A method and system for selective laser melting defect monitoring based on multiple sensors

Abstract

This invention discloses a multi-sensor-based method and system for monitoring defects in selected area laser melting. The invention employs multiple sensors to monitor the molten pool image, acoustic signals, and photoelectric signals during the printing process. Signal acquisition and preprocessing are performed on the molten pool image, acoustic signals, and photoelectric signals. Molten pool features are extracted from the molten pool image and unified in the temporal domain. Typical defect feature parameters are extracted from the acoustic signals, photoelectric signals, and the unified temporal-domain molten pool features, and these parameters are combined with the corresponding scanning positions to generate a defect feature location map based on the acoustic signals, photoelectric signals, and molten pool features. Based on the defect feature location map, a threshold defect judgment method is used to locate the defect location, which is then displayed and statistically analyzed. When the number of defects exceeds a preset range, the printing parameters are optimized and adjusted. This invention offers high real-time performance and is simple to operate.

Description

Technical Field

[0001] This invention relates to the field of metal 3D printing monitoring technology, specifically a method and system for monitoring selected area laser melting defects based on multiple sensors. Background Technology

[0002] Selective Laser Melting (SLM) is an important component of rapid prototyping technology. It is a rapid manufacturing technology that has developed in recent years, and compared to other rapid prototyping technologies, SLM is more efficient, convenient, and has a broader development prospect. However, the quality and reusability of metal parts have consistently hindered the application of SLM. SLM is a complex, highly dynamic process involving the coupling of multiple physical fields. The formed parts are prone to macroscopic defects such as warping, spheroidization, and cracking, as well as internal metallurgical defects such as porosity, inclusions, and lack of fusion, affecting the accuracy and reliability of the metal parts.

[0003] To overcome these shortcomings and manufacture high-quality parts, monitoring the selective laser melting (SLM) process is crucial. Current SLM monitoring primarily focuses on the acoustic, optical, thermal, and vibration signals emitted during the printing process, using sensors corresponding to the signal sources for signal acquisition. However, much research concentrates on single-sensor monitoring of a single signal source or multi-sensor monitoring of a single signal source, resulting in incomplete information and insufficient accuracy. Furthermore, it can only assess the overall quality of the printed part, failing to pinpoint the specific location of defects. Multi-sensor monitoring generates a large amount of homogeneous but heterogeneous data, making it difficult to establish correlations between data points, and the typically time-consuming processing methods severely impact real-time monitoring. Regarding parameter optimization, the overall quality of the finished part is usually used as the evaluation criterion for parameter optimization, making the experimental process cumbersome and wasting significant human and material resources. Summary of the Invention

[0004] Purpose of the invention: This invention addresses the problems existing in the prior art by providing a multi-sensor-based method and system for monitoring selected area laser melting defects that is highly real-time and easy to operate.

[0005] Technical solution: The multi-sensor-based selective laser melting defect monitoring method of the present invention includes the following steps:

[0006] (1) Use multiple sensors to monitor the molten pool image, acoustic signal and photoelectric signal during the printing process;

[0007] (2) The molten pool image, acoustic signal and photoelectric signal are acquired and preprocessed;

[0008] (3) Extract the molten pool features from the molten pool image and unify the molten pool features in the time domain;

[0009] (4) Extract typical defect feature parameters from acoustic signals, photoelectric signals and time-domain unified molten pool features, and combine the defect feature parameters with the corresponding scanning positions to generate a defect feature location map of acoustic signals, photoelectric signals and molten pool features;

[0010] (5) Based on the defect feature location map, the threshold defect judgment method is used to locate the defect location, and then the location is displayed and statistically analyzed;

[0011] (6) When the number of defects exceeds the preset range, the printing parameters are optimized and adjusted.

[0012] Furthermore, step (1) specifically includes:

[0013] (1-1) A high-speed camera is used to capture images of the molten pool during the printing process. The high-speed camera is mounted outside the chamber of the selective laser printer using a rangefinder mounting method.

[0014] (1-2) An acoustic microphone is used to monitor the acoustic signals generated during the printing process, and the acoustic sensor is installed in the printer compartment in a cross-axis manner;

[0015] (1-3) A photodiode is used to monitor the photoelectric signals generated during the printing process. The photodiode is installed in the printer compartment in a cross-axis manner.

[0016] Furthermore, step (3) specifically includes:

[0017] (3-1) Calculate the time for each molten pool image based on the set high-speed camera frame rate;

[0018] (3-2) Extract the molten pool width, molten pool area, and number of spatters from each molten pool image as molten pool features;

[0019] (3-3) Based on the extracted molten pool width, molten pool area, and number of spatters at each time point, respectively, fit the relationship curves between molten pool width, molten pool area, number of spatters and time.

[0020] Furthermore, the formula for calculating the time of the molten pool image is as follows:

[0021]

[0022] In the formula, t k is the time of the k-th molten pool image, f is the frame rate of the high-speed camera, and k represents the k-th molten pool image.

[0023] Furthermore, step (4) specifically includes:

[0024] (4-1) Extract typical defect feature parameters from acoustic signals, photoelectric signals and time-domain unified molten pool features respectively;

[0025] (4-2) Establish a scanning path grid plane based on the current printed layer shape and laser scanning strategy. Each grid in the scanning path plane represents the corresponding scanning position on the current printed layer.

[0026] (4-3) Obtain the minimum and maximum values ​​of the typical defect feature parameters of the acoustic signal, divide the range between the minimum and maximum values ​​into several equidistant intervals, assign a non-zero gray value to the typical defect feature parameter value of the acoustic signal located in each interval, fill it into the grid corresponding to the acoustic signal scanning position in the first scanning path grid plane, and generate the defect feature location map of the acoustic signal.

[0027] (4-4) Obtain the defect feature location map of photoelectric signal and the defect feature location map of molten pool feature by following the same steps.

[0028] Furthermore, the method for obtaining the grid corresponding to the acoustic signal scanning position in step (4-3) is as follows:

[0029] A. The scanning time t and the initial scanning time t0 for acquiring typical defect characteristic parameter values ​​of acoustic signals;

[0030] B. Calculate the difference between the scan time t and the initial scan time t0, and use it as the scan duration τ;

[0031] C. Multiply the scanning speed v and the scanning duration τ, and then divide by the scanning path length to obtain the quotient n and the remainder m. If m = 0, the scanning position corresponds to the last grid in the nth row of the scanning direction in the scanning path grid plane. If m ≠ 0, the scanning position corresponds to the mth grid in the (n+1)th row of the scanning direction in the scanning path grid plane.

[0032] Furthermore, step (5) specifically includes:

[0033] (5-1) Compare the gray value corresponding to each grid in the defect feature location map of acoustic signal, photoelectric signal and molten pool feature with the preset threshold range. When it exceeds the preset threshold range, it is determined that the current grid has a defect and the gray value of the current grid is updated to 0.

[0034] (5-2) The updated acoustic signal, photoelectric signal and molten pool feature defect location map are merged into a single defect location map. The location with a gray value of 0 on the map is the defect location, and the number of defects is counted.

[0035] Furthermore, step (6) specifically includes:

[0036] (6-1) When the number of defects exceeds the set range, design a four-factor, five-level orthogonal process parameter group. The four factors are laser power, scanning speed, scanning spacing and slice layer thickness. The five levels are the average of the set range of each factor from low to high, divided into five levels.

[0037] (6-2) Print 20 layers for each process parameter group and record the number of defects generated during the printing process;

[0038] (6-3) Using the number of defects as the evaluation criterion for parameter optimization, a multi-objective optimization mathematical model is established. The genetic algorithm is used to solve the multi-objective optimization mathematical model to obtain the optimal combination of laser power, scanning speed, scanning spacing and slice thickness.

[0039] The multi-sensor-based selective laser melting defect monitoring system of the present invention includes:

[0040] A multi-sensor module is used to monitor the molten pool image, acoustic signals, and photoelectric signals during the printing process;

[0041] The signal acquisition module is used to acquire the molten pool image, acoustic signal, and photoelectric signal.

[0042] The data processing module is used to perform signal preprocessing and extract molten pool features from the molten pool image. The molten pool features are unified in the time domain. Then, typical defect feature parameters are extracted from the acoustic signal, photoelectric signal and the unified molten pool features in the time domain. The defect feature parameters are combined with the corresponding scanning position to generate a defect feature location map. Finally, the defect location is located based on the defect feature location map using a threshold defect judgment method, and the location is displayed and statistically analyzed.

[0043] The parameter optimization module is used to optimize and adjust printing parameters when the number of defects exceeds the preset range.

[0044] Furthermore, the multi-sensor module specifically includes:

[0045] A high-speed camera, mounted on a rangefinder, is installed outside the chamber of the selective laser printer to capture images of the molten pool during the printing process.

[0046] An acoustic microphone, mounted in a rangefinder configuration inside the printer compartment, is used to monitor acoustic signals generated during the printing process.

[0047] The photodiode, mounted off-axis in the printer compartment, is used to monitor photoelectric signals generated during the printing process.

[0048] Beneficial Effects: Compared with existing technologies, the significant advantages of this invention are: it can more comprehensively monitor the selective laser melting process, resulting in more accurate defect monitoring and overcoming the problems of incomplete and inaccurate defect monitoring by single sensors. Employing paraxial monitoring, the system is relatively easy to set up and implement. By unifying the dimensions of homogeneous and heterogeneous data, connections between data are established, enabling efficient data processing and improving the accuracy of defect detection. This invention achieves knowability and controllability of the selective laser melting process, enabling rapid defect location and optimization of printing parameters, playing a crucial role in improving printing quality and process optimization. Attached Figure Description

[0049] Figure 1 Diagram of a hardware system for multi-sensor selected area laser melting defect monitoring;

[0050] Figure 2 A schematic flowchart of the multi-sensor-based selective laser melting defect monitoring method provided by the present invention;

[0051] Figure 3 A schematic diagram of the dimensional unification process;

[0052] Figure 4 A schematic diagram of the defect location and display process;

[0053] Figure 5 A schematic diagram of the modules of the multi-sensor-based selective laser melting defect monitoring system provided by the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] Example 1

[0056] This embodiment provides a multi-sensor-based method for monitoring selected area laser melting defects. Figure 1The selective laser printer shown is monitored. This selective laser printer includes: a laser 1, a scanning galvanometer 2, a doctor blade 3, a powder supply hopper 4, a forming hopper 5, a substrate 6, a recovery tube 7, a powder recovery hopper 71, and a sealing plug 8. In this embodiment, the selective laser printer used is an iSLM150. The laser 1 is located at the top of the selective laser melting printer. The emitted laser is controlled by the scanning galvanometer 2 mounted in front of the laser 1 and irradiates the metal powder surface above the substrate 6 in the forming hopper 5 according to a specific scanning strategy. The forming hopper 5 is located in the lower middle part of the entire chamber. On its left side is a powder supply hopper 4, and the powder in the powder supply hopper 4 is delivered to the forming hopper 5 via the doctor blade 3 above. On the right side of the forming hopper 5 is a powder recovery tube 7, and below the recovery tube 7 is a powder recovery hopper 71, used to collect excess powder generated during the leveling of the substrate 6.

[0057] like Figure 2 As shown, the multi-sensor-based selected area laser melting defect monitoring method provided in this embodiment includes the following steps:

[0058] (1) Use multiple sensors to monitor the molten pool image, acoustic signal and photoelectric signal during the printing process.

[0059] refer to Figure 1 The system utilizes multiple sensors, specifically a high-speed camera 9, an acoustic microphone 10, and a photodiode 11, to monitor the molten pool image, acoustic signal, and photoelectric signal during the printing process, respectively. A bracket 12 is mounted on the right side of the recycle tube 7, housing the acoustic microphone 10 and photodiode 11 for monitoring the printing process. The microphone is an MPA416, and the photodiode is a DT-30V. This photodiode, through its built-in module, can directly convert light intensity signals into voltage signals, facilitating subsequent data acquisition. The high-speed camera 9 is mounted outside the printer compartment. The high-speed camera 9 is an ISP502, and the compartment door has been replaced with a transparent one. At the end of the recycle tube 7 is a sealing plug 8 with two cable holes for easy sensor cable routing. Simultaneously, to ensure airtightness throughout the compartment during printing, the sealing plug 8 guarantees airtightness, preventing oxidation of the printed parts. The bracket 12 has two rotatable clips 121 for fixing the acoustic microphone 10 and photodiode 11, and allows adjustment of the monitoring angles of the two sensors. The high-speed camera 9 is used to monitor morphological defects, such as holes and depressions. The acoustic microphone 10 reflects the porosity problem during the printing process. The photodiode 11 is used to determine the energy absorption of the molten pool and reflect defects such as poor fusion.

[0060] (2) The molten pool image, acoustic signal and photoelectric signal are acquired and preprocessed.

[0061] Data acquisition is achieved through a data acquisition card 13 and a host computer 14. The ribbon cables of the acoustic microphone 10 and photodiode 11 are connected to the external data acquisition card 13 via a recycling tube 7, and the high-speed camera 9 is directly connected to the host computer 14. The data acquisition card 13 converts electrical signals into digital signals and transmits them to the host computer 14. In this embodiment, an NI data acquisition card is used, and the host computer is a regular desktop computer. Data acquisition and data processing programs are written using the LabVIEW programming environment.

[0062] (3) Extract the features of the molten pool from the molten pool image and unify the molten pool features in the time domain.

[0063] In practice, the time for each molten pool image is calculated based on the set frame rate of the high-speed camera. The calculation formula is as follows:

[0064]

[0065] In the formula, t k is the time of the k-th molten pool image, f is the frame rate of the high-speed camera, and k represents the k-th molten pool image.

[0066] Then, using OpenCV, the molten pool width, molten pool area, and number of spatters were extracted from each molten pool image as molten pool features. Based on the extracted molten pool width, molten pool area, and number of spatters at each time point, curves relating molten pool width, molten pool area, and number of spatters to time were fitted by plotting points, as shown below. Figure 3 As shown.

[0067] (4) Extract typical defect feature parameters from acoustic signals, photoelectric signals and time-domain unified molten pool features, and combine the defect feature parameters with the corresponding scanning positions to generate a defect feature location map of acoustic signals, photoelectric signals and molten pool features.

[0068] This step specifically includes:

[0069] (4-1) Typical defect characteristic parameters were extracted from acoustic signals, photoelectric signals, and time-domain unified molten pool characteristics, respectively. Typical defect characteristic parameters can be obtained experimentally or extracted using existing methods. Preliminary experiments showed that comparing quantitative information such as molten pool size with standard molten pool size can identify defect anomalies; abrupt changes in sound pressure in the acoustic signal reflect the generation of fracture and porosity defects, and the variance reflects the stability of sound pressure to a certain extent; the photoelectric signal directly reflects the energy absorption of the molten pool, and the average value over a certain time interval can reflect the molten pool condition well. Therefore, the difference between the quantitative information of the molten pool and the standard molten pool was used as the characteristic parameter for each sampling moment; the variance of every 10 sampling points in the sound pressure signal was used as the characteristic parameter for this sampling time period; and the average value of every 10 sampling points in the photoelectric signal was used as the characteristic parameter for this sampling time period. Of course, the selection of characteristic parameters in actual situations is not limited to this.

[0070] (4-2) Establish a scanning path grid plane based on the current printed layer shape and laser scanning strategy. Each grid in the scanning path plane represents the corresponding scanning position on the current printed layer.

[0071] (4-3) Obtain the minimum and maximum values ​​of the typical defect feature parameters of the acoustic signal. Divide the range between the minimum and maximum values ​​into several equidistant intervals. Assign a non-zero grayscale value to the typical defect feature parameter value of the acoustic signal located in each interval and fill it into the grid corresponding to the acoustic signal scanning position in the first scanning path grid plane to generate a defect feature location map of the acoustic signal. In specific implementation, it can be divided into 50 equidistant intervals, and the grayscale values ​​in the range of 131-180 can be assigned values ​​according to the parameter values. The method for obtaining the grid corresponding to the acoustic signal scanning position is as follows: A. Obtain the scanning time t and the initial scanning time t0 of the typical defect characteristic parameter values ​​of the acoustic signal; B. Calculate the difference between the scanning time t and the initial scanning time t0 as the scanning duration τ; C. Multiply the scanning speed v and the scanning duration τ and divide by the scanning path length to obtain the quotient n and the remainder m. If m = 0, the scanning position corresponds to the last grid in the nth row of the scanning direction in the scanning path grid plane. If m ≠ 0, the scanning position corresponds to the mth grid in the (n+1)th row of the scanning direction in the scanning path grid plane.

[0072] (4-4) Following the same steps, the defect feature location maps of the photoelectric signal and the molten pool are obtained, such as... Figure 4 As shown.

[0073] (5) Based on the defect feature location map, the threshold defect judgment method is used to locate the defect location, and then display and count the defects.

[0074] In practice, the grayscale value corresponding to each grid in the defect feature location maps of acoustic signals, photoelectric signals, and molten pool features is compared with a preset threshold range. When the value exceeds the preset threshold range, it is determined that a defect has occurred in the current grid, and the grayscale value of the current grid is updated to 0. The updated defect feature location maps of acoustic signals, photoelectric signals, and molten pool features are then merged into a single defect location map. The locations with a grayscale value of 0 on the map represent defect locations, and the number of defects is counted. Figure 4 As shown, areas without defects are displayed in white, while areas with defects are displayed in black.

[0075] (6) When the number of defects exceeds the preset range, the printing parameters are optimized and adjusted.

[0076] When the number of defects exceeds the set range, an orthogonal process parameter set with four factors and five levels is designed. The four factors are laser power, scanning speed, scanning spacing, and slice layer thickness. The five levels are the average of the setting range of each factor from low to high, divided into five levels. Each process parameter set is used to print 20 layers, and the number of defects generated during the printing process is recorded. The number of defects is used as the evaluation criterion for parameter optimization. A multi-objective optimization mathematical model is established, and a genetic algorithm is used to solve the multi-objective optimization mathematical model to obtain the optimal parameter combination of laser power, scanning speed, scanning spacing, and slice layer thickness.

[0077] This invention allows for more comprehensive monitoring of the selective laser melting process, resulting in more accurate defect detection and overcoming the limitations of single-sensor monitoring in terms of completeness and accuracy. Utilizing a paraxial monitoring system, the system is relatively easy to set up and implement. By unifying the dimensions of heterogeneous data from the same source, connections between data points are established, enabling efficient data processing and improving the accuracy of defect detection. This invention achieves knowable and controllable selective laser melting processes, enabling rapid defect location and optimizing printing parameters, thus playing a significant role in improving printing quality and process optimization.

[0078] Example 2

[0079] This embodiment provides a multi-sensor-based selected area laser melting defect monitoring system, such as... Figure 5 As shown, it includes:

[0080] A multi-sensor module is used to monitor the molten pool image, acoustic signals, and photoelectric signals during the printing process;

[0081] The signal acquisition module is used to acquire the molten pool image, acoustic signal, and photoelectric signal.

[0082] The data processing module is used to perform signal preprocessing and extract molten pool features from the molten pool image. The molten pool features are unified in the time domain. Then, typical defect feature parameters are extracted from the acoustic signal, photoelectric signal and the unified molten pool features in the time domain. The defect feature parameters are combined with the corresponding scanning position to generate a defect feature location map. Finally, the defect location is located based on the defect feature location map using a threshold defect judgment method, and the location is displayed and statistically analyzed.

[0083] The parameter optimization module is used to optimize and adjust printing parameters when the number of defects exceeds the preset range.

[0084] The multi-sensor module specifically includes a high-speed camera, an acoustic microphone, and a photodiode: the high-speed camera is mounted outside the selective laser printer compartment in a rangefinder configuration to capture images of the molten pool during printing; the acoustic microphone is mounted inside the printer compartment in a rangefinder configuration to monitor acoustic signals generated during printing; and the photodiode is mounted inside the printer compartment in a rangefinder configuration to monitor photoelectric signals generated during printing.

[0085] The signal acquisition module includes an acquisition card and a host computer.

[0086] The data processing module includes a temporal unification unit, a defect feature location map acquisition unit, and a defect localization unit. The temporal unification unit calculates the time for each molten pool image based on the set high-speed camera frame rate, and extracts the molten pool width, molten pool area, and number of spatters from each image as molten pool features. Based on the extracted molten pool width, molten pool area, and number of spatters at each time point, it fits curves relating the molten pool width, molten pool area, and number of spatters to time. The defect feature location map acquisition unit extracts typical defect feature parameters from acoustic signals, photoelectric signals, and time-domain unified molten pool features, respectively. It establishes a scanning path grid plane based on the current printed layer shape and laser scanning strategy, with each grid representing a corresponding scanning position on the current printed layer. It acquires the minimum and maximum values ​​of the typical defect feature parameters of the acoustic signal, divides the range between the minimum and maximum values ​​into several equidistant intervals, assigns a non-zero grayscale value to the typical defect feature parameter value of the acoustic signal located in each interval, and fills it into the grid corresponding to the acoustic signal scanning position in the first scanning path grid plane, generating the defect feature location map of the acoustic signal. The same steps are followed to obtain the defect feature location maps of the photoelectric signal and the molten pool features. The defect location unit compares the gray value corresponding to each grid in the defect feature location map of acoustic signal, photoelectric signal and molten pool feature with a preset threshold range. When the gray value exceeds the preset threshold range, it is determined that there is a defect in the current grid and the gray value of the current grid is updated to 0. The updated defect feature location maps of acoustic signal, photoelectric signal and molten pool feature are merged into a single defect location map. The location with a gray value of 0 on the map is the defect location, and the number of defects is counted.

[0087] The parameter optimization module designs a four-factor, five-level orthogonal process parameter set when the number of defects exceeds the set range. The four factors are laser power, scanning speed, scanning spacing, and slice layer thickness. The five levels are achieved by dividing the setting range of each factor into five equal levels from low to high. Each process parameter set is used to print 20 layers, and the number of defects generated during the printing process is recorded. Using the number of defect occurrences as the evaluation criterion for parameter optimization, a multi-objective optimization mathematical model is established. A genetic algorithm is used to solve the multi-objective optimization mathematical model to obtain the optimal parameter combination of laser power, scanning speed, scanning spacing, and slice layer thickness.

[0088] The apparatus provided in this embodiment of the invention can be used to execute the method provided in Embodiment 1 of the invention, and has the corresponding functions and beneficial effects of executing the method. For details not covered herein, please refer to Embodiment 1, which will not be repeated here.

[0089] It is worth noting that in the embodiments of the above-mentioned determining device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0090] Each unit or module may be software, hardware, or a combination of software and hardware. Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

Claims

1. A method for monitoring selected area laser melting defects based on multiple sensors, characterized in that... The following steps are involved: (1) Use multiple sensors to monitor the molten pool image, acoustic signal and photoelectric signal during the printing process; (2) The molten pool image, acoustic signal and photoelectric signal are acquired and preprocessed; (3) Extract the molten pool features from the molten pool image and perform temporal unification on the molten pool features; (4) Extract typical defect feature parameters from acoustic signals, photoelectric signals and time-domain unified molten pool features, and combine the defect feature parameters with the corresponding scanning positions to generate a defect feature location map of acoustic signals, photoelectric signals and molten pool features; (5) Based on the defect feature location map, the threshold defect judgment method is used to locate the defect location, and then the location is displayed and statistically analyzed; (6) When the number of defects exceeds the preset range, optimize and adjust the printing parameters; Step (4) specifically includes: (4-1) Extract typical defect feature parameters from acoustic signals, photoelectric signals and time-domain unified molten pool features respectively; (4-2) Establish a scanning path grid plane based on the current printed layer shape and laser scanning strategy. Each grid in the scanning path plane represents the corresponding scanning position on the current printed layer. (4-3) Obtain the minimum and maximum values ​​of the typical defect feature parameters of the acoustic signal, divide the range between the minimum and maximum values ​​into several equidistant intervals, assign a non-zero gray value to the typical defect feature parameter value of the acoustic signal located in each interval, fill it into the grid corresponding to the scanning position of the acoustic signal in the first scanning path grid plane, and generate the defect feature location map of the acoustic signal. (4-4) Obtain the defect feature location map of the photoelectric signal and the defect feature location map of the molten pool by following the same steps; (5) Specifically includes: (5-1) Compare the gray value corresponding to each grid in the defect feature location map of acoustic signal, photoelectric signal and molten pool feature with the preset threshold range. When it exceeds the preset threshold range, it is determined that the current grid has a defect and the gray value of the current grid is updated to 0. (5-2) The updated acoustic signal, photoelectric signal and molten pool feature defect location map are merged into a single defect location map. The location with a gray value of 0 on the map is the defect location, and the number of defects is counted.

2. The method for monitoring selected area laser melting defects based on multiple sensors according to claim 1, characterized in that: Step (1) specifically includes: (1-1) A high-speed camera is used to capture images of the molten pool during the printing process. The high-speed camera is mounted outside the chamber of the selective laser printer using a rangefinder mounting method. (1-2) An acoustic microphone is used to monitor the acoustic signals generated during the printing process. The acoustic microphone is installed in the printer compartment in a rangefinder manner. (1-3) A photodiode is used to monitor the photoelectric signals generated during the printing process. The photodiode is installed in the printer compartment in a cross-axis manner.

3. The method for monitoring selected area laser melting defects based on multiple sensors according to claim 1, characterized in that: Step (3) specifically includes: (3-1) Calculate the time for each molten pool image based on the set high-speed camera frame rate; (3-2) Extract the molten pool width, molten pool area, and number of spatters from each molten pool image as molten pool features; (3-3) Based on the extracted molten pool width, molten pool area, and number of spatters at each time, respectively fit the relationship curves between molten pool width, molten pool area, number of spatters and time.

4. The method for monitoring selected area laser melting defects based on multiple sensors according to claim 3, characterized in that: The formula for calculating the time of the molten pool image is: ， In the formula, The time for the k-th melt pool image, f is the frame rate of the high-speed camera, and k represents the k-th molten pool image.

5. The method for monitoring selected area laser melting defects based on multiple sensors according to claim 1, characterized in that: The method for obtaining the grid corresponding to the acoustic signal scanning position in step (4-3) is as follows: A. The scanning time t and the initial scanning time t0 for acquiring typical defect characteristic parameter values ​​of acoustic signals; B. Calculate the difference between the scan time t and the initial scan time t0, and use it as the scan duration τ; C. Scanning speed v Multiply the scan duration τ by the scan path length to obtain the quotient n and remainder m. If m = 0, the scan position corresponds to the last grid in the nth row of the scan path grid plane. If the scan position corresponds to the m-th grid in the (n+1)-th row of the scan path grid plane, then the scan position is the m-th grid in the scan direction.

6. The method for selective laser melting defect monitoring based on multiple sensors according to claim 1, characterized in that: Step (6) specifically includes: (6-1) When the number of defects exceeds the set range, design a four-factor, five-level orthogonal process parameter group. The four factors are laser power, scanning speed, scanning spacing and slice layer thickness. The five levels are the average of the setting range of each factor from low to high, divided into five levels. (6-2) Print 20 layers for each process parameter group and record the number of defects generated during the printing process; (6-3) Using the number of defects as the evaluation criterion for parameter optimization, a multi-objective optimization mathematical model is established. The genetic algorithm is used to solve the multi-objective optimization mathematical model to obtain the optimal combination of laser power, scanning speed, scanning spacing and slice thickness.

7. A multi-sensor-based selective laser melting defect monitoring system, characterized in that... include: A multi-sensor module is used to monitor the molten pool image, acoustic signals, and photoelectric signals during the printing process; The signal acquisition module is used to acquire the molten pool image, acoustic signal, and photoelectric signal. The data processing module is used to perform signal preprocessing and extract molten pool features from the molten pool image. The molten pool features are unified in the time domain. Then, typical defect feature parameters are extracted from the acoustic signal, photoelectric signal and the unified molten pool features in the time domain. The defect feature parameters are combined with the corresponding scanning position to generate a defect feature location map. Finally, the defect location is located based on the defect feature location map using a threshold defect judgment method, and the location is displayed and statistically analyzed. The parameter optimization module is used to optimize and adjust the printing parameters when the number of defects exceeds the preset range. The data processing module includes a time-domain unification unit, a defect feature location map acquisition unit, and a defect localization unit; The defect feature location map acquisition unit is used to extract typical defect feature parameters from acoustic signals, photoelectric signals, and time-domain unified melt pool features, respectively; to establish a scanning path grid plane according to the current printing layer shape and laser scanning strategy, where each grid in the scanning path plane represents the corresponding scanning position on the current printing layer; to obtain the minimum and maximum values ​​of the typical defect feature parameters of the acoustic signal, divide the range between the minimum and maximum values ​​into several equidistant intervals, assign a non-zero grayscale value to the typical defect feature parameter value of the acoustic signal located in each interval, fill it into the grid corresponding to the acoustic signal scanning position in the first scanning path grid plane, and generate the defect feature location map of the acoustic signal; The same steps were followed to obtain the defect feature location map of the photoelectric signal and the defect feature location map of the molten pool. The defect location unit is used to compare the gray value corresponding to each grid in the defect feature location map of acoustic signal, photoelectric signal and molten pool feature with a preset threshold range. When the gray value exceeds the preset threshold range, it is determined that there is a defect in the current grid and the gray value of the current grid is updated to 0. The updated defect feature location maps of acoustic signal, photoelectric signal and molten pool feature are merged into a single defect location map. The location with a gray value of 0 on the map is the defect location, and the number of defects is counted.

8. The multi-sensor-based selective laser melting defect monitoring system according to claim 7, characterized in that: The multi-sensor module specifically includes: A high-speed camera, mounted on a rangefinder, is installed outside the chamber of the selective laser printer to capture images of the molten pool during the printing process. An acoustic microphone, mounted in a rangefinder configuration inside the printer compartment, is used to monitor acoustic signals generated during the printing process. The photodiode, mounted off-axis in the printer compartment, is used to monitor photoelectric signals generated during the printing process.

Images