Additive Product Production Process Monitoring System and Method Based on Edge Computing

Abstract

The present invention discloses an additive product production process monitoring system and method based on edge computing, which relates to the technical field of electrical digital data processing. The additive product production process monitoring system based on edge computing includes: a data acquisition module, an abnormal hot spot identification module, a time-frequency analysis module, and an edge node optimization module. The present invention identifies abnormal hot spots through the acquired thermal imaging data, and at the same time determines whether to upload to the cloud server based on the identification result of the abnormal hot spot, then performs time-frequency analysis according to the acquired acoustic signal data, and finally determines whether to optimize the edge node according to the acquired working parameters based on the time-frequency analysis result, thereby realizing an improvement in the accuracy of identifying abnormal operating states during the monitoring process of the additive product production process, and solving the problem of low accuracy in the collaborative identification of abnormal operating states and acoustic signal states during the monitoring process of the additive product production process in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical digital data processing, and particularly to an additive product production process monitoring system and method based on edge computing. Background Art

[0002] With the rapid development of additive manufacturing (3D printing) technology and its wide application in fields such as aerospace, medical, automotive, and molds, higher requirements are put forward for the real-time monitoring and quality control of the production process. Traditional production monitoring methods usually rely on a centralized cloud computing architecture, which has problems such as high data transmission latency, insufficient real-time performance, and large bandwidth pressure, and it is difficult to meet the requirements of high-precision and high-real-time monitoring in the additive manufacturing process. Edge computing, as a distributed computing paradigm, provides an effective way to solve the above problems by sinking computing, storage, and network functions to edge nodes close to the data source.

[0003] Existing technologies use big data and artificial intelligence technologies to deeply analyze the collected production data, and at the same time feedback the results of the deep analysis to the edge nodes, realizing comprehensive monitoring of the additive manufacturing process.

[0004] For example, the intelligent monitoring and management method and system for an exterior wall decorative panel production line disclosed in the invention patent announcement with the publication number: CN119167503B includes: before the decorative panel composite process, using UG modeling software to build models of the decorative panel and the composite process equipment to obtain model files, and obtaining reference data thresholds through simulation; during the actual composite process, using three-dimensional imaging analysis technology to process the three-dimensional data set obtained from the glue application and bonding process and the edge overflow glue amount image obtained from the pressurization and curing process, comparing and analyzing the results of the data processing with the reference data thresholds, and displaying the warning information on the human-computer interaction interface; after the composite process is completed, comprehensively detecting the finished decorative panel and establishing a quality inspection database to record the inspection results.

[0005] For example, a production supply chain monitoring and management system and method based on industrial Internet disclosed in the invention patent announcement with the publication number: CN116306325B includes: real-time monitoring of receiving information, determining a customer distribution map according to the receiving information; performing in-map clustering on the customer distribution maps for different time periods and sorting according to the time period to obtain a supply layer group; obtaining a product supply chain containing time information, and establishing a mapping model according to the supply layer group and the product supply chain; receiving the supply chain change information input by the user, and inputting the supply chain change information into the mapping model to obtain a theoretical distribution layer.

[0006] However, in the process of implementing the technical solutions of the present invention in the embodiments of the present application, it is found that the above technologies have at least the following technical problems:

[0007] In the prior art, there may be a lack of an adaptive adjustment mechanism for dynamic changes in the additive manufacturing process, resulting in the inability to flexibly adjust the recognition strategy when facing different process parameters and environmental conditions, reducing the accuracy of abnormal state recognition, and there is a problem of low accuracy in the collaborative recognition of abnormal operating states and acoustic signal states during the monitoring process of the additive product production process. Summary of the Invention

[0008] By providing a monitoring system and method for the additive product production process based on edge computing, the embodiments of the present application solve the problem of low accuracy in the collaborative recognition of abnormal operating states and acoustic signal states during the monitoring process of the additive product production process in the prior art, and achieve an improvement in the accuracy of abnormal operating state recognition during the monitoring process of the additive product production process.

[0009] The embodiments of the present application provide a monitoring system for the additive product production process based on edge computing, including: a data acquisition module, an abnormal hot spot recognition module, a time-frequency analysis module, and an edge node optimization module; wherein, the data acquisition module is used to collect multi-modal data of a specified additive product in a target monitoring area in real time at the end of the additive manufacturing period, and the multi-modal data is collected in real time through an edge computing node with sensors and actuators deployed in the target monitoring area, and the multi-modal data includes thermal imaging data and acoustic signal data; the abnormal hot spot recognition module is used to perform abnormal hot spot recognition based on the acquired thermal imaging data to obtain an abnormal hot spot recognition result, and the abnormal hot spot recognition is used to analyze the temperature change situation of the specified additive product during the additive manufacturing process; the time-frequency analysis module is used to determine whether to upload to the cloud server based on the abnormal hot spot recognition result, if so, perform time-frequency analysis based on the acquired acoustic signal data to obtain a time-frequency analysis result, otherwise perform step size optimization, and the step size optimization means to improve the heat dissipation capacity during the 3D printing process by adjusting the step size of the 3D printing path, and the time-frequency analysis is used to analyze the operating state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process; the edge node optimization module is used to determine whether to send it to the edge node based on the time-frequency analysis result, if so, perform edge node optimization based on the acquired working parameters, otherwise continue to monitor the production process of the specified additive product in the next additive manufacturing period, and the working parameters are used to reflect the fault repair state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process, and the edge node optimization means to improve the load balance of the edge node by adjusting the memory allocation and the number of threads.

[0010] The embodiment of the present application provides a method for monitoring the additive product production process based on edge computing, including the following steps: Step 1, collect in real time multi-modal data of a specified additive product in the target monitoring area at the end of the additive manufacturing period. The multi-modal data is collected in real time by an edge computing node with sensors and actuators deployed in the target monitoring area. The multi-modal data includes thermal imaging data and acoustic signal data; Step 2, perform abnormal hot spot identification based on the obtained thermal imaging data to obtain an abnormal hot spot identification result. The abnormal hot spot identification is used to analyze the temperature change of the specified additive product during the additive manufacturing process; Step 3, judge whether to upload to the cloud server based on the abnormal hot spot identification result. If so, perform time-frequency analysis on the obtained acoustic signal data to obtain a time-frequency analysis result. Otherwise, perform step size optimization. The step size optimization means adjusting the step size of the 3D printing path to improve the heat dissipation capacity during the 3D printing process. The time-frequency analysis is used to analyze the operating state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process; Step 4, judge whether to send it to the edge node based on the time-frequency analysis result. If so, perform edge node optimization based on the obtained working parameters. Otherwise, continue to monitor the production process of the specified additive product in the next additive manufacturing period. The working parameters are used to reflect the fault repair status of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process. The edge node optimization means adjusting the memory allocation and the number of threads to improve the load balancing of the edge node.

[0011] One or more technical solutions provided in the embodiment of the present application have at least the following technical effects or advantages:

[0012] 1. By performing abnormal hot spot identification on the obtained thermal imaging data, and then judging whether to upload to the cloud server based on the abnormal hot spot identification result. If so, perform time-frequency analysis on the obtained acoustic signal data. Otherwise, perform step size optimization. Finally, judge whether to perform edge node optimization based on the obtained working parameters according to the time-frequency analysis result, thus realizing the accurate identification and efficient processing of abnormal operating states in the production process of specified additive products, better adapting to the complex changes in the additive product production process, and effectively solving the problem of low accuracy in the collaborative identification of abnormal operating states and acoustic signal states during the monitoring of the additive product production process in the prior art.

[0013] 2. Compensate for the difference between the obtained molten pool coverage area and the set value of the molten pool coverage area in the database through the molten pool coverage area compensation factor to obtain the molten pool coverage area index. At the same time, couple the obtained molten pool coverage area index, molten pool average temperature index, and thermal gradient value index to obtain the abnormal hot spot determination value, thereby improving the accuracy of obtaining the abnormal hot spot determination value. This step effectively eliminates measurement errors caused by equipment aging, material differences, or environmental factors, thus effectively improving the accuracy and reliability of abnormal hot spot determination.

[0014] 3. By obtaining the difference between the energy frequency peak value and the maximum allowable energy frequency peak value in the database, and performing compensation operations in combination with the introduced energy frequency peak value compensation factor to obtain the energy frequency peak value index. At the same time, couple the result of the inverse proportion processing of the total energy index of the time-frequency diagram obtained, the obtained energy frequency peak value index, and the abnormal frequency proportion index to obtain the time-frequency analysis determination value, thereby improving the accuracy of obtaining the time-frequency analysis determination value. This process effectively eliminates measurement errors caused by factors such as equipment performance fluctuations, environmental noise interference, or signal transmission losses.

[0015] 4. Through the first edge node optimization and the second edge node optimization, the system can dynamically adjust the memory allocation and the number of threads according to the real-time obtained working parameters, ensuring that the edge node can operate efficiently when processing the data of the additive product production process, reducing problems such as delays or insufficient processing capabilities caused by improper resource allocation, thereby improving the accuracy of abnormal state recognition, enhancing the stability and reliability of the system, and providing strong support for the intelligent upgrade of the additive manufacturing industry. Brief Description of the Drawings

[0016] Figure 1 It is a schematic structural diagram of the additive product production process monitoring system based on edge computing provided by an embodiment of the present application;

[0017] Figure 2 It is a monitoring flow chart of the specified additive product production process provided by an embodiment of the present application. Detailed Embodiments

[0018] Embodiments of the present application provide a monitoring system and method for the additive manufacturing process of additive products based on edge computing, which solves the problem of low accuracy in the collaborative recognition of abnormal operating states and acoustic signal states during the monitoring of the additive manufacturing process of additive products. The data acquisition module collects multi-modal data of a specified additive product in the target monitoring area at the end of the additive manufacturing period in real time. At the same time, the abnormal hot spot recognition module performs abnormal hot spot recognition based on the acquired thermal imaging data to obtain an abnormal hot spot recognition result. Then, the time-frequency analysis module determines whether to upload it to the cloud server based on the abnormal hot spot recognition result. If so, time-frequency analysis is performed on the acquired acoustic signal data to obtain a time-frequency analysis result. Otherwise, step size optimization is performed. Finally, the edge node optimization module determines whether to send it to the edge node based on the time-frequency analysis result. If so, edge node optimization is performed based on the acquired working parameters. Otherwise, the production process of the specified additive product in the next additive manufacturing period is continuously monitored, achieving an improvement in the accuracy of identifying abnormal operating states during the monitoring of the additive manufacturing process of additive products.

[0019] The technical solution in the embodiments of the present application aims to solve the problem of low accuracy in the collaborative recognition of abnormal operating states and acoustic signal states during the monitoring of the additive manufacturing process of additive products. The general idea is as follows:

[0020] Abnormal hot spot recognition is performed based on the acquired thermal imaging data, and then it is determined whether to upload it to the cloud server based on the abnormal hot spot recognition result. If so, time-frequency analysis is performed on the acquired acoustic signal data. Otherwise, step size optimization is performed. Finally, it is determined whether to perform edge node optimization based on the acquired working parameters based on the time-frequency analysis result, achieving the effect of improving the accuracy of identifying abnormal operating states during the monitoring of the additive manufacturing process of additive products.

[0021] To better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0022] As Figure 1 shown, it is a schematic structural diagram of a monitoring system for the additive manufacturing process of additive products based on edge computing provided by an embodiment of the present application. The monitoring system for the additive manufacturing process of additive products based on edge computing provided by an embodiment of the present application includes: a data acquisition module, an abnormal hot spot recognition module, a time-frequency analysis module, and an edge node optimization module.

[0023] Among them, the data acquisition module is used to collect in real time the multi-modal data of the specified additive product in the target monitoring area at the end of the additive manufacturing period. The multi-modal data includes thermal imaging data and acoustic signal data, which are collected in real time through the edge computing nodes deployed in the target monitoring area. The thermal imaging data includes the molten pool coverage area, the average molten pool temperature, and the thermal gradient value, which are used to reflect the temperature change of the specified additive product in the target monitoring area during the additive manufacturing process. The acoustic signal data includes the peak energy frequency, the total energy of the time-frequency diagram, and the proportion of abnormal frequencies, which are used to reflect the ultrasonic change of the specified additive product in the target monitoring area during the 3D printing process. The peak energy frequency is used to quantify the energy concentration degree of the acoustic signal corresponding to the specified additive product in the target monitoring area during the 3D printing period, such as the energy intensity at a specific frequency (set by the preset personnel according to the actual 3D printing scenario), which reflects whether there is abnormal vibration or resonance during the 3D printing process. The total energy of the time-frequency diagram is used to quantify the overall energy distribution of the acoustic signal corresponding to the specified additive product (such as molten polymer material, polylactide) in the target monitoring area during the 3D printing period, such as the sum of the energies of all frequency components in the time-frequency diagram, which reflects the overall acoustic characteristics of the 3D printing process. The proportion of abnormal frequencies is used to quantify the proportion of abnormal frequency components in the acoustic signal corresponding to the specified additive product in the target monitoring area during the 3D printing period, such as the proportion of frequency components exceeding the frequency allowable range, which reflects whether there are defects or faults during the 3D printing process.

[0024] The abnormal hot spot identification module is used to identify abnormal hot spots based on the acquired thermal imaging data to obtain the abnormal hot spot identification result. The abnormal hot spot identification is used to analyze the temperature change of the specified additive product during the additive manufacturing process.

[0025] The time-frequency analysis module is used to judge whether to upload to the cloud server based on the abnormal hot spot identification result. If so, it performs time-frequency analysis based on the acquired acoustic signal data to obtain the time-frequency analysis result. Otherwise, it performs step size optimization. The step size optimization means adjusting the step size of the 3D printing path to improve the heat dissipation capacity during the 3D printing process. The time-frequency analysis is used to analyze the operating state of the additive manufacturing equipment corresponding to the specified additive product during the additive manufacturing process.

[0026] The edge node optimization module is used to judge whether to send it to the edge node based on the time-frequency analysis result. If so, it optimizes the edge node according to the acquired working parameters. Otherwise, it continues to monitor the production process of the specified additive product in the next additive manufacturing period. The working parameters are used to reflect the fault repair state of the additive manufacturing equipment corresponding to the specified additive product during the additive manufacturing process. The edge node optimization means adjusting the memory allocation and the number of threads to improve the load balance of the edge node.

[0027] Specifically, the edge computing node is equipped with a variety of sensors, including but not limited to vision sensors, infrared thermal imagers, acceleration sensors, and time sensors. Among them, the vision sensor is used to monitor the area covered by the molten pool, the infrared thermal imager is used to monitor the average temperature and thermal gradient value of the molten pool, the acceleration sensor is used to monitor the peak energy frequency, the total energy of the time-frequency diagram, and the proportion of abnormal frequencies, and the time sensor is used to monitor the average failure working duration and the average failure repair duration.

[0028] Before designing the additive product production process monitoring system based on edge computing, a database for storing various setting data is established. The database includes but is not limited to preset abnormal hot spot determination values, preset time-frequency analysis determination values, additive manufacturing time periods, and 3D printing time periods. All kinds of values therein are directly set by technicians. Among them, the setting basis of the preset abnormal hot spot determination value can be determined according to the actual production scenario of the additive product. For example, the preset abnormal hot spot determination value is represented by the result of summing and averaging the historical abnormal hot spot determination values of the specified additive product in the target monitoring area in the database at the end of the historical additive manufacturing time period. In addition, all kinds of values in the database can be set and fine-tuned by technicians according to actual debugging.

[0029] In this embodiment, the edge computing node is directly deployed in the target monitoring area without transmitting data to the cloud or remote server, thus significantly reducing the latency of multi-modal data collection. Secondly, the edge computing node can simultaneously collect and process thermal imaging data and acoustic signal data, realize the real-time fusion and preliminary analysis of multi-modal data, and form a distributed monitoring network, which helps to ensure the continuity of the system, improve the accuracy and reliability of anomaly detection. At the same time, combined with the advantages of accurate identification of abnormal hot spots, optimization of decision-making based on time-frequency analysis, improvement of step size and heat dissipation capacity, and improvement of the load balance of edge nodes, etc., the intelligent monitoring and management of the additive product production process are realized, providing strong support for the intelligent upgrading of the additive manufacturing industry.

[0030] Further, the specific process for obtaining thermal imaging data is as follows: E1, obtain the area of the molten pool coverage region of the specified additive product within the target monitoring area at the end of the additive manufacturing period. If the obtained area of the molten pool coverage region is greater than the set value of the molten pool coverage region area in the database and does not exceed the maximum allowable molten pool coverage region area in the database, it indicates that the acquisition of the molten pool coverage region is effective and E2 is executed. Otherwise, the deviation of the obtained area of the molten pool coverage region is input into the spot size algorithm to output the increased amplitude of the spot size until the obtained area of the molten pool coverage region is greater than the set value of the molten pool coverage region area in the database and then E2 is executed. The area of the molten pool coverage region is used to quantify the fusion effect of the specified additive product within the target monitoring area during the additive manufacturing period. The set value of the molten pool coverage region area is represented by the result of summing and averaging the historical molten pool coverage region areas of the specified additive product within the target monitoring area at the end of the historical additive manufacturing periods in the database. The maximum allowable molten pool coverage region area is represented by the result of summing and averaging the maximum values of the historical molten pool coverage region areas of the specified additive product within the target monitoring area at the end of each historical additive manufacturing period in the database; E2, obtain the average temperature of the molten pool (not exceeding the set value of the average temperature of the molten pool, such as +100°C) and the thermal gradient value (not exceeding the set value of the thermal gradient value, such as 5°C / mm) of the specified additive product within the molten pool coverage region at the end of the additive manufacturing period. The average temperature of the molten pool is used to quantify the stability of the energy input of the specified additive product within the target monitoring area during the additive manufacturing period, and the thermal gradient value is used to quantify the thermal stress distribution of the specified additive product within the target monitoring area during the additive manufacturing period.

[0031] In this embodiment, through the accurate acquisition and judgment of the area of the molten pool coverage region, the fusion effect of the specified additive product during the additive manufacturing process can be accurately evaluated. Based on the quantitative analysis of the average temperature of the molten pool, the stability of the energy input during additive manufacturing can be understood in real time. The acquisition and quantification of the thermal gradient value enable the operator to clearly master the thermal stress distribution during the additive manufacturing process. This acquisition process can ensure the unified and accurate evaluation of the fusion effect, energy input stability, and thermal stress distribution of the additive product under different batches and different production conditions, which helps to improve the consistency and reliability of the product and meet the high requirements of customers for product quality.

[0032] Further, the specific process for identifying abnormal hot spots based on the obtained thermal imaging data is as follows: First, compensate for the degree of difference between the obtained area of the molten pool coverage region and the set value of the molten pool coverage region area in the database through the molten pool coverage region area compensation factor to obtain the molten pool coverage region area index. The molten pool coverage region area index has the following specific limiting expression: , where Represents the molten pool coverage area index of a specified additive product in the target monitoring area at the end of the additive manufacturing period. Represents the molten pool coverage area compensation factor. Represents the molten pool coverage area of a specified additive product in the target monitoring area at the end of the additive manufacturing period. Represents the maximum allowable molten pool coverage area. Represents the set value of the molten pool coverage area. The units of the molten pool coverage area, the maximum allowable molten pool coverage area, and the set value of the molten pool coverage area are the same, all in square millimeters (mm²).

[0033] Then, compensate for the degree of difference between the obtained average molten pool temperature and the set value of the average molten pool temperature in the database through the average molten pool temperature compensation factor to obtain the average molten pool temperature index. The average molten pool temperature index The specific limiting expression is: , where Represents the average molten pool temperature of a specified additive product in the molten pool coverage area of the target monitoring area at the end of the additive manufacturing period. Represents the average molten pool temperature compensation factor. Represents the average molten pool temperature of a specified additive product in the target monitoring area at the end of the additive manufacturing period. Represents the set value of the average molten pool temperature. The units of the average molten pool temperature and the set value of the average molten pool temperature are the same, both in degrees Celsius (°C). The set value of the average molten pool temperature is represented by the result of summing and averaging the historical average molten pool temperatures of the specified additive product in the target monitoring area at the end of the historical additive manufacturing periods in the database.

[0034] Next, compensate for the degree of difference between the obtained heat gradient value and the set value of the heat gradient value in the database through the heat gradient value compensation factor to obtain the heat gradient value index. The heat gradient value index The specific limiting expression is: , where Represents the heat gradient value index of a specified additive product in the molten pool coverage area of the target monitoring area at the end of the additive manufacturing period. Represents the heat gradient value compensation factor. Represents the heat gradient value of a specified additive product in the target monitoring area at the end of the additive manufacturing period. Represents the set value of the heat gradient value. The units of the heat gradient value and the set value of the heat gradient value are the same, both in degrees Celsius per millimeter (°C / mm). The set value of the heat gradient value is represented by the result of summing and averaging the historical heat gradient values of the specified additive product in the target monitoring area at the end of the historical additive manufacturing periods in the database.

[0035] Finally, the obtained melt pool coverage area index, melt pool average temperature index and thermal gradient value index are coupled to obtain the abnormal hot spot judgment value, which is used to reflect the abnormal degree of the working temperature of the corresponding additive production equipment in the additive manufacturing process of the specified additive product according to the thermal imaging data. The thermal imaging data includes the melt pool coverage area, the melt pool average temperature and the thermal gradient value. The abnormal hot spot judgment value The specific restriction expression is: , where Represents the abnormal hot spot determination value of the specified additive product in the target monitoring area at the end of the additive manufacturing period.

[0036] The database stores preset compensation factors that are closely related to the abnormal hotspot judgment values. A pre-defined mapping relationship is established between these compensation factors and the corresponding melt pool coverage area, melt pool average temperature, and thermal gradient value. It is worth noting that this mapping is not set arbitrarily. It can be one-to-one or many-to-one. For example, in actual applications, when it is necessary to analyze the temperature changes of a specified additive product during the additive manufacturing process, the real-time melt pool coverage area, melt pool average temperature, and thermal gradient value can be directly input into this preset mapping relationship, so that the melt pool coverage area compensation factor, melt pool average temperature compensation factor, and thermal gradient value compensation factor that match the melt pool coverage area, melt pool average temperature, and thermal gradient value can be quickly and accurately obtained.

[0037] It is particularly important that in order to ensure the consistency and comparability of the evaluation, the value ranges of the melt pool coverage area compensation factor, the melt pool average temperature compensation factor and the thermal gradient value compensation factor in this example are all limited to between 0 and 1, and the sum of the three is 1.

[0038] In this embodiment, the abnormal hotspot judgment value increases with the increase of the area covered by the molten pool, the average temperature of the molten pool and the thermal gradient value. Among them, the increase in the area covered by the molten pool may require higher energy input to maintain the melting state, thereby causing an increase in the average temperature of the molten pool. Conversely, the increase in the average temperature of the molten pool may also promote the expansion of the area covered by the molten pool, because high temperature helps the melting and flow of materials.

[0039] Secondly, the increase in the area covered by the molten pool may lead to more uneven heat distribution, especially in the edge area, where the temperature changes may be more drastic, thereby increasing the thermal gradient value; the increase in the average temperature of the molten pool may lead to changes in heat conduction and heat convection, thereby affecting the distribution of the thermal gradient value.

[0040] By considering the interaction mechanism among the molten pool coverage area, the average temperature of the molten pool, and the thermal gradient value, it helps to more comprehensively evaluate the thermal state during the additive manufacturing process, thereby more accurately identifying the location and cause of abnormal hot spots, optimizing the formation and stability of the molten pool, reducing the generation of abnormal hot spots, improving the accuracy and reliability of monitoring, and effectively solving the problem of low accuracy in the collaborative identification of abnormal operating states and acoustic signal states during the monitoring of the additive product production process in the prior art.

[0041] Furthermore, based on the abnormal hot spot identification result, it is determined whether to upload it to the cloud server. The specific process is as follows: If the obtained abnormal hot spot determination value is not greater than the preset abnormal hot spot determination value in the database, the abnormal hot spot identification result is recorded as no abnormal temperature component in the molten pool coverage area, and at the same time, the corresponding thermal imaging data is uploaded to the cloud server. Otherwise, the abnormal hot spot identification result is recorded as having an abnormal temperature component in the molten pool coverage area and the step size is adjusted.

[0042] Among them, the specific process of step size adjustment is as follows: The obtained hot spot abnormal determination value deviation and step size deviation are input into the step size PID (Proportional-Integral-Derivative) control algorithm of the additive manufacturing equipment to output the actual step size adjustment amplitude. The hot spot abnormal determination value deviation is used to quantify the difference degree between the obtained abnormal hot spot determination value and the preset abnormal hot spot determination value, that is, the difference between the obtained abnormal hot spot determination value and the preset abnormal hot spot determination value. The step size deviation is used to quantify the difference degree between the actual step size and the target step size of the corresponding 3D printing path at the end of the additive manufacturing period, that is, the absolute value of the difference between the target step size and the actual step size. When the newly obtained hot spot abnormal determination value after step size adjustment is not greater than the preset hot spot abnormal determination value in the database and the adjustment times do not exceed the preset times, the step size adjustment is completed and the newly obtained thermal imaging data is uploaded to the cloud server. Otherwise, a shutdown instruction is sent and the preset personnel are prompted to perform maintenance.

[0043] In this embodiment, through the precise comparison and quantitative analysis of the abnormal hot spot determination value, it is possible to timely and accurately identify the abnormal temperature components in the molten pool coverage area, avoiding quality problems of additive manufacturing products caused by temperature abnormalities, such as internal defects of materials and insufficient structural strength. At the same time, using the step size PID control algorithm, the step size is dynamically adjusted according to the hot spot abnormal determination value deviation and step size deviation, making the step size in the additive manufacturing process more accurate and meeting the requirements of the target step size, further improving the dimensional accuracy and shape accuracy of the product. When the step size adjustment cannot meet the requirements, a shutdown instruction is sent in a timely manner and the preset personnel are prompted to perform maintenance, effectively avoiding the expansion of faults and equipment damage that may be caused by the equipment continuing to operate in an abnormal state, and ensuring the safe operation of the equipment.

[0044] Further, perform time-frequency analysis on the acquired acoustic signal data. The specific process is as follows: First, obtain the difference degree between the peak energy frequency and the maximum allowable peak energy frequency in the database, and perform compensation operations in combination with the introduced peak energy frequency compensation factor to obtain the peak energy frequency index. The specific limit expression of the peak energy frequency index is: , where represents the peak energy frequency index of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the 3D printing period, represents the peak energy frequency compensation factor, represents the peak energy frequency of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the 3D printing period, represents the maximum allowable peak energy frequency. The peak energy frequency and the maximum allowable peak energy frequency have the same unit, both in joules per hertz (J / Hz). The maximum allowable peak energy frequency is represented by the result of summing and averaging the maximum values of the historical peak energy frequencies of the specified additive product in the molten pool coverage area of the target monitoring area in each historical 3D printing period in the database.

[0045] Then, obtain the difference degree between the total energy of the time-frequency diagram and the total energy of the reference time-frequency diagram in the database, and perform compensation operations in combination with the introduced total energy compensation factor of the time-frequency diagram to obtain the total energy index of the time-frequency diagram. The specific limit expression of the total energy index of the time-frequency diagram is: , where represents the total energy index of the time-frequency diagram of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the 3D printing period, represents the total energy compensation factor of the time-frequency diagram, represents the total energy of the time-frequency diagram of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the 3D printing period, represents the total energy of the reference time-frequency diagram. The total energy of the time-frequency diagram and the total energy of the reference time-frequency diagram have the same unit, both in joules (J). The total energy of the reference time-frequency diagram is represented by the result of summing and averaging the historical total energies of the time-frequency diagrams of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the historical 3D printing periods in the database.

[0046] Next, obtain the difference degree between the abnormal frequency proportion and the maximum allowable abnormal frequency proportion in the database, and perform compensation operations in combination with the introduced abnormal frequency proportion compensation factor to obtain the abnormal frequency proportion index. The specific limit expression of the abnormal frequency proportion index is: , where It indicates the abnormal frequency ratio index of the specified additive product in the molten pool coverage area of the target monitoring area at the end of the 3D printing period. Indicates the abnormal frequency ratio compensation factor, It indicates the abnormal frequency ratio of the specified additive product in the molten pool coverage area in the target monitoring area at the end of the 3D printing period. It represents the maximum allowable abnormal frequency ratio. The unit of abnormal frequency ratio and maximum allowable abnormal frequency ratio is the same, both are percentage (%). The maximum allowable abnormal frequency ratio is represented by the maximum value of the historical abnormal frequency ratio of the specified additive product in the molten pool coverage area in the target monitoring area in the database at the end of each historical 3D printing period.

[0047] Finally, the result of inverse proportional processing of the total energy index of the time-frequency graph is coupled with the energy frequency peak index and the abnormal frequency ratio index to obtain the time-frequency analysis judgment value. The time-frequency analysis judgment value is used to reflect the abnormal degree of the acoustic wave signal data corresponding to the acoustic wave signal of the specified additive product in the 3D printing process. The specific restriction expression is: , where It represents the time-frequency analysis judgment value of the specified additive product in the molten pool coverage area in the target monitoring area at the end of the 3D printing period.

[0048] The database stores preset compensation factors that are closely related to the time-frequency analysis judgment values. A predefined mapping relationship is established between these compensation factors and the corresponding energy frequency peak values, total energy of the time-frequency graph and abnormal frequency ratio. It is worth noting that this mapping is not set arbitrarily. It can be one-to-one or many-to-one. For example, in practical applications, when it is necessary to analyze the abnormal conditions of the acoustic wave signal of a specified additive product during the 3D printing process, the energy frequency peak value, total energy of the time-frequency graph and abnormal frequency ratio obtained in real time can be directly input into this preset mapping relationship, so that the energy frequency peak compensation factor, total energy compensation factor of the time-frequency graph and abnormal frequency ratio that match the energy frequency peak value, total energy of the time-frequency graph and abnormal frequency ratio can be quickly and accurately obtained.

[0049] It is particularly important that in order to ensure the consistency and comparability of the evaluation, the value ranges of the energy frequency peak compensation factor, the total energy compensation factor of the time-frequency graph, and the abnormal frequency ratio compensation factor in this example are all limited to between 0 and 1, and the sum of the three is 1.

[0050] In this embodiment, the time-frequency analysis determination value increases with the increase of the peak energy frequency, the total energy of the time-frequency diagram, and the abnormal frequency ratio. Among them, when the total energy of the time-frequency diagram increases, it may be caused by an increase in the normal workload, and there is no obvious change in the peak energy frequency and the abnormal frequency ratio. Then, it can be judged that the node is operating normally. On the contrary, if all three show abnormal changes at the same time, it is more likely that there is a problem with the node.

[0051] When the total energy of the time-frequency diagram increases, the threshold of the abnormal frequency ratio can be appropriately increased to avoid misjudgment caused by the increase in the total energy. When it is found that the peak energy frequency and the abnormal frequency ratio increase at the same time, the possible location of the fault can be further determined. If the peak energy frequency continues to increase and the abnormal frequency ratio gradually rises, then it can be predicted that the fault may deteriorate further, and measures can be taken in advance for maintenance and repair.

[0052] According to the analysis of the mutual influence mechanism, the increase in the peak energy frequency causes the total energy of the time-frequency diagram to rise, and since the energy is concentrated in the abnormal frequency range, the abnormal frequency ratio increases significantly. Combining the changes in these indicators, it is judged that there may be a hardware fault or software anomaly in the edge computing node. After further inspection, it is found that the temperature of a certain chip of the node is too high, resulting in abnormal signal processing, thus verifying the accuracy of identifying the abnormal operating state by considering the mutual influence mechanism, avoiding the further expansion of the node fault, and effectively solving the problem of low accuracy in the collaborative identification of the abnormal operating state and the acoustic signal state in the monitoring process of the additive product production process in the prior art.

[0053] Further, it is determined whether to send it to the edge node based on the time-frequency analysis result. The specific process is as follows: If the obtained time-frequency analysis determination value is not greater than the preset time-frequency analysis determination value in the database, the time-frequency analysis result is recorded as no abnormal frequency components in the molten pool coverage area, and a production process monitoring instruction for the next additive manufacturing period is sent. Otherwise, the time-frequency analysis identification result is recorded as having abnormal frequency components in the molten pool coverage area, and the FFT (Fast Fourier Transform) point number is adjusted; the preset time-frequency analysis determination value is represented by the result of summing and averaging the historical time-frequency analysis determination values of the specified additive product in the molten pool coverage area of the target monitoring area in the database at the end of the historical 3D printing period; the FFT point number adjustment means optimizing the balance relationship between time and frequency resolution by the obtained FFT point number adjustment amplitude; the FFT point number adjustment amplitude is the result obtained by inputting the obtained time-frequency analysis determination value deviation and frequency resolution deviation into the LMS (Least Mean Square) algorithm; the time-frequency analysis determination value deviation is used to quantify the degree of difference between the obtained time-frequency analysis determination value and the preset time-frequency analysis determination value, that is, the difference between the obtained time-frequency analysis determination value and the preset time-frequency analysis determination value; the frequency resolution deviation is used to quantify the degree of difference between the actual frequency resolution and the target frequency resolution of the adaptive adjustment filter corresponding to the time-frequency diagram of the molten pool coverage area at the end of the 3D printing period, that is, the absolute value of the difference between the actual frequency resolution and the target frequency resolution.

[0054] In this embodiment, by adjusting the FFT point number, the balance relationship between time and frequency resolution is optimized. During the additive manufacturing process, different application scenarios have different requirements for time resolution and frequency resolution. For example, for rapidly changing signals, higher time resolution is required; for signals with rich frequency components, higher frequency resolution is required. By adjusting the FFT point number, different analysis requirements can be met according to the actual situation, realizing real-time monitoring and feedback of the production process, and improving the intelligent level of additive product production management.

[0055] Further, the edge node optimization includes the first edge node optimization and the second edge node optimization; the first edge node optimization refers to the optimization corresponding to when the obtained harmonic mean of the working parameters is greater than the maximum value of the historical harmonic mean of the working parameters in the database. The maximum value of the historical harmonic mean of the working parameters represents the maximum value of the historical harmonic mean of the working parameters of the corresponding edge node in the target monitoring area in the database during the historical first edge node optimization process.

[0056] Specifically, the specific process of optimizing the first edge node is as follows: F11, input the obtained working parameter deviation and working parameter average value into the memory allocation adjustment algorithm to output the memory allocation increase amplitude. After one memory allocation increase, determine whether the corresponding decrease amplitude of the obtained first working parameter harmonic average deviation is greater than the preset decrease amplitude in the database (set by the preset personnel according to the actual situation). The working parameter harmonic average represents the result of performing harmonic averaging on the obtained working parameters, and the working parameters include the average failure working duration and the average failure repair duration; F12, if the corresponding decrease amplitude of the obtained working parameter harmonic average deviation is greater than the preset decrease amplitude in the database, it indicates that the memory allocation increase is effective and continue to increase the memory allocation by the preset amplitude until the obtained working parameter harmonic average is within the allowable range of the working parameter harmonic average in the database; F13, if the corresponding decrease amplitude of the obtained working parameter harmonic average deviation is not greater than the preset decrease amplitude in the database, input the working parameter deviation and working parameter average value obtained after one memory allocation increase into the thread number adjustment algorithm to output the thread number decrease amplitude and return to F11. The allowable range of the working parameter harmonic average represents the range corresponding to the maximum and minimum values of the historical working parameter harmonic average of the corresponding edge node in the target monitoring area in the database during the historical second edge node optimization process, usually including the cases equal to the maximum and minimum values of the historical working parameter harmonic average. The working parameter deviation represents the absolute value of the difference between the obtained working parameter harmonic average and the working parameter average value, and the working parameter average value represents the result of summing and averaging the obtained average failure working duration and average repair duration. The first working parameter harmonic average deviation represents the difference between the obtained working parameter harmonic average and the maximum value of the historical working parameter harmonic average.

[0057] The optimization of the second edge node represents the optimization corresponding to the case where the obtained working parameter harmonic average is less than the minimum value of the historical working parameter harmonic average in the database. The minimum value of the historical working parameter harmonic average represents the minimum value of the historical working parameter harmonic average of the corresponding edge node in the target monitoring area in the database during the historical second edge node optimization process.

[0058] Specifically, the specific process of optimizing the second edge node is as follows: F21, input the obtained working parameter deviation and working parameter average value into the memory allocation adjustment algorithm to output the memory allocation reduction amplitude. After a memory allocation reduction, determine whether the reduction amplitude corresponding to the obtained harmonic mean deviation of the second working parameter is greater than the preset reduction amplitude in the database. The harmonic mean deviation of the second working parameter represents the difference between the minimum value of the historical working parameter harmonic mean and the obtained harmonic mean of the working parameters; F22, if the reduction amplitude corresponding to the obtained harmonic mean deviation of the working parameters is greater than the preset reduction amplitude in the database, it indicates that the memory allocation reduction is effective and continue to reduce the memory allocation by the preset amplitude until the obtained harmonic mean of the working parameters is within the allowable range of the harmonic mean of the working parameters in the database; F23, if the reduction amplitude corresponding to the obtained harmonic mean deviation of the working parameters is not greater than the preset reduction amplitude in the database, input the working parameter deviation and working parameter average value re-obtained after a memory allocation reduction into the thread number adjustment algorithm to output the thread number increase amplitude and return to F21.

[0059] In this embodiment, by dynamically adjusting the memory allocation and the number of threads, resources can be reasonably allocated according to the actual working conditions of the edge nodes, avoiding resource waste or insufficiency. By monitoring and optimizing the working parameters (average failure working duration and average failure repair duration), potential system problems can be timely detected and measures can be taken for adjustment. When new edge nodes need to be added or existing nodes need to be upgraded, reasonable resource allocation and scheduling can be achieved through a unified resource management mechanism, improving the overall scalability of the system.

[0060] Such as Figure 2As shown in the figure, it is a monitoring flowchart of the production process of a specified additive product provided by an embodiment of the present application. The method for monitoring the additive product production process based on edge computing provided by the embodiment of the present application includes the following steps: Step 1, collect in real time the multi-modal data of the specified additive product in the target monitoring area at the end of the additive manufacturing period. The multi-modal data is collected in real time through an edge computing node with sensors and actuators deployed in the target monitoring area. The multi-modal data includes thermal imaging data and acoustic signal data; Step 2, perform abnormal hot spot identification based on the obtained thermal imaging data to obtain the abnormal hot spot identification result. The abnormal hot spot identification is used to analyze the temperature change of the specified additive product during the additive manufacturing process; Step 3, based on the abnormal hot spot identification result, determine whether to upload it to the cloud server. If so, perform time-frequency analysis on the obtained acoustic signal data to obtain the time-frequency analysis result. Otherwise, perform step size optimization. The step size optimization means adjusting the step size of the 3D printing path to improve the heat dissipation capacity during the 3D printing process. The time-frequency analysis is used to analyze the operating state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process; Step 4, based on the time-frequency analysis result, determine whether to send it to the edge node. If so, perform edge node optimization according to the obtained working parameters. Otherwise, continue to monitor the production process of the specified additive product in the next additive manufacturing period. The working parameters are used to reflect the fault repair state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process. The edge node optimization means adjusting the memory allocation and the number of threads to improve the load balancing of the edge node.

[0061] At the beginning of the process, the multi-modal data is collected in real time through the data acquisition module, and the thermal imaging data is processed by the abnormal hot spot identification module to detect whether there are abnormal hot spots. If abnormal hot spots are detected, the process enters the time-frequency analysis module to analyze the acoustic signals; if no abnormal hot spots are detected, step size optimization is performed, and then the data in the next period is continuously monitored. After being processed by the time-frequency analysis module, it is judged whether to upload the data to the cloud. If uploading is selected, it enters the edge node optimization module for optimization processing; if uploading is not selected, step size optimization is performed, and the next period is continuously monitored. After being processed by the edge node optimization module, the process ends.

[0062] The design logic of the entire flowchart is to realize the intelligent monitoring and optimization of the system through the collection and analysis of multi-modal data, combined with anomaly detection and signal processing technologies.

[0063] In this embodiment, through steps such as real-time collection of multi-modal data, abnormal hot spot identification and time-frequency analysis, as well as step size optimization and edge node optimization, the comprehensive monitoring and optimization of the additive product production process are realized. Compared with the prior art, this method has higher monitoring accuracy, system stability, data processing efficiency, intelligent production level and lower operation and maintenance costs.

[0064] In summary, the embodiments of the present application identify abnormal hotspots through the acquired thermal imaging data, and then determine whether to upload them to the cloud server based on the identification results of the abnormal hotspots. If so, time-frequency analysis is performed according to the acquired acoustic signal data; otherwise, step size optimization is carried out. Finally, it is determined whether to perform edge node optimization according to the acquired working parameters based on the time-frequency analysis results, thereby realizing the accurate identification and efficient processing of abnormal operating states in the production process of specified additive products, better adapting to the complex changes in the additive product production process, and effectively solving the problem of low accuracy in the collaborative identification of abnormal operating states and acoustic signal states during the monitoring of the additive product production process in the prior art.

[0065] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] The present invention is described with reference to the flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0067] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Therefore, the instructions executed on the computer or other programmable device provide for implementing the specified functions in Figure 1steps of one or more processes and / or boxes Figure 1 steps of functions specified in one or more boxes

[0069] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0070] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. An additive product production process monitoring system based on edge computing, characterized in that, Including: A data acquisition module, an abnormal hot spot identification module, a time-frequency analysis module, and an edge node optimization module; Among them, the data acquisition module is used to collect multi-modal data of a specified additive product in a target monitoring area in real time at the end of the additive manufacturing period. The multi-modal data is collected in real time by an edge computing node with sensors and actuators deployed in the target monitoring area. The multi-modal data includes thermal imaging data and acoustic signal data; The abnormal hot spot identification module is used to identify abnormal hot spots based on the acquired thermal imaging data to obtain an abnormal hot spot identification result. The abnormal hot spot identification is used to analyze the temperature change of the specified additive product during the additive manufacturing process; The time-frequency analysis module is used to determine whether to upload to the cloud server based on the abnormal hot spot identification result. If so, time-frequency analysis is performed on the acquired acoustic signal data to obtain a time-frequency analysis result. Otherwise, step size optimization is performed. The step size optimization means adjusting the step size of the 3D printing path to improve the heat dissipation capacity during the 3D printing process. The time-frequency analysis is used to analyze the operating state of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process; The edge node optimization module is used to determine whether to send it to the edge node based on the time-frequency analysis result. If so, edge node optimization is performed according to the acquired working parameters. Otherwise, continue to monitor the production process of the specified additive product in the next additive manufacturing period. The working parameters are used to reflect the fault repair status of the corresponding additive manufacturing equipment of the specified additive product during the additive manufacturing process. The edge node optimization means adjusting the memory allocation and the number of threads to improve the load balance of the edge node; The process of determining whether to upload to the cloud server based on the abnormal hot spot identification result is as follows: If the obtained abnormal hot spot determination value is not greater than the preset abnormal hot spot determination value in the database, the abnormal hot spot identification result is recorded as no abnormal temperature component in the molten pool coverage area, and at the same time, the corresponding thermal imaging data is uploaded to the cloud server. Otherwise, the abnormal hot spot identification result is recorded as having an abnormal temperature component in the molten pool coverage area and step size adjustment is performed; The specific process of the step size adjustment is as follows: Input the obtained hot spot abnormal determination value deviation and step size deviation into the step size PID control algorithm of the additive manufacturing equipment to output the actual step size adjustment amplitude; When the re-obtained hot spot abnormal determination value after step size adjustment is not greater than the preset hot spot abnormal determination value in the database and the adjustment times do not exceed the preset times, the step size adjustment is completed and the re-obtained thermal imaging data is uploaded to the cloud server. Otherwise, a shutdown instruction is sent and a preset person is prompted to perform maintenance; The hot spot abnormal determination value deviation is used to quantify the difference degree between the obtained abnormal hot spot determination value and the preset abnormal hot spot determination value; The step size deviation is used to quantify the difference degree between the actual step size of the corresponding 3D printing path at the end of the additive manufacturing period and the target step size.

2. The additive product production process monitoring system based on edge computing according to claim 1, characterized in that The thermal imaging data is used to reflect the temperature change of the specified additive product in the target monitoring area during the additive manufacturing process; The acoustic signal data is used to reflect the ultrasonic changes of a specified additive product in the target monitoring area during the 3D printing process. The acoustic signal data includes the peak energy frequency, the total energy of the time-frequency diagram, and the proportion of abnormal frequencies. The peak energy frequency is used to quantify the degree of energy concentration of the corresponding acoustic signal of the specified additive product in the target monitoring area during the 3D printing period. The total energy of the time-frequency diagram is used to quantify the overall energy distribution of the corresponding acoustic signal of the specified additive product in the target monitoring area during the 3D printing period. The proportion of abnormal frequencies is used to quantify the proportion of abnormal frequency components in the corresponding acoustic signal of the specified additive product in the target monitoring area during the 3D printing period.

3. The additive product production process monitoring system based on edge computing according to claim 2, wherein, The specific acquisition process of the thermal imaging data is as follows: E1. Obtain the area of the molten pool coverage area of the specified additive product in the target monitoring area at the end of the additive manufacturing period. If the obtained area of the molten pool coverage area is greater than the set value of the molten pool coverage area in the database, it indicates that the acquisition of the molten pool coverage area is effective and E2 is executed. Otherwise, the deviation of the obtained area of the molten pool coverage area is input into the spot size algorithm to output the increase amplitude of the spot size until the obtained area of the molten pool coverage area is greater than the set value of the molten pool coverage area in the database and then E2 is executed. The area of the molten pool coverage area is used to quantify the fusion effect of the specified additive product in the target monitoring area during the additive manufacturing period. E2. Obtain the average temperature and thermal gradient value of the molten pool of the specified additive product in the molten pool coverage area at the end of the additive manufacturing period. The average temperature of the molten pool is used to quantify the stability of energy input of the specified additive product in the target monitoring area during the additive manufacturing period, and the thermal gradient value is used to quantify the thermal stress distribution of the specified additive product in the target monitoring area during the additive manufacturing period.

4. The additive product production process monitoring system based on edge computing according to claim 3, characterized in that, The specific process of identifying abnormal hot spots based on the obtained thermal imaging data is as follows: Compensate for the difference between the obtained area of the molten pool coverage area and the set value of the molten pool coverage area in the database through the molten pool coverage area compensation factor to obtain the molten pool coverage area index. Compensate for the difference between the obtained average temperature of the molten pool and the set value of the average temperature of the molten pool in the database through the average temperature compensation factor of the molten pool to obtain the average temperature index of the molten pool. Compensate for the difference between the obtained thermal gradient value and the set value of the thermal gradient value in the database through the thermal gradient value compensation factor to obtain the thermal gradient value index. Couple the obtained molten pool coverage area index, average temperature index of the molten pool, and thermal gradient value index to obtain an abnormal hot spot determination value. The abnormal hot spot determination value is used to reflect the degree of abnormal working temperature of the corresponding additive production equipment of the specified additive product during the additive manufacturing process. The thermal imaging data includes the area of the molten pool coverage area, the average temperature of the molten pool, and the thermal gradient value.

5. The additive product production process monitoring system based on edge computing according to claim 2, characterized in that, The specific process of performing time-frequency analysis based on the obtained acoustic signal data is as follows: Obtain the degree of difference between the peak energy frequency and the maximum allowable peak energy frequency in the database, and perform compensation operations in combination with the introduced peak energy frequency compensation factor to obtain the peak energy frequency index. Obtain the degree of difference between the total energy of the time-frequency diagram obtained and the total energy of the reference time-frequency diagram in the database, and perform a compensation operation in combination with the introduced total energy compensation factor of the time-frequency diagram to obtain the total energy index of the time-frequency diagram; Obtain the degree of difference between the abnormal frequency ratio obtained and the maximum allowable abnormal frequency ratio in the database, and perform a compensation operation in combination with the introduced abnormal frequency ratio compensation factor to obtain the abnormal frequency ratio index; Perform a coupling process on the result of the inverse proportional processing of the obtained total energy index of the time-frequency diagram and the obtained energy frequency peak index and abnormal frequency ratio index to obtain a time-frequency analysis determination value, and the time-frequency analysis determination value is used to reflect the abnormal degree of the acoustic signal data corresponding to the specified additive product during 3D printing for the corresponding acoustic signal.

6. The additive product production process monitoring system based on edge computing according to claim 5, wherein, The process of judging whether to send to the edge node based on the time-frequency analysis result is as follows: If the obtained time-frequency analysis determination value is not greater than the preset time-frequency analysis determination value in the database, record the time-frequency analysis result as no abnormal frequency components in the molten pool coverage area and send a production process monitoring instruction for the next additive manufacturing period; otherwise, record the time-frequency analysis recognition result as having abnormal frequency components in the molten pool coverage area and perform an FFT point number adjustment; The FFT point number adjustment means optimizing the balance relationship between time and frequency resolution through the obtained FFT point number adjustment amplitude; The FFT point number adjustment amplitude means the result obtained by inputting the obtained time-frequency analysis determination value deviation and frequency resolution deviation into the LMS algorithm; The time-frequency analysis determination value deviation is used to quantify the degree of difference between the obtained time-frequency analysis determination value and the preset time-frequency analysis determination value; The frequency resolution deviation is used to quantify the degree of difference between the actual frequency resolution and the target frequency resolution of the adaptive adjustment filter corresponding to the time-frequency diagram in the molten pool coverage area at the end of the 3D printing period.

7. The additive product production process monitoring system based on edge computing according to claim 1, wherein The edge node optimization means the first edge node optimization and the second edge node optimization; The first edge node optimization means the optimization corresponding to when the obtained harmonic mean of the working parameters is greater than the maximum value of the historical harmonic mean of the working parameters in the database; The specific process of the first edge node optimization is as follows: F11, input the obtained working parameter deviation and working parameter average value into the memory allocation adjustment algorithm to output the memory allocation increase amplitude. After one memory allocation increase, judge whether the corresponding decrease amplitude of the obtained first working parameter harmonic mean deviation is greater than the preset decrease amplitude in the database; F12, if the corresponding decrease amplitude of the obtained working parameter harmonic mean deviation is greater than the preset decrease amplitude in the database, continue to increase the memory allocation by the preset amplitude until the obtained working parameter harmonic mean is within the allowable range of the working parameter harmonic mean in the database; F13, if the corresponding decrease amplitude of the obtained working parameter harmonic mean deviation is not greater than the preset decrease amplitude in the database, input the working parameter deviation and working parameter average value re-obtained after one memory allocation increase into the thread number adjustment algorithm to output the thread number decrease amplitude and return to F11; The harmonic mean of the working parameters means the result of performing a harmonic mean process on the obtained working parameters; The working parameters include the mean time between failures and the mean time to repair. The deviation of the harmonic mean of the first working parameter represents the difference between the harmonic mean of the obtained working parameters and the maximum value of the harmonic mean of the historical working parameters.

8. The additive product production process monitoring system based on edge computing according to claim 7, wherein, The optimization of the second edge node represents the optimization corresponding to the situation where the harmonic mean of the obtained working parameters is less than the minimum value of the harmonic mean of the historical working parameters in the database. The specific process of the optimization of the second edge node is as follows: F21, input the obtained working parameter deviation and the working parameter average value into the memory allocation adjustment algorithm to output the reduction amplitude of the memory allocation. After one reduction of the memory allocation, determine whether the reduction amplitude corresponding to the obtained harmonic mean deviation of the second working parameter is greater than the preset reduction amplitude in the database. F22, if the reduction amplitude corresponding to the obtained harmonic mean deviation of the working parameters is greater than the preset reduction amplitude in the database, continue to reduce the memory allocation by the preset amplitude until the obtained harmonic mean of the working parameters is within the allowable range of the harmonic mean of the working parameters in the database. F23, if the reduction amplitude corresponding to the obtained harmonic mean deviation of the working parameters is not greater than the preset reduction amplitude in the database, input the working parameter deviation and the working parameter average value re-obtained after one reduction of the memory allocation into the thread number adjustment algorithm to output the increase amplitude of the thread number and return to F21. The deviation of the harmonic mean of the second working parameter represents the difference between the minimum value of the harmonic mean of the historical working parameters and the harmonic mean of the obtained working parameters.

9. An additive product production process monitoring method based on edge computing, characterized in that, It includes the following steps: Step 1, collect in real time the multi-modal data of the specified additive product in the target monitoring area at the end of the additive manufacturing period. The multi-modal data is collected in real time by the edge computing nodes with sensors and actuators deployed in the target monitoring area. The multi-modal data includes thermal imaging data and acoustic signal data. Step 2, perform abnormal hot spot identification based on the obtained thermal imaging data to obtain the abnormal hot spot identification result. The abnormal hot spot identification is used to analyze the temperature change of the specified additive product during the additive manufacturing process. Step 3, based on the abnormal hot spot identification result, judge whether to upload it to the cloud server. If so, perform time-frequency analysis on the obtained acoustic signal data to obtain the time-frequency analysis result. Otherwise, perform step size optimization. The step size optimization means adjusting the step size of the 3D printing path to improve the heat dissipation capacity during the 3D printing process. The time-frequency analysis is used to analyze the operating state of the corresponding additive manufacturing equipment during the additive manufacturing process of the specified additive product. Step 4, based on the time-frequency analysis result, judge whether to send it to the edge node. If so, perform edge node optimization according to the obtained working parameters. Otherwise, continue to monitor the production process of the specified additive product in the next additive manufacturing period. The working parameters are used to reflect the fault repair status of the corresponding additive manufacturing equipment during the additive manufacturing process of the specified additive product. The edge node optimization means adjusting the memory allocation and the number of threads to improve the load balance of the edge node. The specific process of judging whether to upload to the cloud server based on the abnormal hot spot identification result is as follows: If the obtained abnormal hot spot determination value is not greater than the preset abnormal hot spot determination value in the database, the abnormal hot spot recognition result is recorded as no abnormal temperature component in the molten pool coverage area, and the corresponding thermal imaging data is uploaded to the cloud server. Otherwise, the abnormal hot spot recognition result is recorded as having an abnormal temperature component in the molten pool coverage area and the step size is adjusted. The specific process of the step size adjustment is as follows: the obtained hot spot abnormal determination value deviation and step size deviation are input into the step size PID control algorithm of the additive manufacturing equipment to output the actual step size adjustment amplitude; when the re-obtained hot spot abnormal determination value after the step size adjustment is not greater than the preset hot spot abnormal determination value in the database and the adjustment times do not exceed the preset times, the step size adjustment is completed and the re-obtained thermal imaging data is uploaded to the cloud server. Otherwise, a shutdown instruction is sent and a preset person is prompted to perform maintenance; the hot spot abnormal determination value deviation is used to quantify the difference degree between the obtained abnormal hot spot determination value and the preset abnormal hot spot determination value; the step size deviation is used to quantify the difference degree between the actual step size and the target step size of the corresponding 3D printing path at the end of the additive manufacturing period.

Images