and implementation are that a single-channel multilayer weld stacking process experiment is designed according to experimental conditions such as the load performance of a welding power supply, the thickness of a welding plate, the material and the diameter of a welding wire, and batch experiments are carried out by using selected electric arc additive manufacturing equipment.

And (3) preprocessing process parameters, namely preprocessing process experiment data into structured data required by a weld forming shape data set, wherein the process parameter data content related to the process experiment comprises welding current, welding voltage, welding speed, weld forming size and collapse compensation distance at the arc quenching position of the weld. For example, the process parameters may include welding current, welding voltage, welding speed, etc., and the experimental result data may include the weld forming size and the collapse compensation distance at the arc quenching of the weld.

In some embodiments, the weld forming dimensions include weld middle residual height, weld lowest residual height, and compensation residual height difference. And the collapse compensation distance at the arc quenching position of the welding line is the compensation distance with the optimal welding line smoothness after corresponding compensation.

In other embodiments, the specific steps of designing a corresponding process experiment according to the welding process include: on the basis of corresponding materials and equipment, in the process parameter range of ensuring good weld forming, an orthogonal experiment method is adopted to design a multivariable multi-level large-range additive manufacturing process experiment, on the premise of ensuring the experiment effect, the process experiment times and the training difficulty of a neural network are reduced as much as possible, and meanwhile, the precision and the efficiency of a collapse compensation distance prediction neural network model at the arc quenching position of the weld are ensured to the maximum extent on the basis of avoiding the condition that the neural network is not fitted or is over-fitted due to too few data sets.

Step 103: and establishing a collapse compensation distance prediction data set at the arc quenching position of the welding seam according to the data combination mode of the welding current, the welding voltage, the welding speed, the welding seam forming size and the collapse compensation distance at the arc quenching position of the welding seam.

Specifically, a weld forming experiment data set is collected and established, and a collapse compensation distance prediction data set at the arc blowout position of the weld is established according to collected process parameters and experiment result data in a data combination mode of welding current, welding voltage, welding speed, weld forming size and collapse compensation distance at the arc blowout position of the weld.

Step 104: and constructing a deep neural network which takes the welding current, the welding voltage and the welding speed as input and takes the welding seam forming size and the collapse compensation distance at the arc blowout position of the welding seam as output, and training and evaluating the deep neural network by utilizing an experimental data set to divide a training set and a verification set to obtain a collapse compensation distance prediction model at the arc blowout position of the welding seam.

Specifically, the neural network is built, the functions of building a neural network structure, adjusting neural network training parameters, evaluating network training efficiency and model prediction energy efficiency, optimizing the network structure and the training parameters and the like are designed and realized, the network structure and the training parameters are adjusted according to the error condition of a prediction result, and the collapse compensation distance prediction model at the arc quenching position of the welding line is obtained.

In some embodiments, the deep neural network is specifically a multi-input multi-output network, which is composed of an input layer, a hidden layer and an output layer, wherein the neuron number of the output layer is determined by process parameters, the output layer comprises a weld forming size and a collapse compensation distance at an arc quenching position of a weld, the layer number and the neuron number of the hidden layer are determined by the process parameters to the non-linearity degree and the forming prediction accuracy in the multi-input multi-output network composed of weld forming results, and the calculation efficiency of network model weights and threshold values, and the model structure and the network parameters are optimized and adjusted according to prediction errors.

In some embodiments, the step of optimally adjusting the model structure and the network parameters specifically includes: and optimizing the model structure and the network parameters according to the size of the data set, the network structure, the model calculation speed and/or the prediction result precision. The optimization and adjustment of the model structure and the network parameters mainly aim at adjusting according to the size of a data set, prediction errors and calculation efficiency, and comprehensive control of feedback and adjustment of the neural network forming prediction model parameters is achieved.

In other embodiments, evaluating the neural network specifically includes: the method comprises the steps of evaluating the size of a network structure of the neural network, evaluating training time when the neural network is built, evaluating the calculation efficiency of a neural network model and evaluating the precision of a prediction result of the neural network, comprehensively evaluating the calculation cost and efficiency of a compensation distance prediction model at an arc blowout position, ensuring the precision and stability of the neural network prediction model, and determining the precision of the network model by performing linear regression analysis on prediction result data and experimental test data.

Step 105: collecting the welding current, the welding voltage and the welding speed obtained in the welding line forming process, inputting the process parameters into the welding line arc quenching part collapse compensation distance prediction model, and outputting the corresponding welding line forming size and the welding line arc quenching part collapse compensation distance, thereby accurately predicting the welding line arc quenching part collapse compensation distance.

Specifically, deploying a weld forming result parameter prediction model, deploying the weld arc quenching collapse compensation distance prediction model to a computer platform, designing a human-computer interaction interface and a port, wherein an input port is welding equipment, material types, structural forms, welding modes and process parameters, an output port is a weld forming size and a weld arc quenching collapse compensation distance, and an expansion port is reserved.

For a better understanding of the present invention, the logic of the operation of the inventive weld blowout collapse compensation distance prediction model is set forth below.

Referring to fig. 2, first, a plurality of sets of corresponding data of the process parameters and the weld forming data are read from the welding parameters and the forming data set; secondly, presetting a neural network structure and training parameters; thirdly, based on welding process parameters and welding seam forming data, placing a preset neural network structure and training parameters in a welding seam arc-quenching collapse compensation distance prediction neural network model for training; then, evaluating the effectiveness of the neural network, if the evaluation fails, carrying out optimization adjustment on the network structure and the network training parameters, feeding back the optimization adjustment to the preset stages of the network structure and the network training parameters, and carrying out neural network training again until the neural network model passes the effectiveness evaluation; and finally, after the evaluation is passed, deploying a collapse compensation distance prediction model at the arc blowout position of the weld joint of the neural network.

For better understanding of the present invention, the method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to the present invention is further described below by way of specific examples.

The embodiment of the invention provides a neural network-based prediction system for the collapse compensation distance at the arc extinction position of the MIG welding line of an ER308L stainless steel welding wire, wherein the model of welding equipment is TM-1400 GIII, the model of an arc welding robot is Panasonic robot-AUR01062, the model of a demonstrator system is a windows CE system, the model of a welding power supply is YD-500GS, the model of a wire is ER308L, and the size of a test panel is 400mm multiplied by 200mm multiplied by 10mm.

The structure form is flat single-pass multilayer accumulation, the welding method is front wire flat welding, a welding gun is perpendicular to a test plate, the type of an electric arc is consumable electrode inert gas shielded welding (MIG), a substrate is Q345 low-carbon steel, the welding wire is an ER308L stainless steel welding wire, the diameter is 1.2mm, the value of an arc voltage is 22.5V, the welding speed is 0.5-0.8 m/min, the value range of a welding current is 160-220A, the dry elongation of the welding wire is 15mm, the protective gas is (98 Ar < 2 > CO2 >) mixed gas, and the gas flow is 20L/min.

The forming size data of the ER308L stainless steel wire MIG welding line is shown in the table 1, and the corresponding welding line morphology photo data is shown in the figure 4.

TABLE 1ER308L stainless steel welding wire MIG weld forming dimension data

The neural network prediction model is a three-input two-output network, wherein the input is welding voltage, welding current and welding speed, and the output is welding seam forming size and collapse compensation distance at the welding seam arc quenching position. In order to reduce the value influence of the characteristic values and ensure the performance of the neural network, the characteristic data is processed by using a maximum and minimum normalization method, namely for each characteristic value x, after calculating a mean value mean (x), a maximum value max (x) and a minimum value min (x), linear transformation is carried out, namely normalization:

the data are processed into a [0,1] interval through normalization so as to reduce the training difficulty of the neural network model, ensure the iteration speed of the neural network model and prevent the overfitting phenomenon.

Referring to fig. 3, fig. 3 is a neural network structure for predicting the collapse compensation distance of the ER308L stainless steel wire MIG weld, and the neural network structure for predicting the collapse compensation distance at the arc blowout position of the weld is composed of an input layer, a hidden layer i, a hidden layer ii and an output layer, wherein the number of neurons in the input layer is 3, the number of neurons in the hidden layer is 11, and the number of neurons in the output layer is 2. The final training parameter of the neural network, batch _ size, epochs, and training set size are 6, 2000, the training set size is 50, the test set size is 6, the learning rate is 0.1, and in order to measure the prediction accuracy of the neural network more significantly, two parameters, namely mean square error MSE and regression value R, are used as the criteria for measuring the accuracy of the neural network prediction model, and can be expressed as:

the prediction error result of the ER308L stainless steel welding wire MIG weld prediction neural network is shown in fig. 5, and for the final weld blowout collapse compensation distance prediction neural network to be deployed, the neural network belongs to a multi-input multi-output height nonlinear function relationship from welding voltage U, welding current I and welding speed V to the weld residual height difference and weld compensation distance in the weld forming dimension, and can be written as:

in order to evaluate the effectiveness and the accuracy of a neural network prediction model, a large number of data points are intensively generated in reasonable value ranges of three process parameters of welding current I, welding voltage U and welding speed V for prediction, and are compared with experimental data after normalization processing, then prediction is carried out, and the correlation and the error between an actually measured value and a predicted value are analyzed.

And (3) importing the data into a neural network prediction model, and operating the trained model to obtain the mean square error MSE of the model of 0.0052606. The correlation coefficient of each data set is shown in the neural network regression fitting diagram of fig. 5 (a), and it can be seen that the regression value R of all data sets is greater than 0.9 and close to 1, which indicates that the accuracy and effect of the established model prediction are excellent. Fig. 5 (b) is a comparison graph of the neural network prediction result and the actual value, the prediction value 1 and the actual value 1 refer to the prediction value and the actual value of the weld residual height difference, the maximum error of the prediction value and the actual value < =0.2mm, the prediction value 2 and the actual value 2 refer to the prediction value and the actual value of the collapse compensation distance at the arc blowout position of the weld, and the maximum error of the prediction value and the actual value < =0.2mm. It can be seen that; the coincidence degree between the actual value and the predicted value of the two output parameters is very good, so that the error evaluation of the neural network prediction model is qualified and can be used.

Referring to fig. 6, fig. 6 is a comparative graph illustrating the MIG weld application results of ER308L stainless steel wire in accordance with an embodiment of the present invention. Predicting the compensation distance and the welding process parameters by a trained BP neural network prediction model, wherein the program instruction is A = [15;150;22.5;0.5] and y = sim (net 0, a), net0 is a trained neural network, the obtained results are 0.1899 and 0.3088, the predicted value is relatively small, therefore, the distance of the backward welding is set to 15mm according to the predicted result, the current during the backward welding is adjusted to 150A, the welding voltage and the welding speed are not changed, the parameters are consistent with the previous parameters, the forming test of the single-channel multilayer straight wall body is carried out, the welding current is reduced to 150A while the arc is not extinguished at the arc blowout part and the backward welding is carried out for 15mm during the forming of the same layer, and then the arc blowout is carried out, and the interlayer temperature is controlled to be about 150 degrees. Fig. 6 is a graph showing a comparison between the experimental result of the compensation method of the present invention and the conventional stacking experimental result, in which fig. 6 (a) is a forming result of the conventional stacking forming method and fig. 6 (b) is a forming result of the collapse compensation method. It can be seen from fig. 6 that the surface of the formed part obtained by the collapse compensation method is relatively flat, and the two ends of the formed part are not obviously collapsed, so that the forming result is greatly improved compared with the forming result obtained by the traditional method.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Moreover, various embodiments or examples and features of various embodiments or examples described in this specification can be combined and combined by one skilled in the art without being mutually inconsistent.

The above embodiments are only for illustrating the technical idea and features of the present invention, and the purpose of the present invention is to enable those skilled in the art to understand the content of the present invention and implement the present invention, and not to limit the protection scope of the present invention by this. All equivalent changes or modifications made in accordance with the spirit of the present disclosure are intended to be covered by the scope of the present disclosure.

Claims (8)

1. The method for compensating collapse at the arc blowout position of the single-channel multilayer welding seam in electric arc additive manufacturing is characterized by comprising the following steps of: the method comprises the following steps:

designing a corresponding welding process according to welding equipment, welding materials and a welding method;

designing corresponding process experiments according to the welding process, and implementing the process experiments in batches to further obtain process experiment data; the process experimental data comprise welding current, welding voltage, welding speed, welding seam forming size and collapse compensation distance at the arc quenching position of the welding seam;

establishing a collapse compensation distance prediction data set at the arc quenching position of the welding seam according to the data combination mode of the welding current, the welding voltage, the welding speed, the welding seam forming size and the collapse compensation distance at the arc quenching position of the welding seam;

constructing a deep neural network which takes the welding current, the welding voltage and the welding speed as input and takes the welding seam forming size and the collapse compensation distance at the arc blowout position of the welding seam as output, and training and evaluating the deep neural network by utilizing an experimental data set to divide a training set and a verification set to obtain a collapse compensation distance prediction model at the arc blowout position of the welding seam;

collecting the welding current, the welding voltage and the welding speed obtained in the welding line forming process, inputting the process parameters into the welding line arc quenching part collapse compensation distance prediction model, and outputting the corresponding welding line forming size and the welding line arc quenching part collapse compensation distance, thereby accurately predicting the welding line arc quenching part collapse compensation distance.

2. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, characterized in that: the weld forming dimensions mainly include: any one or any combination of weld widening, excess height of the middle position of the weld and collapse excess height of the arc quenching position of the weld.

3. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, characterized in that: the step of designing a corresponding process experiment according to the welding process specifically comprises: on the basis of corresponding materials and equipment, a process parameter range capable of ensuring good weld joint forming is selected, a multivariable multilevel process experiment is designed by adopting an orthogonal experiment method, and the process experiment times and the training difficulty in building a neural network are reduced to the maximum extent.

4. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, characterized in that: the deep neural network is specifically a multi-input multi-output network and comprises an input layer, a hidden layer and an output layer, the neuron number of the output layer is determined by process parameters, the output layer comprises a welding seam forming size and a welding seam arc quenching collapse compensation distance, the layer number and the neuron number of the hidden layer are determined by the process parameters to the calculation efficiency of a network model weight and a threshold value in the multi-input multi-output network formed by welding seam forming results, the non-linearity degree, the forming prediction accuracy and the network model weight are determined together, and the model structure and the network parameters are optimized and adjusted according to prediction errors.

5. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 4, characterized in that: the step of optimizing and adjusting the model structure and the network parameters specifically comprises: and optimizing the model structure and the network parameters according to the size of the data set, the network structure, the model calculation speed and/or the prediction result precision.

6. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, characterized in that: the step of evaluating the neural network specifically includes: the size of a network mechanism, network training time, model calculation efficiency and forming prediction precision are comprehensively considered, so that the calculation cost and efficiency of the welding seam forming prediction model are evaluated, and the accuracy and stability of the model are guaranteed, wherein the model accuracy is determined by linear regression analysis of prediction data and actual data.

7. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, characterized in that: the structural form of the welding seam comprises: the welding method comprises any one or any combination of non-consumable electrode gas shielded welding TIG, consumable electrode gas shielded welding MIG \ MAG, cold metal transition welding CMT or plasma arc welding PAW.

8. The method for compensating collapse at arc-quenching of an arc additive manufacturing single-pass multi-layer weld according to claim 1, wherein: and the collapse compensation distance at the arc quenching position of the welding line is the compensation distance with the optimal welding line smoothness after corresponding compensation.

Images