An online monitoring device and method for internal defects of arc additively formed metal components

Abstract

The present invention discloses an online monitoring device and method for internal defects of arc additively formed metal components, including an arc additively formed system, a signal acquisition system and a signal processing system. The arc additively formed system is used to realize 3D printing of metal components. The signal acquisition system is connected to the arc additively formed system to obtain acoustic emission signals during the printing process of the metal components. The signal processing system is connected to the signal acquisition system to extract characteristic parameters and establish a machine learning model. The present invention constructs metal components with different defect types or levels by adjusting the printing parameters of the arc additively formed system. The signal acquisition system collects the corresponding acoustic emission signals during the printing process, uses the signal processing system to extract characteristic parameters and constructs a defect discrimination model through a machine learning model; the defect discrimination model discriminates and classifies unknown signals during the printing process, and realizes online monitoring of internal defects of arc additively formed metal components.

Description

Technical Field

[0001] The present invention relates to the technical field of additive manufacturing, and in particular to an online monitoring device and method for internal defects of arc additively formed metal components. Background Art

[0002] Arc additive manufacturing is a type of additive manufacturing technology with the advantages of high forming efficiency and low material cost, and is very suitable for the production of large-sized parts. However, due to its large arc and large molten pool characteristics, it leads to high heat output, uneven temperature, unstable molten pool, and is prone to various external and internal defects, such as warping, faults, holes, cracks, etc. Carrying out online monitoring and detection of forming defects in the arc additive process is of great significance to ensure the forming quality.

[0003] At present, it is relatively easy to monitor external defects such as warping and faults generated by the arc additive forming process online, while internal defects such as holes and cracks are difficult to detect online. Acoustic emission refers to the phenomenon that when a component is deformed or broken by external or internal forces, stress concentration occurs in a local area of ​​the component material, energy is released quickly, and transient elastic waves are generated. Acoustic emission technology can monitor crack initiation and expansion, hole formation, etc. online, making it possible to monitor internal and dynamic defects in the arc additive forming process online. At present, acoustic emission technology has been used in a small number of online monitoring of arc additive manufacturing, such as the invention patent No. 201710095553.3 "A method for detecting and evaluating the stability of arc three-dimensional rapid prototyping manufacturing process based on acoustic emission signals" and the invention patent No. 201710095559.0 "A method for monitoring arc length based on arc acoustic emission signal sensing". However, they use acoustic emission technology to monitor the stability of arc additive manufacturing process and arc length, and have not yet involved the use of acoustic emission technology to monitor internal defects such as pores and cracks in arc additive manufacturing workpieces. The reason is that there are the following problems: (1) During the printing process, the temperature of the substrate near the forming area can be as high as 200°C or more, and it is difficult to directly attach a suitable acoustic emission probe to the substrate; (2) The acoustic emission signal is complex, and most studies only analyze conventional characteristic parameters and use the change of a single parameter to characterize the defects of the printing process, which will affect the judgment results and cannot achieve automatic online monitoring. Summary of the invention

[0004] The purpose of the present invention is to establish a set of internal defect monitoring devices for arc additive forming process based on acoustic emission technology, and propose a machine learning model that couples multiple acoustic emission characteristic parameters to solve the current problem of difficulty in online monitoring of internal defects in arc additive forming process, and realize high-precision online monitoring of internal defects of arc additive forming components.

[0005] The technical solution adopted by the present invention is:

[0006] An online monitoring device for internal defects of an arc additively formed metal component, comprising an arc additively formed system, a signal acquisition system and a signal processing system; the arc additively formed system is used to realize 3D printing of metal components; the arc additively formed system is connected to the signal acquisition system; the signal acquisition system is used to obtain acoustic emission signals during the printing process; the signal processing system is used to extract characteristic parameters, establish a machine learning classification model and distinguish unknown signals;

[0007] The arc additive forming system includes a robot arm, a control host, a wire loop and its electric drive device, a welding wire extrusion device, a water-cooled argon arc welding gun, a range hood, an air purifier, a water tank and an argon protection device; the water-cooled argon arc welding gun is mounted at the tip of the robot arm; the wire loop and its electric drive device transport the welding wire to the water-cooled argon arc welding gun along the robot arm through the welding wire extrusion device; the water tank is distributed throughout the arc additive system to reduce the temperature of each component during the printing process; the range hood and the air purifier are located next to the forming workbench to absorb harmful gases; the robot arm, the water-cooled argon arc welding gun, the wire loop and its electric drive device, the range hood and the air purifier are uniformly deployed by the control host; the argon protection device leads to the forming workbench;

[0008] The signal acquisition system comprises a waveguide rod, an acoustic emission probe, a preamplifier and a signal acquisition card; one side of the waveguide rod is fixed on a substrate by a clamp, the acoustic emission probe is coupled to the other side of the waveguide rod by vaseline, the output end of the acoustic emission probe is connected to the input end of the preamplifier by a coaxial cable, the output end of the preamplifier is connected to the input end of the signal acquisition card by a coaxial cable, and the output end of the signal acquisition card is connected to the signal processing system;

[0009] The signal processing system includes feature parameter extraction and machine learning models; the feature parameter extraction module is used to extract time domain feature parameters, frequency domain feature parameters, wavelet feature parameters and quantitative recursive feature parameters from the acoustic emission waveform signal; the machine learning module is used to classify the collected acoustic emission signals and establish a classification discrimination model.

[0010] Furthermore, the waveguide rod includes an upper surface of the waveguide rod, a metal rod of the waveguide rod and a lower surface of the waveguide rod. The upper surface of the waveguide rod is connected to the acoustic emission probe using a vaseline patch, and the lower surface of the waveguide rod is coupled to the substrate through a clamp. The metal rod of the waveguide rod is used to transmit the acoustic emission signal on the substrate to the acoustic emission probe.

[0011] Furthermore, the molding workbench is provided with a base hole, and four bolts pass through the base holes through four clamp gaskets and cooperate with nuts to fix the substrate between the clamp gasket and the base; the lower surface of the waveguide rod is fixed between the clamp gasket and the substrate, and the triangular pad is arranged next to the clamp gasket.

[0012] Furthermore, the waveguide rod has a diameter of 10-20 mm.

[0013] Furthermore, the forming workbench is a rotatable base.

[0014] The present invention also provides an online monitoring method for internal defects of an arc additively formed metal component, comprising the following steps:

[0015] (1) Connect the components of the signal acquisition system in sequence; perform a 45° lead breaking test in the substrate forming area to ensure that the acoustic emission probe and the acoustic emission signal source are well coupled;

[0016] (2) Turn on each component of the arc additive manufacturing system in turn, including the welding system, wire feeding system, robotic arm movement, range hood, and air purifier. After they are stable, test the environmental noise characteristics generated by each component, including amplitude and frequency range;

[0017] (3) Open the signal acquisition software and set the acquisition parameters: amplitude threshold value;

[0018] (4) Setting different process parameters to form metal components with different defect types or levels, and collecting the acoustic emission signals corresponding to the forming process;

[0019] (5) Transmitting the acoustic emission signal to the signal processing system to obtain the corresponding characteristic parameters, including time domain characteristic parameters, frequency domain characteristic parameters, wavelet characteristic parameters and quantitative recursive characteristic parameters;

[0020] (6) Use linear discriminant analysis to classify feature parameters and establish corresponding machine learning models;

[0021] (7) Substitute the collected unknown signals into the signal processing system for discriminant analysis to achieve online monitoring of the defect type or level of the forming process;

[0022] (8) Integrate the predicted and verified unknown signals into the classification discriminant model, and continuously expand and update the machine learning discriminant model.

[0023] Furthermore, the amplitude of the signal received by the acoustic emission probe is not less than 90 dB.

[0024] Furthermore, the acoustic emission probe has a resonance frequency of 150 KHz, an operating frequency range of 60 KHz-1000 KHz, and a sensitivity peak value of >115 dB.

[0025] Furthermore, the time domain characteristic parameters include rise time Rt, ring count C, duration T, amplitude A, relative energy Re and absolute energy Ae.

[0026] Furthermore, the frequency domain characteristic parameters include amplitude frequency Af, root mean square frequency Rf, peak frequency Pf, and center frequency Cf.

[0027] Further, the wavelet characteristic parameters include d1, d2, d3, d4, d5, d6, d7 and d8.

[0028] Furthermore, the recursive characteristic parameters include recursive rate R, determination rate D, Shannon entropy E and average diagonal length L mean .

[0029] The beneficial effects produced by the present invention are:

[0030] 1. The present invention overcomes the problem that the surface temperature near the substrate of arc additive forming is high and the acoustic emission probe cannot be directly placed on its surface by designing an acoustic emission signal waveguide device, thereby achieving high-quality collection of acoustic emission signals during the arc additive forming process.

[0031] 2. The present invention establishes a machine learning model by coupling multiple acoustic emission characteristic parameters, thereby overcoming the problem that traditional single characteristic parameters are difficult to effectively classify acoustic emission signals, and achieving high-precision classification of complex acoustic emission signals.

[0032] 3. The present invention establishes a machine learning model based on the linear discriminant method of supervised learning, which can realize automatic online monitoring and detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a structural schematic diagram of the present invention;

[0034] Figure 2a It is a three-dimensional picture of the acoustic emission probe;

[0035] Figure 2b This is the internal structure diagram of the acoustic emission probe;

[0036] Figure 3 It is a three-dimensional diagram of the waveguide rod;

[0037] Figure 4a For base-substrate-waveguide rod matching Figure 1 ;

[0038] Figure 4b The base-substrate-waveguide rod matching diagram 2;

[0039] Figure 5a This is the actual image printed under the first type of arc additive parameters;

[0040] Figure 5b This is the actual image printed under the second arc additive parameters;

[0041] Figure 5c This is the actual image printed under the third type of arc additive parameters;

[0042] Figure 5d This is the actual image printed under the fourth arc additive parameters;

[0043] Figure 6a This is the higher level hole image Level 1 under the light microscope;

[0044] Figure 6b This is the lower level hole image Level 2 under the light microscope;

[0045] Figure 6c Level 3 image with basically no holes under light microscope;

[0046] Figure 6d It is the hole image of the target component under the optical microscope;

[0047] Figure 6e The discrimination results of the acoustic emission signals corresponding to the target components in the machine learning model;

[0048] In the figure: 1-control host; 2-teaching box; 3-water tank; 4-argon protection device; 5-metal wire ring and its electric drive device; 6-welding wire extrusion device; 7-mechanical arm; 8-water-cooled argon arc welding gun; 9-forming component; 10-forming workbench; 11-substrate; 12-waveguide rod; 12-1-upper surface of waveguide rod; 12-2-waveguide metal rod; 12-3-lower surface of waveguide rod; 13-signal processing system; 14-signal acquisition card; 15-preamplifier; 16-acoustic emission probe; 16-1-terminal; 16-2-housing; 16-3-piezoelectric ceramic; 16-4-coupling surface; 17-range hood; 18-air purifier; J-1-clamp gasket; J-2-bolt; J-3-triangular gasket; J-4-nut. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the accompanying drawings.

[0050] like Figure 1 As shown, an online monitoring device for internal defects of arc additively formed metal components is used. An arc additive system is used to form metal components with different defect types or levels. A signal acquisition system is used to obtain corresponding acoustic emission signals. A signal processing system extracts characteristic parameters and a defect discrimination model is established using a machine learning method. The established defect discrimination model is used to discriminate and classify unknown signals in the printing process, thereby realizing online monitoring of internal defects of arc additively formed metal components.

[0051] The invention comprises an arc additive forming system, a signal acquisition system and a signal processing system.

[0052] The arc additive forming system includes a robot and its control system, a wire feeding system, an electric welding system, an atmosphere protection system, and a forming workbench. The hardware of the robot and its control system mainly consists of a robot arm 7 and a control host 1; the wire feeding system consists of a metal wire ring and its electric drive device 5, and a welding wire extrusion device 6; the electric welding system consists of a water-cooled argon arc welding gun 8 and its housing; and the atmosphere protection system consists of a range hood 17, an air purifier 18, a water tank 3, and an argon protection device 4. The tip of the mechanical arm 7 is equipped with a water-cooled argon arc welding gun 8. The metal wire ring and its electric drive device 5 transport the metal welding wire squeezed by the welding wire squeezing device 6 to the water-cooled argon arc welding gun 8 along the mechanical arm. The water tank 4 is spread throughout the arc additive system through water pipes to reduce the temperature of each component during the printing process. The range hood 17 and the air purifier 18 are located beside the forming workbench 10 to absorb harmful gases. The mechanical arm 7, the water-cooled argon arc welding gun 8, the metal wire ring and its electric drive device 5, the range hood 17, and the air purifier 18 are all connected to the control host 1, are uniformly deployed by the control host 1, and are directly controlled by the teaching box 2. The forming workbench 10 is a rotatable base.

[0053] The signal acquisition system includes a waveguide rod 12, an acoustic emission probe 16, a preamplifier 15 and a signal acquisition card 14; one side of the waveguide rod 12 is fixed to the substrate 11 by a clamp, the acoustic emission probe 16 is coupled to the other side of the waveguide rod 12 by vaseline, the output end of the acoustic emission probe 16 is connected to the input end of the preamplifier 15 through a coaxial cable, the output end of the preamplifier 15 is connected to the input end of the signal acquisition card 14 through a coaxial cable, and the output end of the signal acquisition card 14 is connected to the signal processing system 13.

[0054] The signal processing system 13 includes a feature parameter extraction and a machine learning model. The feature parameter extraction module is used to extract time domain feature parameters, frequency domain feature parameters, wavelet feature parameters and quantitative recursive feature parameters from the acoustic emission waveform signal. The machine learning module is used to classify the collected acoustic emission signals and establish a classification discrimination model.

[0055] like Figure 2a and Figure 2b As shown, the internal structure of the acoustic emission probe 16 includes a terminal 16-1, a housing 16-2, a piezoelectric ceramic 16-3, and a coupling surface 16-4. Its working principle is that the coupling surface 16-4 contacts the sensing interface. When the acoustic emission phenomenon occurs, a stress wave is emitted. The propagation of the stress wave is essentially the movement of particles. When the movement of the particles is transmitted to the contact surface of the probe, the protons on the piezoelectric ceramic 16-3 are driven to move, thereby producing a compression and stretching effect on the piezoelectric ceramic 16-3, which is then converted into a voltage signal and sent to the signal processor to complete the transformation process of the stress wave to the electrical signal wave.

[0056] like Figure 3As shown, the waveguide rod is composed of three parts, namely, the upper surface 12-1 of the waveguide rod, the metal rod 12-2 of the waveguide rod, and the lower surface 12-3 of the waveguide rod. The upper surface 12-1 of the waveguide rod is connected to the acoustic emission probe by using vaseline, the lower surface 12-3 of the waveguide rod is coupled to the substrate by a clamp, and the metal rod 12-2 of the waveguide rod is used to transmit the acoustic emission signal on the substrate to the acoustic emission probe.

[0057] As shown in Fig. 4, the base-substrate-waveguide rod matching fixing structure includes a clamp gasket J-1, a bolt J-2, a triangular pad J-3, a nut J-4, a base 11, a base hole 10-1, an upper surface 12-1 of the waveguide rod, a lower surface 12-3 of the waveguide rod, and an acoustic emission probe 16. The substrate-waveguide rod-probe matching method is that the bolt J-2 passes through the base hole 10-1 through a clamp gasket J-1 and cooperates with the nut J-4 to fix the substrate 11 and the lower surface 12-3 of the waveguide rod between the clamp gasket J-1 and the base. The triangular pad J-3 can adapt to substrates 11 and waveguide rods of different thicknesses, so that the clamp gasket J-1 remains horizontal when fixed. At least four sets of clamps are required for each fixation to ensure that the substrate 11 and the waveguide rod are completely fixed. The upper surface 12-1 of the waveguide rod is coupled with the acoustic emission probe 16 through a vaseline patch to collect the acoustic emission signal generated on the substrate 11 during the printing process.

[0058] An online monitoring method for internal defects of an arc additively formed metal component comprises the following steps:

[0059] Connect the hardware of each system as follows and determine the parameters: the tip of the robot arm 7 is equipped with a water-cooled argon arc welding gun 8, the metal wire ring and its electric drive device 5 transport the metal welding wire squeezed by the welding wire squeezing device 6 to the water-cooled argon arc welding gun 8 along the robot arm, the water tank 4 is spread throughout the arc additive system through water pipes to reduce the temperature of each component during the printing process, the range hood 17 and the air purifier 18 are located next to the forming workbench to absorb harmful gases, the robot arm 7, the water-cooled argon arc welding gun 8, the metal wire ring and its electric drive device 5, the range hood 17, and the air purifier 18 are all connected The water enters the control host 1 and is uniformly allocated by the controlled host 1; wherein, the water tank 3 is spread throughout the various parts of the arc additive system through the water pipe, and the working flow rate is 1.7GPM, which is used to reduce the temperature of various components in the printing process; the argon protection device 4 is used in that it neither reacts chemically with the metal nor dissolves in the metal, so that the metallurgical reaction of the molten pool in the printing process appears simple and easy to control, thus providing good conditions for forming high-quality components; the range hood 17 is used to remove various fumes and exhaust gases generated during the printing process; the air purifier 18 processes the smoke and exhaust gases drawn in. The bolt J-2 passes through the base hole 10-1 through a clamp gasket J-1 and cooperates with the nut J-4 to fix the substrate 11 and the lower surface 12-3 of the waveguide rod between the clamp gasket J-1 and the base. The triangular pad J-3 can adapt to substrates 11 and waveguide rods of different thicknesses so that the clamp gasket J-1 remains horizontal when fixed. At least four sets of clamps are required each time to ensure that the substrate 11 and the waveguide rod are completely fixed. The upper surface 12-1 of the waveguide rod is coupled with the acoustic emission probe 16 through Vaseline to collect the acoustic emission signals occurring on the substrate 11 during the printing process. Among them, the upper surface 12-1 of the waveguide rod is a square surface with a side length of 50mm and a thickness of 1mm, the lower surface 12-3 of the waveguide rod is a square surface with a side length of 80mm and a thickness of 1mm, and the waveguide rod metal rod 12-2 is a cylindrical A3 steel metal tube with a length of 200mm and a diameter of Φ15; the thickness of the forming workbench 10 is 5mm, the base hole diameter is 2cm, the bolt J-2 uses a full-thread bolt with a length of 100mm and a hexagonal head diameter of 20mm, the nut J-4 uses a matching nut, the diameter of the slot hole of the fixture gasket J-1 is slightly less than 20mm, the triangular pad is 60mm long, wide, and 60mm thick, and is a right-angled triangular prism. By adjusting the position with the fixture gasket J-1, substrates and waveguide rods of various thicknesses can be configured. The acoustic emission probe 16 uses a split piezoelectric probe with a resonant frequency of 150KHz, an operating frequency range of 60KHz-1000KHz, and a sensitivity peak value of >115dB. The preamplifier 15 uses a 2 / 4 / 6 type amplifier, which can provide amplification gains of 20dB, 40dB and 60dB, and the gain of the preamplifier 15 is set at 60dB. The input end of the signal acquisition card 14 is connected to the input end of the preamplifier 15.The signal acquisition card 14 has a 10M / s sampling rate, 16-bit sampling accuracy, low system noise and high dynamic range. Each signal acquisition card comes with a 1Gb waveform buffer.

[0060] After the hardware connection is completed, the lead is broken continuously for 10 times in the forming area on the substrate 11. The extended length of the lead core is about 2.5 mm, and the angle with the surface of the iron plate is 45°. The average signal amplitude received by the acoustic emission probe is not less than 90 dB. It is determined that the acoustic emission probe 16 is well coupled with the upper surface 12-1 of the waveguide rod, the lower surface 12-3 of the waveguide rod and the substrate 11, and the acoustic emission signal on the substrate 11 can be collected.

[0061] The robot is a FANUC robot, and its forming workbench 10 is placed on a horizontal ground. The working mode of the FANUC robot in arc additive is:

[0062] 1. Model and data processing: First, use 3D software to perform 3D modeling on the components to be printed, and export them in STL file format; import the STL file into the slicing software to perform layered slicing, area division, path planning, and calibration of the printing coordinate system in sequence, and output it in LS file format; import the LS file format into the simulation software to perform path simulation of the printing process, and export it in LP file format; import the LP file format into the printing arc additive main box for printing operations.

[0063] 2. Printing operation: The robot arm 7 and the forming workbench 10 are calibrated according to the coordinates in the model, and then the actual operation simulation is carried out according to the simulation path in the simulation software. After confirming that the printing path is correct, open the water tank 3, the argon protection device 4, the metal wire ring and its electric drive device 5, the welding wire extrusion device 6, the range hood 17, and the air purifier 18. After ensuring that each system is working stably, start printing; after clicking on the teaching box 2 to start printing, the water-cooled argon arc welding gun 8 starts working, and the 5356 aluminum welding wire is melted between the substrate 10 to form an arc. After scanning a layer according to the predetermined path, the water-cooled argon arc welding gun 8 stops working, and the robot arm 7 is lifted for a distance. After the molten pool cools down for a certain period of time, the robot arm 7 is lowered back to the arc starting point, and the water-cooled argon arc welding gun 8 re-starts the arc and scans the next layer according to the predetermined path. Repeat this process until the component is printed.

[0064] To ensure that the arc additive system can work normally, the signals generated by the operation of each system are first collected, including the signal generated when the water tank 3 is working stably, the signal generated when the argon protection device 4 is working stably, the signal generated when the wire ring and its electric drive device 5 are feeding, the signal generated when the welding wire extrusion device 6 is extruding, the signal generated when the robot arm 7 moves along the predetermined path, the signal generated when the water-cooled argon arc welding gun 8 is turned on and off, the signal generated when the range hood 17 is working stably, and the signal generated when the air purifier 18 is working stably. After the acoustic emission probe 16 collects the noise generated by various internal operations, the acquisition software is opened and the acquisition parameter amplitude threshold is set to 40 dB to remove environmental noise.

[0065] Adjust the printing parameters, print 5356 aluminum alloy components with different porosities, and collect the corresponding acoustic emission signals. Arc additive parameters 1: wire feeding speed 4m / min, scanning speed 40mm / s, layer thickness 5mm, argon gas flow 15L / min, interlayer cooling time 8s; see the actual printed picture Figure 5a ; After cutting, grinding, polishing and etching the components, the metallographic diagram is shown in Figure 6a There are many pits on the surface, and the hole level is Level 1, which means there are a lot of holes. Arc additive parameters 2: wire feeding speed 4m / min, scanning speed 30mm / s, layer thickness 5mm, argon gas flow 15L / min, interlayer cooling time 8s; see the actual print picture Figure 5b ; After cutting, grinding, polishing and etching the components, the metallographic diagram is shown in Figure 6b There are a few pits on the surface, and the hole level is Level 2, which means there are fewer holes. Arc additive parameters 3: wire feeding speed 4m / min, scanning speed 20mm / s, layer thickness 5mm, argon gas flow 15L / min, interlayer cooling time 8s; see the actual print picture Figure 5c After cutting, grinding, polishing and etching the components, the metallographic diagram is shown in Figure 6c , it is very dense under the light microscope, and the pore level is Level 3, which means there are no pores.

[0066] The signal processing system 13 processes the collected signal to obtain time domain characteristic parameters, frequency domain characteristic parameters, wavelet characteristic parameters and recursive characteristic parameters. The time domain characteristic parameters include rise time Rt, ring count C, duration T, amplitude A, relative energy Re, and absolute energy Ae; the frequency domain characteristic parameters include amplitude frequency Af, root mean square frequency Rf, peak frequency Pf, and center frequency Cf; the wavelet characteristic parameters are the 8-layer wavelet decomposition of the signal using the db3 function (Bechy limit phase wavelet), where d1, d2, and d3 represent short-time scale events, and d4, d5, d6, d7, and d8 represent medium and long-time events; the quantitative recursive characteristic parameters include recursive rate R, determination rate D, Shannon entropy E, and average diagonal length L mean.

[0067] The signal processing system 13 uses the linear discriminant analysis method to perform machine learning classification modeling on the collected acoustic emission signals, and its modeling data finally selects 22 characteristic parameters: rise time Rt, ring count C, duration T, amplitude A, relative energy Re, absolute energy Ae, amplitude frequency Af, root mean square frequency Rf, peak frequency Pf, center frequency Cf, wavelet characteristic parameters d1, d2, d3, d4, d5, d6, d7, d8, recursion rate R, determination rate D, Shannon entropy E, average diagonal length L mean A machine learning discriminant classification model was established with a classification accuracy of 97.8%. Figure 6e As shown. Among them, the specific discriminant function is as follows:

[0068] LDA1=5.112Rt+0.144C+0.23T-0.587A+0.386Re+0.87Ae+0.707Af-5.23Rf+2.31Pf-0.32Cf+8.749 d1-0.381d2+1.269d3-1.331d4+2.646d5+1.144d6-0.676d7-0.0011d8+5.433R-18.745D-6.06×10 -3 L mean .

[0069] LDA2=2.112Rt+0.244C-0.43T+0.287A-0.86Re+0.37Ae-0.407Af-2.23Rf+4.31Pf-1.32Cf+5.74 9d1+2.381d2-0.269d3-3.331d4+0.646d5+3.144d6+1.676d7-0.0011d8-2.433R+8.75D+2.06×10 -3 L mean .

[0070] Input the arc additive parameters of the target component: wire feeding speed 5m / min, scanning speed 25mm / s, layer thickness 5mm, argon gas pressure 15L / min, interlayer cooling time 8s, and print the target workpiece. The signal acquisition card 14 collects the corresponding acoustic emission signal, and the signal processing system 13 processes the signal to obtain time domain feature parameters, frequency domain feature parameters, wavelet feature parameters and recursive feature parameters, and input them into the machine learning model for discriminant analysis. The discriminant results are as follows: Figure 6e shown.

[0071] The actual picture of the formed target component is as follows Figure 5d After cutting, grinding, polishing and etching the components, the metallographic diagram is as follows: Figure 6dThe results show that the identification results of internal defects in the arc additive forming process using acoustic emission technology are consistent with the metallographic photograph results.

[0072] The contents described in the implementation of the present invention are merely an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms described in the embodiments. The protection scope of the present invention and the equivalent technical means that can be thought of by those skilled in the art based on the inventive concept.

Claims

1. An online monitoring method for internal defects of an arc additively formed metal component, the device used includes an arc additive forming system, a signal acquisition system and a signal processing system, the arc additive forming system is used to realize 3D printing of metal components; the arc additive forming system is connected to the signal acquisition system; the signal acquisition system is used to obtain acoustic emission signals during the printing process; the signal processing system is used to extract characteristic parameters, establish a machine learning classification model and distinguish unknown signals; The arc additive forming system includes a robot arm, a control host, a metal wire ring and its electric drive device, a welding wire extrusion device, a water-cooled argon arc welding gun, a range hood, an air purifier, a water tank and an argon gas protection device; The water-cooled argon arc welding gun is mounted on the tip of the robotic arm; The wire ring and its electric drive device transport the welding wire to the water-cooled argon arc welding gun along the mechanical arm through the welding wire extrusion device; the water tank is distributed throughout the arc additive system to reduce the temperature of each component during the printing process; the range hood and the air purifier are located next to the forming workbench to absorb harmful gases; the mechanical arm, the water-cooled argon arc welding gun, the wire ring and its electric drive device, the range hood and the air purifier are uniformly deployed by the controlled host; the argon protection device leads to the forming workbench; The signal acquisition system comprises a waveguide rod, an acoustic emission probe, a preamplifier and a signal acquisition card; one side of the waveguide rod is fixed on a substrate by a clamp, the acoustic emission probe is coupled to the other side of the waveguide rod by vaseline, the output end of the acoustic emission probe is connected to the input end of the preamplifier by a coaxial cable, the output end of the preamplifier is connected to the input end of the signal acquisition card by a coaxial cable, and the output end of the signal acquisition card is connected to the signal processing system; The signal processing system includes a feature parameter extraction and a machine learning model; the feature parameter extraction model is used to extract time domain feature parameters, frequency domain feature parameters, wavelet feature parameters and quantitative recursive feature parameters from the acoustic emission waveform signal; the machine learning model is used to classify the collected acoustic emission signals and establish a classification discrimination model; The method for monitoring using the above device comprises the following steps: (1) Connect the components of the signal acquisition system in sequence; perform a 45° lead breaking test in the substrate forming area to ensure that the acoustic emission probe and the acoustic emission signal source are well coupled; The acoustic emission probe is a broadband differential piezoelectric sensor with a resonant frequency of 150 KHz, an operating frequency range of 60 KHz-1000 KHz, a sensitivity peak value of >115 dB, and a signal amplitude received by the acoustic emission probe in a lead breaking experiment of not less than 90 dB; (2) Turn on each component of the arc additive manufacturing system in turn, including the welding system, wire feeding system, robotic arm movement, range hood, and air purifier. After they are stable, test the environmental noise characteristics generated by each component, including amplitude and frequency range; (3) Open the signal acquisition software and set the acquisition parameters: amplitude threshold value; (4) Setting different process parameters to form metal components with different defect types or levels, and collecting the acoustic emission signals corresponding to the forming process; (5) Transmitting the acoustic emission signal to the signal processing system to obtain the corresponding characteristic parameters, including time domain characteristic parameters, frequency domain characteristic parameters, wavelet characteristic parameters and quantitative recursive characteristic parameters; The time domain characteristic parameters include rise time Rt, ring count C, duration T, amplitude A, relative energy Re and absolute energy Ae; The frequency domain characteristic parameters include amplitude frequency Af, root mean square frequency Rf, peak frequency Pf and center frequency Cf; The wavelet characteristic parameters are obtained by performing 8-layer wavelet decomposition on the collected signal using the db3 function to obtain characteristic parameters d1, d2, d3, d4, d5, d6, d7 and d8; The quantitative recursive characteristic parameters include recursive rate R, determination rate D, Shannon entropy E and average diagonal length L mean ; (6) Use linear discriminant analysis to classify feature parameters and establish corresponding machine learning models; (7) Substitute the collected unknown signals into the signal processing system for discriminant analysis to achieve online monitoring of the defect type or level of the forming process; (8) Integrate the predicted and verified unknown signals into the classification discriminant model, and continuously expand and update the machine learning discriminant model.

2. The online monitoring device for internal defects of arc additively formed metal components according to claim 1, characterized in that: The waveguide rod comprises an upper surface of the waveguide rod, a metal rod of the waveguide rod and a lower surface of the waveguide rod. The upper surface of the waveguide rod is connected to the acoustic emission probe by using vaseline, the lower surface of the waveguide rod is coupled to the substrate by a clamp, and the metal rod of the waveguide rod is used to transmit the acoustic emission signal on the substrate to the acoustic emission probe.

3. The online monitoring device for internal defects of arc additively formed metal components according to claim 1, characterized in that: The forming workbench is provided with a base hole, and four bolts pass through the base holes through four clamp gaskets and cooperate with nuts to fix the substrate between the clamp gasket and the base; the lower surface of the waveguide rod is fixed between the clamp gasket and the substrate, and the triangular pad is arranged next to the clamp gasket.

4. The online monitoring device for internal defects of arc additively formed metal components according to claim 1, characterized in that: The waveguide rod has a diameter of 10-20 mm, and the forming workbench is a rotatable base.

Images