A multi-target real-time monitoring device and system for additive manufacturing

Abstract

The present invention discloses a multi-target real-time monitoring device and system for additive manufacturing, including a multi-target real-time monitoring device and a brain-like computing chip with a heterogeneous fusion architecture. The monitoring device includes an excitation laser generator, a compressed sensing camera, an infrared thermometer, an infrared thermal imager, a photodiode, a galvanometer, and a beam splitter optical path component composed of three beam splitters. The excitation laser emitted by the excitation laser generator reaches the galvanometer through the beam splitter optical path component, and is scanned on the detection object by the galvanometer as an excitation signal; after receiving the excitation signal, the detection object generates reflected light, which is divided into multiple light paths with different wavelengths through the beam splitter optical path component, and reaches the corresponding light sensor respectively for corresponding monitoring. The present invention can compress sampling, high-speed transmission and reconstruction of image data of the additive manufacturing process, monitor the molten pool temperature in real time and extract image video features, and realize real-time monitoring of multi-target information across scales of molten pool temperature, sputtering, smoke plume and component layer contour.

Description

Technical Field

[0001] The present invention belongs to the field of additive manufacturing, and relates to an additive manufacturing molten pool monitoring technology, and specifically to a multi-target real-time monitoring device and system for additive manufacturing. Background Art

[0002] In laser powder bed fusion additive manufacturing, the behavior of variables such as molten pool morphology and temperature, sputtering and surface morphology directly affects the performance of manufactured parts. At present, the quality inspection of products mainly relies on offline methods, which requires a lot of material, financial and human resources, and cannot fully guarantee the quality of products under optimized processes. The online monitoring and feedback control of key variables through various monitoring equipment such as high-speed cameras and photodiodes is the mainstream method to solve metallurgical defects and shape control. However, due to the speed limit of imaging devices, the existing imaging monitoring method is difficult to continuously capture the change process within the microsecond time scale. At the same time, conventional monitoring methods will inevitably generate massive data and increase data storage and processing time under the premise of achieving high precision, which is far from meeting the needs of real-time monitoring and control of additive manufacturing processes. Under limited hardware equipment technology, how to efficiently sample and restore the monitored multivariate signals, a key foundation of modern signal processing is Shannon sampling theory, that is, the number of discrete samples required for a signal to be reconstructed without distortion requires that its sampling frequency must be greater than or equal to twice the highest frequency of the signal. However, according to Shannon sampling theory, the huge data signal of the monitored video image is difficult to meet the real-time requirements of the additive manufacturing process. In addition, the metal additive manufacturing process involves multi-scale and multi-physical field problems, including surface defects and temperature information of the molten pool. Currently, there is a lack of systems that can comprehensively and simultaneously monitor and feedback the surface defects and temperature signals of the molten pool. Moreover, the signal processing, artificial intelligence and other image processing algorithms in these online monitoring processes also face the problems of fast target movement speed of part geometric features and low transmission processing efficiency.

[0003] In addition, the multiple parameter monitoring equipment has a complex structure and a large size, which seriously affects the setting of the optical path and powder feeding mechanism of metal additive manufacturing. Therefore, it is urgent to develop a set of data compression sampling devices and systems that can efficiently and quickly collect and reconstruct the original multivariate signals with high quality, so as to realize real-time in-situ monitoring of additive manufacturing, reduce the space occupied by detection equipment, and improve the flexibility of laser cladding and powder feeding. Summary of the invention

[0004] The present invention aims to solve the problems of harsh online visual monitoring conditions and low processing efficiency in the laser powder bed fusion manufacturing process, and builds a multi-target real-time monitoring device and system for additive manufacturing. The system uses an infrared temperature measurement system and a photodiode to efficiently monitor the temperature signal of the molten pool in order to solve the problems of small molten pool area, fast solidification, high temperature and many interference factors in online monitoring. Furthermore, in order to solve the problems of large amount of monitoring signal data, low transmission efficiency and low signal-to-noise ratio in the laser powder bed fusion additive manufacturing process, the present invention designs and builds a high-speed imaging camera based on compressed sensing technology to achieve real-time and continuous capture of the processing process, and uses a DMD mask matrix for the molten pool signal to effectively remove redundant data in the signal without affecting the detection accuracy of the system. At the decoding end, the original image is reconstructed by random demodulation and reconstruction technology to shorten the image transmission time, and realize real-time monitoring of the processing process. Compressed sensing technology provides a solution to the difficult problems of low resolution and large amount of data in the effective monitoring signal in the laser powder bed fusion additive manufacturing process. At the same time, during the monitoring process, the edge computing method of complex signal features is integrated into the brain-like computing chip of the heterogeneous fusion architecture. With the help of the integrated storage and computing design of the brain-like computing chip, the inherent access rate limit of the von Neumann architecture can be broken, and TB-level monitoring signal processing can be completed per minute. Finally, the compressed sensing camera, temperature monitoring optical path, and brain-like computing chip are integrated to realize the real-time monitoring of cross-scale multi-target information of the melt pool temperature, sputtering, smoke plume, and component level contour during the processing process.

[0005] The above-mentioned object of the present invention is achieved through the following scheme:

[0006] A multi-target real-time monitoring device, comprising:

[0007] A galvanometer, used to scan the excitation laser and collect reflected light;

[0008] An excitation laser generator is used to emit an excitation laser of a set wavelength and is connected to the galvanometer through a beam splitting optical path component;

[0009] The detection component is connected to the galvanometer through the splitter optical path component, and includes a plurality of light sensors for detecting different parameters. The splitter optical path component splits the working light path according to the wavelength, so that the plurality of light sensors work in different bands without interfering with each other;

[0010] The excitation laser emitted by the excitation laser generator reaches the galvanometer through the splitter optical path component, and is scanned on the detection object by the galvanometer as an excitation signal; after receiving the excitation signal, the detection object generates reflected light, and the reflected light is divided into multiple optical paths with different wavelengths by the splitter optical path component, and reaches the corresponding light sensors respectively, thereby realizing real-time synchronous monitoring of multiple targets of the same object.

[0011] Preferably, the three optical sensors are a compressed sensing camera, an infrared thermometer and a photodiode.

[0012] Preferably, the light splitting optical path component comprises three mutually parallel and semi-transparent and semi-reflective beam splitters, the first beam splitter is a bandpass beam splitter, the second beam splitter and the third beam splitter are long-pass beam splitters with successively increasing working wavelengths or short-pass beam splitters with successively decreasing working wavelengths;

[0013] The excitation laser generator and the galvanometer are respectively arranged on two opposite sides of the first beam splitter, and the bandpass wavelength of the first beam splitter matches the excitation laser; the three light sensors are arranged in turn on the reflection light path of the second beam splitter, the reflection light path of the third beam splitter, and the transmission light path of the third beam splitter; the working wavelengths of the three light sensors increase or decrease in turn according to the types of the second beam splitter and the third beam splitter, and are staggered with the wavelength of the excitation laser;

[0014] The excitation laser emitted by the excitation laser generator passes through the first bandpass beam splitter to reach the galvanometer, and generates an excitation signal on the measurement object through the scanning of the galvanometer; after receiving the excitation signal, the detection object generates reflected signal light, and the reflected signal light is collected by the galvanometer and reaches the first beam splitter. After being reflected by the first beam splitter, the reflected signal light is separated into different wavelengths through the second and third beam splitters and sent to the corresponding light sensors.

[0015] Preferably, the operating wavelengths of the compressed sensing camera, infrared thermometer and photodiode increase sequentially, and correspondingly, the second beam splitter and the third beam splitter are long-pass beam splitters with successively increasing operating wavelengths, and the operating wavelengths are respectively denoted as A1 and A2; the compressed sensing camera is arranged in the reflection light path of the second beam splitter, and the operating wavelength is less than or equal to A1, the infrared thermometer is arranged in the reflection light path of the third beam splitter, and the operating wavelength is between A1 and A2, and the photodiode is arranged in the transmission light path of the third beam splitter, and the operating wavelength is greater than A2.

[0016] Preferably, the multi-target real-time monitoring device also includes an infrared thermal imager for synchronously aligning the detection object.

[0017] Preferably, the excitation laser generator comprises a laser and a beam expander, and the beam expander is used to expand the laser light emitted by the laser before emitting it.

[0018] Preferably, the compressed sensing camera comprises a high-speed photodetector and a femtosecond pulse laser, a beam splitter, a digital micromirror array and an acousto-optic deflector which are sequentially arranged on the same optical path; the high-speed photodetector is arranged on the reflection optical path of the beam splitter;

[0019] The femtosecond pulse laser emits a femtosecond pulse laser which passes through a beam splitter, a digital micromirror array and an acousto-optic deflector in sequence as an excitation signal for a sensing camera, and is reflected to a galvanometer through a second beam splitter and a first beam splitter in sequence, and is scanned on a detection object through the galvanometer;

[0020] The detection object receives the excitation signal of the sensing camera and generates a reflected signal. After being collected by the galvanometer, it is reflected by the first and second beam splitters and reaches the compressed sensing camera. The reflected signal passes through the acousto-optic deflector, digital micromirror array and beam splitter in turn and is detected by the high-speed photodetector.

[0021] Preferably, a total reflection mirror for changing the direction of the light path and a lens for focusing light are provided between the acousto-optic deflector and the second beam splitter of the compressed sensing camera.

[0022] The present invention also protects a multi-target real-time monitoring system for additive manufacturing, comprising any one of the multi-target real-time monitoring devices and a brain-like computing chip with a heterogeneous fusion architecture, wherein the brain-like computing chip has a built-in neural network algorithm for extracting features from information collected by a compressed sensing camera.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The compressed sensing camera designed by the present invention can realize the compressed sampling of the surface signal of the molten pool during the processing. By designing a mask matrix for the molten pool signal, the redundant data in the signal is effectively removed without affecting the detection accuracy of the system, and the image reconstruction algorithm is optimized to shorten the image transmission time, so as to realize the real-time monitoring of the processing process.

[0025] The molten pool temperature monitoring optical path designed by the present invention, the multi-spectral temperature detector combined with the infrared temperature measurement system and the photodiode complement each other's advantages, and can collect the higher temperature information of the molten pool at high speed;

[0026] The present invention uses the compressed sensing camera to compress and sample the molten pool morphology signal, and then extracts the surface features of the molten pool and reconstructs the original signal at high speed and high efficiency through the brain-like chip (48TFLOPS@FP16) based on the neural network and deep compressed sensing network at the output end, greatly improving the algorithm operation rate. The brain-like chip uses the physical characteristics of edge computing and storage, and is naturally compatible with edge computing. At the raw data collection end, it completes the preprocessing of the raw data, reduces the generation of redundant data, and reduces the cost of data storage, processing, and transmission.

[0027] The present invention has built a dynamic vision, infrared thermal imager, diode, and multi-type metal additive manufacturing online monitoring system to obtain the dynamic behavior of the molten pool and the macro-microstructure characteristic information, and can simultaneously monitor the molten pool, smoke plume, and workpiece morphology at multiple spatial scales. A compressed sensing edge computing method for multi-source monitoring signals is proposed, and for the first time, an autonomous and controllable brain-like computing chip is combined to quickly process compressed sensing monitoring signals online at the edge to achieve fast and efficient online detection.

[0028] The multi-target real-time monitoring system constructed by the present invention combines the optical paths of multiple monitoring technologies into one and adopts one galvanometer, which greatly saves floor space, greatly improves the flexibility of additive manufacturing, and helps to improve the additive manufacturing speed and the ability to manufacture complex workpieces. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Shown is a schematic diagram of a multi-target real-time monitoring system for additive manufacturing.

[0030] Figure 2 Shown is the temperature monitoring optical path diagram of the multi-target real-time monitoring device of the present invention.

[0031] Figure 3 Shown is the optical path diagram of the compressed sensing camera of the present invention.

[0032] Figure 4 Shown is the flow chart of compressed sensing AI chip feature extraction.

[0033] 1-photodiode, 2-infrared thermometer, 3-compressed sensing camera, 4-excitation laser generator, 41-laser, 42-beam expander, 5-brain-like computing chip, 6-infrared thermal imager, 7-first beam splitter, 8-second beam splitter, 9-third beam splitter, 10-galvanometer, 11-molten pool, 12-laser line column 1, 13-laser line column 2, 14-lens, 15-reflector, 16-acoustic-optic deflector, 17-beam splitter, 18-digital micromirror array, 19-femtosecond pulse laser, 20-high-speed photodetector, 21-workpiece, 22-working platform. DETAILED DESCRIPTION

[0034] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0035] like Figures 1 to 3 As shown, the present invention provides a multi-target real-time monitoring system for additive manufacturing, including a multi-target real-time monitoring device and a brain-like computing chip 5 with a heterogeneous fusion architecture;

[0036] The multi-target real-time monitoring device comprises a galvanometer 10, a light splitting optical path component, an excitation laser generator 4 and a detection component, wherein the excitation laser generator 4 is used to emit an excitation laser of a set wavelength and is connected to the galvanometer 10 through the light splitting optical path component;

[0037] The detection component is also connected to the galvanometer 10 through the split light path component, which includes a plurality of light sensors for detecting different parameters. The split light path component splits the working light path according to the wavelength, so that the plurality of light sensors work in different bands without interfering with each other;

[0038] The excitation laser emitted by the excitation laser generator 4 reaches the galvanometer 10 through the splitter optical path component, and is scanned on the detection object by the galvanometer 10 as an excitation signal; the detection object generates reflected light after receiving the excitation signal, and the reflected light is divided into multiple optical paths with different wavelengths by the splitter optical path component, and reaches the corresponding light sensors respectively, thereby realizing real-time synchronous monitoring of multiple targets of the same object.

[0039] The brain-like computing chip 5 has a built-in neural network algorithm for extracting features from the information collected by one or more light sensors.

[0040] The present invention utilizes an excitation laser generator 4 to emit an excitation laser, and the spectroscopic optical path component filters the excitation laser and then scans the molten pool 11 and the surrounding area of ​​the additive manufacturing through a galvanometer 10; then, different optical sensors are used to monitor the molten pool 11 and the surrounding area at different spatial scales and target parameters, and the monitoring information obtained by one or more optical sensors is feature extracted through the brain-like computing chip 5 to obtain the temperature information of the molten pool 11, sputtering, smoke plume and cross-scale multi-target information of the component layer contour.

[0041] As a specific embodiment, the optical sensors used in the present invention are a compressed sensing camera 3, an infrared thermometer 2 and a photodiode 1; the present invention measures the temperature information of the molten pool 11 with high speed and accuracy through the infrared thermometer 2 and the photodiode 1. The infrared thermometer 2 can monitor the temperature information of the molten pool 11 at 3000°C, and the photodiode 1 can measure the temperature signal of the molten pool 11 at high speed. The advantages of the two complement each other to achieve rapid acquisition and monitoring of higher temperature characteristics of the molten pool 11.

[0042] The compressed sensing camera 3 module performs sparse sampling of the original signal through a dynamic mask generated by a digital micromirror array 18 (DMD), which greatly reduces the amount of transmitted data. The decoding end adopts random demodulation and reconstruction technology, and the brain-like computing chip 5 with a built-in neural network algorithm can realize online high-precision reconstruction of the surface morphology of the molten pool 11.

[0043] The brain-like computing chip 5 used in the present invention breaks through the inherent memory access rate limit of the von Neumann architecture by virtue of the integrated storage and computing design of the brain-like computing chip, and completes TB-level monitoring signal processing per minute. The system image feature extraction algorithm and the compressed sensing image reconstruction algorithm are transmitted to the brain-like computing chip 5. The monitoring signal processing efficiency is improved through the coordinated optimization of software and hardware, and multiple types of abnormal manufacturing status features can be identified online.

[0044] As a preferred embodiment, Figure 2 As shown, the light splitting optical path component includes three mutually parallel and semi-transparent and semi-reflective beam splitters (dichroic mirrors), the first beam splitter 7 is a bandpass beam splitter, the second beam splitter 8 and the third beam splitter 9 are long-pass beam splitters with successively increasing working wavelengths or short-pass beam splitters with successively decreasing working wavelengths;

[0045] The excitation laser generator 4 and the galvanometer 10 are respectively arranged on two opposite sides of the first beam splitter 7, and the bandpass wavelength of the first beam splitter 7 matches the excitation laser; the three light sensors are respectively arranged in the reflection light path of the second beam splitter 8, the reflection light path of the third beam splitter 9 and the transmission light path of the third beam splitter 9; the working wavelengths of the three light sensors are increased or decreased in sequence according to the types of the second beam splitter 8 and the third beam splitter 9 to achieve optical path signal distinction; and they are staggered with the wavelength of the excitation laser to prevent signal interference;

[0046] For example, the second beam splitter 8 and the third beam splitter 9 are long-pass beam splitters whose working wavelengths increase sequentially, and the working wavelengths of the corresponding three light sensors increase sequentially according to the types of the second beam splitter 8 and the third beam splitter 9;

[0047] The second beam splitter 8 and the third beam splitter 9 are short-pass beam splitters whose working wavelengths decrease sequentially; the working wavelengths of the corresponding three light sensors decrease sequentially according to the types of the second beam splitter 8 and the third beam splitter 9.

[0048] The signal monitoring principle of the present invention is as follows:

[0049] The excitation laser emitted by the excitation laser generator 4 passes through the bandpass first beam splitter 7 to reach the galvanometer 10, and generates an excitation signal on the measurement object through scanning by the galvanometer 10; after receiving the excitation signal, the detection object generates reflected signal light, and the reflected signal light is collected by the galvanometer 10 and reaches the first beam splitter 7. After being reflected by the first beam splitter 7, the reflected signal light is separated into different wavelengths through the second and third beam splitters 9 and sent to the corresponding light sensors.

[0050] Through the design of the spectroscopic optical path component of the present invention, the three optical sensors (compressed sensing camera 3, infrared thermometer 2 and photodiode 1) of the present invention can share a reflected light signal, which not only greatly reduces the optical path component equipment, but also, most importantly, the three optical sensors truly realize synchronous monitoring, eliminating the time difference from the technical principle, so that the multi-target information extracted by the present invention is monitoring information that completely overlaps in time and space.

[0051] As a preferred embodiment, the excitation laser generator 4 includes a laser 41 and a beam expander 42 for increasing the diameter of the laser spot. The laser 41 is a pulsed laser. The beam expander 42 can ensure that the coverage range of the excitation laser is sufficient to cover the entire molten pool 11 and its vicinity.

[0052] As a preferred embodiment, the present invention additionally designs an infrared thermal imager that is synchronously aimed at the molten pool 11 area. Different from the infrared thermometer 2, the infrared thermal imager has the advantage of a wide measurement range and can obtain temperature distribution information, which is an effective supplement to the monitoring of the additive process. Since the infrared thermal imager itself is a wide-area sensor, it does not need to be added to the spectroscopic optical path component of the present invention to measure the temperature of the molten pool 11 and the surrounding area separately.

[0053] As a preferred embodiment, Figure 3 As shown, the compressed sensing camera 3 includes a high-speed photodetector 20 and a femtosecond pulse laser 19, a beam splitter 17, a digital micromirror array 18 and an acousto-optic deflector 16 which are sequentially arranged on the same optical path; the high-speed photodetector 20 is arranged on the reflection optical path of the beam splitter 17; the femtosecond pulse laser 19 emits a femtosecond pulse laser which sequentially passes through the beam splitter 17, the digital micromirror array 18 and the acousto-optic deflector 16 as a sensing camera excitation signal, which is sequentially reflected to the galvanometer 10 through the second beam splitter 8 and the first beam splitter 7, and is scanned on the detection object by the galvanometer 10. The detection object receives the sensing camera excitation signal to generate a reflection signal, which is collected by the galvanometer 10, and then reflected by the first beam splitter 7 and the second beam splitter 8 to reach the compressed sensing camera 3, and the reflection signal is sequentially detected by the high-speed photodetector 20 after passing through the acousto-optic deflector 16, the digital micromirror array 18 and the beam splitter 17.

[0054] The detection light used by the compressed sensing camera 3 of the present invention is an ultrashort (femtosecond) light pulse, which is combined with the space-time frequency dispersion technology to quickly generate a randomly encoded mask image in the time domain, breaking through the rate limit of the existing encoding device and achieving an ultra-high encoding rate. Combined with the high-speed continuous sampling capability of the high-speed photodetector 20, online high-speed monitoring of the dynamic behavior of the molten pool 11 is achieved. When the monitoring data is transmitted to the signal processing end at high speed, according to the random demodulation and reconstruction technology, the limitation of the Nyquist sampling theorem is broken, and high-precision reconstruction of the monitoring image video information is achieved.

[0055] Further preferably, a total reflection mirror 15 for changing the direction of the light path and a lens 14 for focusing light are provided between the acousto-optic deflector 16 and the second beam splitter 8 of the compressed sensing camera 3 of the present invention, and the femtosecond pulse laser can be focused by the lens 14 .

[0056] It should be noted that the basic technical problem to be solved by the present invention is how to realize simultaneous monitoring of multiple parameters. Therefore, the above-mentioned compressed sensing camera 3 is only a better real-time method and does not represent a limitation on the protection scope of the present invention. Other compressed sensing cameras or photo sensors may also be used as substitutes without affecting the implementation of the technical solution of the present invention.

[0057] Based on the above design, the present invention provides a specific parameter design case of a multi-target real-time monitoring system, which is as follows:

[0058] The infrared temperature measurement system is composed of an infrared thermometer 2 (spectral range: 0.8μm~1.0μm) and an infrared thermal imager 6 (spectral range: 0.5μm~0.54μm, 0.8μm);

[0059] The first beam splitter 7 is a 1.0μm ~ 1.1µm bandpass beam splitter, the second beam splitter 8 is a 0.81μm long pass beam splitter, and the third beam splitter 9 is a 1.1μm long pass beam splitter; the working wavelength of the laser 41 is selected to be 1.0μm ~ 1.1µm, specifically 1064nm; the working spectrum range of the infrared thermometer 2 is 0.81μm ~ 1.0μm; the working spectrum range of the infrared thermal imager 6 is: 0.5μm ~ 0.54μm, 0.8μm; the working spectrum range of the compressed sensing camera 3 is 0.78-0.8μm, specifically 0.8μm can be selected; the working spectrum range of the photodiode 1 is 1.1μm ~ 3.0μm;

[0060] The working wavelength of the galvanometer 10 is 780nm, and the working wavelength of the femtosecond pulse laser 19 is 0.8μm;

[0061] The brain-like computing chip 5 adopts a 48TFLOPS@FP16 model, with built-in image feature extraction algorithm and compressed sensing algorithm (both are known algorithms in the prior art). The brain-like chip is used to improve the system image feature extraction rate. Through the image feature extraction algorithm and the compressed sensing algorithm, the algorithm is integrated into the brain-like chip at the signal output end to quickly reconstruct the original signal, thereby extracting and reconstructing the surface features of the molten pool 11 at a high speed and efficiently.

[0062] Specifically, the laser 41 (wavelength 1064nm) emits a laser beam which is expanded by a beam expander 42 and then passes through a first beam splitter 7 (1.0μm ~1.1µm bandpass) and a galvanometer 10 as an excitation laser to reach a molten pool 11; among the reflected light from the molten pool 11, the light in the spectral range of 0.5μm~0.54μm and 0.8μm is received by the infrared thermal imager 6; the reflected light in the remaining spectral range is deflected by the galvanometer 10 and reaches the first beam splitter 7, with a wavelength of 1.0μm. ~1.1µm are reflected onto the second beam splitter 8; among the reflected light reaching the second beam splitter 8, the wavelength below 0.81μm is totally reflected onto the compressed sensing camera 3 and is used by the compressed sensing camera 3, and the reflected light above 0.81μm all passes through to reach the third beam splitter 9; among the reflected light reaching the third beam splitter 9, the wavelength below 1.1μm is totally reflected onto the infrared thermometer 2 and is used by the infrared thermometer 2 (0.81μm~1.0μm), and the reflected light above 1.1μm all passes through to reach the photodiode 1 and is used by the photodiode 1 (1.1μm ~3.0μm).

[0063] For the compressed sensing camera 3, the working wavelength of the femtosecond pulse laser 19 is 0.8μm; the femtosecond pulse laser 19 outputs an 800nm ​​laser beam, which passes through the DMD, beam splitter 17, acousto-optic deflector 16, reflector 15, lens 14, second beam splitter 8, first beam splitter 7 and galvanometer 10 to reach the surface of the molten pool 11. The light beam (laser line column 2) reflected by the molten pool 11 passes through the galvanometer 10, first beam splitter 7, lens 14, reflector 15, acousto-optic deflector 16, DMD and beam splitter 17, and finally forms an image on the high-speed photodetector 20. The femtosecond pulse laser 19 of the present invention has the advantages of high light output power, stable wavelength, spectrum width up to 80nm, high isolation, ultra-stable laser, mid-infrared tunability, etc., and can be used for rapid monitoring of different temperature environments. The working band of the galvanometer 10 is 780nm, which is used to adjust the beam angle to process and scan the workpiece 21. In order to match sensors with different wavelength ranges, the second beam splitter 8 uses a 810nm long pass to select (reflect and project) light; the lens 14 uses a conventional lens with high transmittance and an operating wavelength of 780nm to converge light; the reflector 15 uses a conventional lens with high transmittance and an operating wavelength of 780nm to reflect light; the acousto-optic deflector 16 uses G&H's AODF4085, which controls the direction of the laser through circuit signals and shortens the reaction delay, and can adjust the beam angle to scan the workpiece 21. The beam splitter 17 has an operating band of 780nm, and by separating the round-trip light, the reflected light (workpiece 21 information) is received by the detector; the DMD uses vialux's V-650L, which selectively reflects the monitored light signal through a mask, thereby realizing image signal compression and decoding at the acquisition end.

[0064] It should be noted that the working wavelength of the above-mentioned light sensor and the working wavelength of the spectroscope are not limited to the above-mentioned example parameters. As long as the reflected light can be split according to the wavelength to the corresponding light sensor and do not interfere with each other, the working wavelength of the light sensor or the spectroscope can be adjusted according to actual conditions.

[0065] The above embodiments are only used to illustrate the present invention, but not to limit the present invention. Although the present invention is described in detail with reference to the embodiments, it should be understood by those skilled in the art that various combinations, modifications or equivalent substitutions of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and should be included in the scope of the claims of the present invention.

Claims

1. A multi-target real-time monitoring device, characterized in that: include A galvanometer, used to scan the excitation laser and collect reflected light; An excitation laser generator is used to emit an excitation laser of a set wavelength and is connected to the galvanometer through a beam splitting optical path component; The detection component is connected to the galvanometer through the splitter optical path component, and includes a plurality of light sensors for detecting different parameters. The splitter optical path component splits the working light path according to the wavelength, so that the plurality of light sensors work in different bands without interfering with each other; The excitation laser emitted by the excitation laser generator reaches the galvanometer through the beam splitter optical path component, and is scanned on the detection object by the galvanometer as an excitation signal; the detection object generates reflected light after receiving the excitation signal, and the reflected light is divided into multiple optical paths with different wavelengths by the beam splitter optical path component, and reaches the corresponding light sensors respectively, so as to realize real-time synchronous monitoring of multiple targets of the same object; The light splitting optical path component comprises three mutually parallel semi-transparent and semi-reflective beam splitters, the first beam splitter is a band-pass beam splitter, and the second and third beam splitters are long-pass beam splitters with successively increasing working wavelengths; The excitation laser generator and the galvanometer are respectively arranged on two opposite sides of the first beam splitter, and the bandpass wavelength of the first beam splitter matches the excitation laser; the three light sensors are arranged in turn on the reflection light path of the second beam splitter, the reflection light path of the third beam splitter, and the transmission light path of the third beam splitter; the working wavelengths of the three light sensors increase in turn according to the types of the second beam splitter and the third beam splitter, and are staggered with the wavelength of the excitation laser; The excitation laser emitted by the excitation laser generator passes through the first bandpass beam splitter to reach the galvanometer, and generates an excitation signal on the measurement object through the scanning of the galvanometer; after receiving the excitation signal, the detection object generates reflected signal light, and the reflected signal light is collected by the galvanometer and reaches the first beam splitter. After being reflected by the first beam splitter, the reflected signal light is separated into different wavelengths through the second and third beam splitters and sent to the corresponding light sensor; The three light sensors are a compressed sensing camera, an infrared thermometer and a photodiode; the working wavelengths of the compressed sensing camera, the infrared thermometer and the photodiode increase sequentially, and correspondingly, the second beam splitter and the third beam splitter are long-pass beam splitters with working wavelengths increasing sequentially, and the working wavelengths are denoted as A1 and A2 respectively; the compressed sensing camera is arranged in the reflection light path of the second beam splitter, and the working wavelength is less than or equal to A1, the infrared thermometer is arranged in the reflection light path of the third beam splitter, and the working wavelength is between A1 and A2, and the photodiode is arranged in the transmission light path of the third beam splitter, and the working wavelength is greater than A2.

2. The multi-target real-time monitoring device according to claim 1, characterized in that: It also includes an infrared thermal imager for synchronously aligning the inspection object.

3. The multi-target real-time monitoring device according to claim 2, characterized in that: The excitation laser generator comprises a laser and a beam expander, and the beam expander is used for expanding the laser light emitted by the laser and then emitting it.

4. The multi-target real-time monitoring device according to claim 1, characterized in that: The compressed sensing camera comprises a high-speed photodetector and a femtosecond pulse laser, a beam splitter, a digital micromirror array and an acousto-optic deflector which are sequentially arranged on the same optical path; the high-speed photodetector is arranged on the reflection optical path of the beam splitter; The femtosecond pulse laser emits a femtosecond pulse laser which passes through a beam splitter, a digital micromirror array and an acousto-optic deflector in sequence as an excitation signal for a sensing camera, and is reflected to a galvanometer through a second beam splitter and a first beam splitter in sequence, and is scanned on a detection object through the galvanometer; The detection object receives the excitation signal of the sensing camera and generates a reflected signal. After being collected by the galvanometer, it is reflected by the first and second beam splitters and reaches the compressed sensing camera. The reflected signal passes through the acousto-optic deflector, digital micromirror array and beam splitter in turn and is detected by the high-speed photodetector.

5. The multi-target real-time monitoring device according to claim 4, characterized in that: A total reflection mirror for changing the direction of the light path and a lens for focusing light are also provided between the acousto-optic deflector and the second beam splitter of the compressed sensing camera.

6. A multi-target real-time monitoring system for additive manufacturing, characterized in that: It comprises the multi-target real-time monitoring device as described in any one of claims 1 to 5 and a brain-like computing chip with a heterogeneous fusion architecture, wherein the brain-like computing chip has a built-in neural network algorithm for extracting features from information collected by a compressed sensing camera.

Images