Crucible bubble detection method and system based on nano-robot and storage medium

Abstract

The invention provides a crucible bubble detection method and system based on a nano-robot and a storage medium, and relates to the technical field of quality detection.The method comprises the steps that a dynamic detection path is analyzed and positioned according to historical crucible detection data; and in the process that the motion driving module guides the nano robot to move on the surface of the quartz crucible to be detected by taking the dynamic detection path as a constraint, the intelligent decision-making module is synchronously operated to carry out bubble detection processing, and a real-time bubble distribution heat map is output. The technical problems that in the prior art, the crucible bubble detection efficiency is low, some hidden or difficult-to-touch bubbles cannot be accurately recognized, and the accuracy and comprehensiveness of the detection result are limited are solved. The technical effects that different types of bubbles are accurately and intelligently recognized and processed, efficient bubble detection and processing are ensured, and meanwhile the overall bubble detection and processing efficiency of the crucible is improved are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of quality detection, and in particular to a crucible bubble detection method, system and storage medium based on nanorobots. Background Art

[0002] Existing technologies have significant limitations in crucible bubble detection, mainly reflected in low detection efficiency and the inability to accurately identify certain hidden or hard-to-reach bubbles.

[0003] Specifically, although traditional optical microscope detection is simple and intuitive, it has limited detection capabilities for tiny bubbles or bubbles inside crucibles, and cannot effectively penetrate complex material structures. Although advanced optical or ultrasonic technologies can achieve non-invasive detection, their application in complex structures or opaque materials is still limited and cannot meet the needs of high-precision detection.

[0004] In addition, although chemical treatment methods can help remove bubbles in some cases, they may damage the crucible material or introduce new impurities, affecting product quality.

[0005] Therefore, the existing technology has deficiencies in bubble detection efficiency and accuracy, especially when detecting hidden or hard-to-reach bubbles, it is unable to provide comprehensive and accurate detection results, which limits the effectiveness and application of bubble detection. Summary of the invention

[0006] The present application provides a nanorobot-based crucible bubble detection method, system and storage medium, which are used to solve the technical problems in the prior art that the crucible bubble detection efficiency is low and certain hidden or hard-to-reach bubbles cannot be accurately identified, resulting in the accuracy and comprehensiveness of the detection results being limited.

[0007] In view of the above problems, the present application provides a crucible bubble detection method, system and storage medium based on nanorobots.

[0008] The first aspect of the present application provides a crucible bubble detection method based on a nanorobot, the method comprising: obtaining historical crucible detection data by calling local detection data; performing bubble distribution analysis based on the historical crucible detection data, and locating multiple bubble detection areas based on the analysis results; performing regional distribution analysis on the multiple bubble detection areas, and generating a dynamic detection path based on the analysis results; pre-embedding an intelligent decision-making module and a motion drive module into the nanorobot; while the motion drive module uses the dynamic detection path as a constraint to guide the nanorobot to move on the surface of the quartz crucible to be detected, synchronously running the intelligent decision-making module to perform bubble detection processing and output a real-time bubble distribution heat map.

[0009] The second aspect of the present application provides a crucible bubble detection system based on nanorobots, the system comprising: a detection data calling unit, used to obtain historical crucible detection data by calling local detection data; a detection area positioning unit, used to perform bubble distribution analysis based on the historical crucible detection data, and locate multiple bubble detection areas based on the analysis results; a detection path generation unit, used to perform regional distribution analysis on the multiple bubble detection areas, and generate dynamic detection paths based on the analysis results; a device configuration processing unit, used to pre-embed an intelligent decision-making module and a motion drive module into the nanorobot; a bubble detection execution unit, used to synchronously run the intelligent decision-making module to perform bubble detection processing and output a real-time bubble distribution heat map while the motion drive module uses the dynamic detection path as a constraint to guide the nanorobot to move on the surface of the quartz crucible to be detected.

[0010] The third aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, and provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the steps of the nanorobot-based crucible bubble detection method described in any one of the first aspects above.

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

[0012] The method provided in the embodiment of the present application obtains historical crucible detection data by calling local detection data; performs bubble distribution analysis based on the historical crucible detection data, and locates multiple bubble detection areas based on the analysis results; performs regional distribution analysis on the multiple bubble detection areas, and generates dynamic detection paths based on the analysis results; pre-embeds an intelligent decision-making module and a motion drive module into the nanorobot; while the motion drive module uses the dynamic detection path as a constraint to guide the nanorobot to move on the surface of the quartz crucible to be detected, the intelligent decision-making module is synchronously operated to perform bubble detection processing and output a real-time bubble distribution heat map. The technical effect of accurately and intelligently identifying and processing different types of bubbles, ensuring efficient bubble detection and processing, and improving the overall crucible bubble detection and processing efficiency is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of the process of the crucible bubble detection method based on nanorobots provided in this application;

[0014] Figure 2 A schematic diagram of a process for locating multiple bubble detection areas in a crucible bubble detection method based on nanorobots provided in the present application;

[0015] Figure 3A schematic structural diagram of the nanorobot-based crucible bubble detection system provided in this application.

[0016] Explanation of the accompanying drawings: detection data calling unit 11, detection area positioning unit 12, detection path generating unit 13, equipment configuration processing unit 14, bubble detection execution unit 15. DETAILED DESCRIPTION

[0017] The present application provides a crucible bubble detection method, system and storage medium based on nanorobots, which are used to solve the technical problems of low crucible bubble detection efficiency and inability to accurately identify certain hidden or hard-to-reach bubbles in the prior art, resulting in limited accuracy and comprehensiveness of the detection results. The invention achieves the technical effect of accurately and intelligently identifying and processing different types of bubbles, ensuring efficient bubble detection and processing, while improving the overall crucible bubble detection and processing efficiency.

[0018] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0019] Embodiment 1, as Figure 1 As shown, the present application provides a crucible bubble detection method based on nanorobots, the method comprising:

[0020] A100: Obtain historical crucible detection data by calling local detection data.

[0021] Specifically, in this embodiment, existing data resources are used to obtain relevant information about quartz crucible detection in the past, and the historical crucible detection data is obtained, wherein the historical crucible detection data includes bubble distribution detection of the same type of quartz crucible after multiple historical productions, and multiple sample bubble distribution data are obtained, wherein each sample bubble distribution data includes multiple bubble quantity, size, distribution, position, and type information of multiple historical detection bubbles.

[0022] A200: Perform bubble distribution analysis based on the historical crucible detection data, and locate multiple bubble detection areas based on the analysis results.

[0023] In one embodiment, Figure 2As shown, based on the historical crucible detection data, bubble distribution analysis is performed, and multiple bubble detection areas are located based on the analysis results. The method step A200 provided in the present application includes:

[0024] A210: Analyze the historical crucible detection data to obtain multiple sample bubble distribution data.

[0025] A220: Modeling the quartz crucible to be tested according to the design information to obtain a real-time crucible model.

[0026] A230: According to the plurality of sample bubble distribution data, bubble distribution fitting is performed in the real-time crucible model to obtain an initial bubble distribution model.

[0027] A240: Divide the detection area in the initial bubble distribution model to obtain the multiple bubble detection areas.

[0028] A250: Analyze the detection priorities of the multiple bubble detection areas, obtain multiple detection priority coefficients, and use the multiple detection priority coefficients to identify the multiple bubble detection areas.

[0029] In one embodiment, the detection priorities of the plurality of bubble detection areas are analyzed to obtain a plurality of detection priority coefficients. The method step A250 provided in the present application includes:

[0030] A251: Perform regional feature extraction on the multiple bubble detection areas to obtain multiple groups of bubble distribution features, wherein each group of bubble distribution features includes bubble density features, bubble type features, bubble size features, bubble shape features, bubble position features, bubble evolution features and bubble clustering features.

[0031] A252: Pre-built priority evaluation function.

[0032] A253: Substitute the multiple groups of bubble distribution characteristics into the priority evaluation function to calculate and obtain the multiple detection priority coefficients.

[0033] In one embodiment, the priority evaluation function is as follows:

[0034]

[0035] Among them, P is the detection priority coefficient, D is the bubble density feature, T is the bubble type feature, S max is the maximum size of the bubble in the bubble size feature, S avg is the average size of bubbles in the bubble size feature, R is the bubble shape feature, L is the bubble position feature, V is the bubble evolution feature, and λ is the bubble clustering feature.

[0036] Specifically, in this embodiment, information about bubble distribution is extracted from historical crucible detection data to obtain multiple sample bubble distribution data. According to the design parameters of the quartz crucible to be detected (such as the shape, size, structure, material, etc. of the crucible), a digital model of the crucible is generated by modeling software or calculation methods (such as CAD, 3D modeling, etc.), and the real-time crucible model is obtained. The real-time crucible model is a digital representation of the physical structure of the crucible, which is used for subsequent bubble distribution fitting and path planning.

[0037] Combine multiple sample bubble distribution data with the real-time crucible model and use a fitting algorithm (such as least squares method, interpolation method or machine learning method) to fit the data. This fitting process will infer the possible distribution pattern of bubbles at different positions and conditions based on the historical bubble distribution data. The final generated bubble distribution initial model will be able to reflect the possible bubble distribution inside the crucible.

[0038] In the initial model of bubble distribution, the density, size, and shape of the bubbles are considered and cluster analysis is used to spatially segment the crucible model to obtain the multiple bubble detection areas. The multiple bubble detection areas will be used as detection target areas for subsequent priority analysis and path planning.

[0039] Perform regional feature extraction on the multiple bubble detection areas to obtain multiple groups of bubble distribution features, wherein each group of bubble distribution features includes bubble density features (the number of bubbles per unit area or unit volume), bubble type features (the number of bubble types in the area), bubble size features, bubble shape features (bubble shape regularity index, aspect ratio or surface area to volume ratio.) It should be understood that the greater the aspect ratio or regularity (i.e., the more irregular) the impact on priority, the higher the priority, bubble position features (normalized distance from the bubble to the key area), bubble evolution features (bubble size change rate) and bubble cluster features (local density of bubbles in the area). Substitute the multiple groups of bubble distribution features into the priority evaluation function to calculate and obtain the multiple detection priority coefficients.

[0040] This embodiment divides the crucible into multiple bubble detection areas by analyzing the historical bubble distribution characteristics, thereby achieving the technical effect of providing basic data for the subsequent construction of the driving trajectory of the nanorobot for bubble detection.

[0041] A300: Perform regional distribution analysis on the multiple bubble detection areas, and generate a dynamic detection path based on the analysis results.

[0042] In one embodiment, the regional distribution analysis of the multiple bubble detection areas is performed, and a dynamic detection path is generated based on the analysis results. The method step A300 provided in the present application includes:

[0043] A310: Sort the multiple bubble detection areas according to the multiple detection priority coefficients to obtain a detection area execution sequence.

[0044] A320: Generate an initial detection path in the real-time crucible model according to the detection area execution sequence.

[0045] A330: Perform obstacle avoidance analysis on the initial detection path to generate the dynamic detection path.

[0046] A340: Load the dynamic detection path into the path control analysis model to obtain a dynamic detection control sequence.

[0047] A350: Associate and store the dynamic detection path and the dynamic detection control sequence.

[0048] Specifically, in this embodiment, areas with higher priority coefficients represent areas with higher bubble density or greater processing difficulty, and these areas need to be detected first. Based on this, the multiple bubble detection areas are sorted according to the multiple detection priority coefficients to obtain a detection area execution sequence.

[0049] According to the obtained detection area execution sequence, each target detection area is determined in order, and a preliminary path is generated in the real-time crucible model. Path generation uses path planning algorithms, such as A* algorithm, Dijkstra algorithm, etc., to calculate the shortest path or optimal path between each detection area. With the assistance of the real-time crucible model, the generated path is ensured to be physically feasible and effective, while the accuracy of path planning is improved, ensuring the smooth progress of the subsequent detection process.

[0050] It should be understood that after the initial path is generated, the path must be analyzed for obstacle avoidance to ensure that the path will not be affected by obstacles during the actual execution of the crucible bubble detection. Based on this, real-time obstacle detection technology (such as ultrasonic sensors or visual sensors) is used to identify possible obstacles inside the crucible (such as crucible walls, support structures, etc.). Through a dynamic obstacle avoidance algorithm, such as a path correction method based on local planning, the conflicting parts in the preliminary path are corrected to obtain the dynamic detection path that effectively avoids obstacles.

[0051] The dynamic detection path after obstacle avoidance analysis and optimization is input into the path control analysis model, which generates specific motion control instructions (such as speed, steering angle, acceleration, etc.) according to the robot's physical characteristics, dynamic constraints and control requirements. These control instructions serve as dynamic detection control sequences to ensure that the nanorobot can accurately follow the path when performing tasks and complete the predetermined detection tasks.

[0052] The dynamic detection path and control sequence are stored together in a database or path management system for associated storage. In this way, when the nanorobot is performing a detection task, the control system of the nanorobot can read these data in real time and adjust the path or control parameters according to the actual situation.

[0053] This embodiment achieves the goal of generating a dynamic detection path that can effectively perform global bubble detection on the crucible, and indirectly achieves the technical effect of ensuring global detection of bubble defects in the crucible.

[0054] A400: Pre-embed intelligent decision-making module and motion drive module into the nanorobot.

[0055] In one embodiment, the nanorobot is pre-embedded with an intelligent decision-making module and a motion driving module. Before that, the method step A400 provided in the present application further includes:

[0056] A410: Interactively obtain M sample bubble perception information sets and M sample bubble feature sets of M sample bubble types.

[0057] A420: Use the M sample bubble perception information sets and the M sample bubble feature sets as training data to construct M bubble recognition units.

[0058] A430: By connecting the M bubble recognition units in parallel, the construction of the crucible bubble recognition model is completed.

[0059] A440: Locate K types of processable bubbles from the M types of sample bubbles, and construct a crucible bubble processing model based on the K types of processable bubbles, where K is a positive integer less than M.

[0060] A450: Before embedding the intelligent decision-making module into the nanorobot, load the crucible bubble recognition model and the crucible bubble processing model into the intelligent decision-making module.

[0061] In one embodiment, K processable bubbles are located from the M sample bubble types, and a crucible bubble processing model is constructed based on the K processable bubbles. The method step A440 provided in the present application further includes:

[0062] A441: Filter and obtain K sample bubble feature sets of the K types of processable bubbles from the M sample bubble feature sets.

[0063] A442: Perform bubble processing information backtracking on the K sample bubble feature sets to obtain K sample processing parameter sets.

[0064] A443: Use the K sample bubble feature sets and K sample processing parameter sets as training data to construct K crucible bubble processing units.

[0065] A444: By connecting the K crucible bubble processing units in parallel, the construction of the crucible bubble processing model is completed.

[0066] Specifically, in this embodiment, first, M sample bubble perception information sets and M sample bubble feature sets of M sample bubble types are obtained through interaction. These perception information sets and feature sets contain perception data and feature data of different types of bubbles, respectively, and provide the basic data sets required for bubble recognition and processing model training. The perception information set may include raw data from sensors (such as optical sensors, acoustic sensors, etc.), while the feature set contains key features of bubbles extracted from these data, such as the gas composition, size, shape, position, density, etc. of the bubbles.

[0067] These perception information sets and feature sets are used as training data to construct M bubble recognition units. Each recognition unit is independently responsible for identifying and classifying a specific type of bubble. Through machine learning techniques (such as neural networks, support vector machines, etc.), each recognition unit is trained to enable it to accurately identify different types of bubbles, as well as the specific size, shape, position, density and other information of the bubbles based on the input perception data and feature data.

[0068] By connecting the M bubble recognition units in parallel, the construction of the crucible bubble recognition model is finally completed. Each bubble recognition unit processes one type of bubble, and these recognition units are connected in parallel to form a complete recognition system. This parallel structure can handle the recognition tasks of multiple bubble types at the same time, improving the efficiency and processing speed of bubble recognition.

[0069] Further, K types of processable bubbles are located from the M types of sample bubbles, and a crucible bubble processing model is constructed based on the K types of processable bubbles, where K is a positive integer less than M. The purpose of this step is to screen out the types of bubbles that can be directly processed and eliminated by the nanorobot, and the number of K is less than M because not all types of bubbles can be directly identified and processed by the nanorobot.

[0070] In order to construct an effective bubble processing model, K sample bubble feature sets of K processable bubbles are obtained by screening from M sample bubble feature sets.

[0071] Based on the selected K sample bubble feature sets, the bubble processing information is backtracked to obtain K sample processing parameter sets. The sample processing parameters are the processing parameter settings used when the historical nanorobot successfully eliminated the bubbles. According to the obtained K sample bubble feature sets and K sample processing parameter sets, K crucible bubble processing units are constructed. Each processing unit generates corresponding control parameters based on the input bubble characteristics for efficient processing of specific types of bubbles. In this way, the processing unit can intelligently generate and optimize the processing control parameters for different bubble types, thereby improving the processing effect and efficiency.

[0072] Finally, the crucible bubble processing model is constructed by connecting the K crucible bubble processing units in parallel. Similar to the recognition model, the processing model also adopts a parallel structure, so that each processing unit can work in parallel, improving the overall processing efficiency and real-time response capability.

[0073] Once the bubble recognition model and bubble processing model are built, they will be loaded into the intelligent decision-making module. The intelligent decision-making module will use these two models to make real-time decisions, select appropriate processing strategies based on the bubble recognition results, and pass relevant processing instructions to the nanorobot for execution. Through this process, the nanorobot can efficiently and automatically complete the bubble detection and processing tasks, ensuring the accuracy of the detection effect and the effectiveness of the processing strategy.

[0074] A500: When the motion driving module guides the nanorobot to move on the surface of the quartz crucible to be detected with the dynamic detection path as a constraint, the intelligent decision-making module is synchronously operated to perform bubble detection processing and output a real-time bubble distribution heat map.

[0075] In one embodiment, the motion driving module uses the dynamic detection path as a constraint to guide the nanorobot to move on the surface of the quartz crucible to be detected, and synchronously runs the intelligent decision-making module to perform bubble detection processing and output a real-time bubble distribution heat map. The method step A500 provided in this application includes:

[0076] A510: When the motion driving module guides the nanorobot to move to the first bubble detection area with the dynamic detection path as a constraint, the micro-sensing unit is activated to sense the first bubble detection area to obtain a first bubble perception information set, wherein the first bubble detection area is the area corresponding to the maximum value of the detection priority coefficient among the multiple bubble detection areas.

[0077] A510: Send the first bubble perception information set to the crucible bubble recognition model of the intelligent decision-making module, perform bubble state recognition via M bubble recognition units in the crucible bubble recognition model, and obtain M groups of bubble feature information.

[0078] A520: Divide the M groups of bubble feature information into MK groups of bubble feature information and K groups of bubble feature information according to the K types of processable bubbles.

[0079] A530: According to the mapping relationship between the K types of processable bubbles and the K crucible bubble processing units in the crucible bubble processing model, the K groups of bubble characteristic information mappings are loaded into the K crucible bubble processing units to obtain K groups of real-time bubble processing parameters.

[0080] A540: Perform bubble processing in the bubble detection area according to the K groups of real-time bubble processing parameters.

[0081] A550: Construct the real-time bubble distribution heat map in the real-time crucible model according to the bubble characteristic information of the MK group.

[0082] Specifically, in this embodiment, first, the nanorobot is guided to the first bubble detection area according to a predetermined dynamic detection path. In this process, the motion drive module ensures that the nanorobot reaches the target detection area smoothly according to dynamic path planning and obstacle avoidance analysis.

[0083] Once the nanorobot reaches the area, the micro-sensing unit (such as optical, acoustic or electrochemical sensor) inside the nanorobot is activated and starts working to detect and collect relevant information of the bubbles in the area, and obtain a first bubble perception information set, wherein the first bubble detection area is the area corresponding to the maximum value of the detection priority coefficient among the multiple bubble detection areas.

[0084] The first bubble sensing information set is sent to a crucible bubble recognition model in an intelligent decision-making module, wherein the model includes M bubble recognition units. Each bubble recognition unit recognizes the bubble state of a specific component through the received bubble sensing information, and outputs M groups of bubble feature information, which includes various features of the bubble, such as the shape, size, density, type, etc. of the bubble.

[0085] After obtaining the M groups of bubble characteristic information, the M groups of bubble characteristic information are divided into MK groups of bubble characteristic information and K groups of bubble characteristic information according to the characteristics of the K types of processable bubbles. The MK group of bubble characteristic information represents bubbles that are not suitable for direct processing, while the K group of bubble characteristic information represents the types of bubbles that can be directly processed by the nanorobot.

[0086] Next, according to the mapping relationship between the K types of processable bubbles and the K bubble processing units in the crucible bubble processing model, the K sets of bubble feature information are loaded into the corresponding processing units. Each bubble processing unit will generate real-time bubble processing parameters for a specific type of bubble based on the input bubble feature information. These parameters may include control parameters such as processing method, time, intensity, etc. to ensure efficient bubble removal.

[0087] Based on the obtained K sets of real-time bubble processing parameters, the nanorobot will perform bubble processing in the corresponding bubble detection area. These processing methods include eliminating, dissolving or other processing methods for bubbles that can be directly removed to improve product quality.

[0088] At the same time, based on the bubble characteristic information of the MK group and the real-time crucible model, a real-time bubble distribution heat map was constructed. The heat map reflects the distribution of bubbles in each area of ​​the crucible, including the treated area and the area to be treated, providing real-time feedback for subsequent detection and processing.

[0089] By analogy, the nanorobot is controlled to traverse the multiple bubble detection areas in sequence to eliminate bubbles, and the bubbles that cannot be eliminated are updated in the real-time bubble distribution heat map.

[0090] This embodiment achieves accurate and intelligent identification and processing of different types of bubbles, ensuring efficient bubble detection and processing, improving the overall detection and processing efficiency, while providing dynamic information on bubble distribution, ensuring that the crucible production process can be optimized and adjusted according to actual conditions.

[0091] Embodiment 2 is based on the same inventive concept as the crucible bubble detection method based on nanorobots in the aforementioned embodiment. Figure 3 As shown, the present application provides a crucible bubble detection system based on nanorobots, wherein the system comprises:

[0092] The detection data calling unit 11 is used to obtain historical crucible detection data by calling local detection data.

[0093] The detection area positioning unit 12 is used to perform bubble distribution analysis based on the historical crucible detection data, and locate multiple bubble detection areas based on the analysis results.

[0094] The detection path generating unit 13 is used to perform regional distribution analysis on the plurality of bubble detection areas and generate a dynamic detection path based on the analysis result.

[0095] The device configuration processing unit 14 is used to pre-embed the intelligent decision-making module and the motion driving module into the nanorobot.

[0096] The bubble detection execution unit 15 is used to synchronously run the intelligent decision-making module to perform bubble detection processing and output a real-time bubble distribution heat map when the motion driving module guides the nanorobot to move on the surface of the quartz crucible to be detected with the dynamic detection path as a constraint.

[0097] In one embodiment, the detection area positioning unit 12 includes:

[0098] The historical crucible detection data is parsed to obtain multiple sample bubble distribution data; a real-time crucible model is obtained by modeling according to the design information of the quartz crucible to be detected; bubble distribution fitting is performed on the real-time crucible model according to the multiple sample bubble distribution data to obtain an initial bubble distribution model; detection areas are divided in the initial bubble distribution model to obtain the multiple bubble detection areas; detection priorities of the multiple bubble detection areas are analyzed to obtain multiple detection priority coefficients, and the multiple bubble detection areas are identified using the multiple detection priority coefficients.

[0099] In one embodiment, the device configuration processing unit 14 includes:

[0100] Interactively obtain M sample bubble perception information sets and M sample bubble feature sets of M sample bubble types; use the M sample bubble perception information sets and M sample bubble feature sets as training data to construct M bubble recognition units; complete the construction of a crucible bubble recognition model by connecting the M bubble recognition units in parallel; locate K processable bubbles from the M sample bubble types, and construct a crucible bubble processing model based on the K processable bubbles, wherein K is a positive integer less than M; and load the crucible bubble recognition model and the crucible bubble processing model into the intelligent decision-making module before embedding the intelligent decision-making module into the nanorobot.

[0101] In one embodiment, the device configuration processing unit 14 includes:

[0102] K sample bubble feature sets of the K types of processable bubbles are obtained by screening from the M sample bubble feature sets; bubble processing information of the K sample bubble feature sets is backtracked to obtain K sample processing parameter sets; the K sample bubble feature sets and the K sample processing parameter sets are used as training data to construct K crucible bubble processing units; and the construction of the crucible bubble processing model is completed by connecting the K crucible bubble processing units in parallel.

[0103] In one embodiment, the bubble detection execution unit 15 includes:

[0104] When the motion driving module guides the nanorobot to move to the first bubble detection area with the dynamic detection path as a constraint, the micro-sensing unit is activated to sense the first bubble detection area to obtain a first bubble perception information set, wherein the first bubble detection area is an area corresponding to the maximum value of the detection priority coefficient among the multiple bubble detection areas; the first bubble perception information set is sent to the crucible bubble recognition model of the intelligent decision-making module, and the bubble state is recognized by M bubble recognition units in the crucible bubble recognition model to obtain M groups of bubble feature information; the M groups of bubble feature information are divided into MK groups of bubble feature information and K groups of bubble feature information according to the K kinds of processable bubbles; according to the mapping relationship between the K kinds of processable bubbles and the K crucible bubble processing units in the crucible bubble processing model, the K groups of bubble feature information are mapped and loaded into the K crucible bubble processing units to obtain K groups of real-time bubble processing parameters; the bubble processing of the bubble detection area is performed according to the K groups of real-time bubble processing parameters; the real-time bubble distribution heat map is constructed in the real-time crucible model according to the MK group of bubble feature information.

[0105] In one embodiment, the detection path generation unit 13 includes:

[0106] The multiple bubble detection areas are sorted according to the multiple detection priority coefficients to obtain a detection area execution sequence; an initial detection path is generated in the real-time crucible model according to the detection area execution sequence; an obstacle avoidance analysis is performed on the initial detection path to generate the dynamic detection path; the dynamic detection path is loaded into a path control analysis model to obtain a dynamic detection control sequence; and the dynamic detection path and the dynamic detection control sequence are stored in association.

[0107] In one embodiment, the detection area positioning unit 12 includes:

[0108] Perform regional feature extraction on the multiple bubble detection areas to obtain multiple groups of bubble distribution features, wherein each group of bubble distribution features includes bubble density features, bubble type features, bubble size features, bubble shape features, bubble position features, bubble evolution features and bubble clustering features; pre-construct a priority evaluation function; substitute the multiple groups of bubble distribution features into the priority evaluation function to calculate and obtain the multiple detection priority coefficients.

[0109] In one embodiment, the detection area positioning unit 12 includes:

[0110] The priority evaluation function is as follows:

[0111]

[0112] Among them, P is the detection priority coefficient, D is the bubble density feature, T is the bubble type feature, S max is the maximum size of the bubble in the bubble size feature, S avg is the average size of bubbles in the bubble size feature, R is the bubble shape feature, L is the bubble position feature, V is the bubble evolution feature, and λ is the bubble clustering feature.

[0113] Embodiment three, based on the same inventive concept as the crucible bubble detection method based on nanorobots in the aforementioned embodiment one, the present application also provides a computer-readable storage medium, on which a computer program is stored, and on which a computer program is stored. When the computer program is executed, the steps of the crucible bubble detection method based on nanorobots in any one of the aforementioned embodiment one are implemented.

[0114] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0115] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the patent protection scope of the present invention.

Claims

1. A crucible bubble detection method based on nanorobots, characterized in that: The method comprises: Obtain historical crucible detection data by calling local detection data; performing bubble distribution analysis based on the historical crucible detection data, and locating multiple bubble detection areas based on the analysis results; Performing regional distribution analysis on the multiple bubble detection areas, and generating a dynamic detection path based on the analysis results; Pre-embed intelligent decision-making modules and motion drive modules into nanorobots; When the motion driving module guides the nanorobot to move on the surface of the quartz crucible to be detected with the dynamic detection path as a constraint, the intelligent decision-making module is synchronously operated to perform bubble detection processing and output a real-time bubble distribution heat map.

2. The nanorobot-based crucible bubble detection method according to claim 1, characterized in that: Based on the historical crucible detection data, bubble distribution analysis is performed, and multiple bubble detection areas are located based on the analysis results. The method includes: Analyze the historical crucible detection data to obtain a plurality of sample bubble distribution data; Modeling the quartz crucible to be tested according to the design information to obtain a real-time crucible model; According to the plurality of sample bubble distribution data, performing bubble distribution fitting on the real-time crucible model to obtain an initial bubble distribution model; Dividing the detection area on the initial bubble distribution model to obtain the plurality of bubble detection areas; The detection priorities of the multiple bubble detection areas are analyzed to obtain multiple detection priority coefficients, and the multiple bubble detection areas are identified using the multiple detection priority coefficients.

3. The method for detecting bubbles in a crucible based on nanorobots as claimed in claim 2, characterized in that: Before pre-embedding the intelligent decision-making module and the motion driving module into the nanorobot, the method further comprises: Interactively obtain M sample bubble perception information sets and M sample bubble feature sets of M sample bubble types; Using the M sample bubble perception information sets and the M sample bubble feature sets as training data, constructing M bubble recognition units; By connecting the M bubble recognition units in parallel, the construction of the crucible bubble recognition model is completed; Locating K processable bubbles from the M sample bubble types, and constructing a crucible bubble processing model based on the K processable bubbles, wherein K is a positive integer less than M; Before embedding the intelligent decision-making module into the nanorobot, the crucible bubble recognition model and the crucible bubble processing model are loaded into the intelligent decision-making module.

4. The method for detecting bubbles in a crucible based on nanorobots as claimed in claim 3, characterized in that: Locating K processable bubbles from the M sample bubble types, and constructing a crucible bubble processing model based on the K processable bubbles, the method further comprising: K sample bubble feature sets of the K processable bubbles are obtained by screening the M sample bubble feature sets; Perform bubble processing information backtracking on the K sample bubble feature sets to obtain K sample processing parameter sets; Using the K sample bubble feature sets and the K sample processing parameter sets as training data, constructing K crucible bubble processing units; The construction of the crucible bubble processing model is completed by connecting the K crucible bubble processing units in parallel.

5. The method for detecting bubbles in a crucible based on nanorobots according to claim 4, characterized in that: In the process of guiding the nanorobot to move on the surface of the quartz crucible to be detected by the motion driving module with the dynamic detection path as a constraint, the intelligent decision-making module is synchronously operated to perform bubble detection processing and output a real-time bubble distribution heat map, the method comprising: When the motion driving module guides the nanorobot to move to a first bubble detection area with the dynamic detection path as a constraint, the micro-sensing unit is activated to sense the first bubble detection area to obtain a first bubble sensing information set, wherein the first bubble detection area is an area corresponding to the maximum value of the detection priority coefficient among the multiple bubble detection areas; Sending the first bubble perception information set to the crucible bubble recognition model of the intelligent decision-making module, performing bubble state recognition via M bubble recognition units in the crucible bubble recognition model, and obtaining M groups of bubble feature information; Dividing the M groups of bubble feature information into MK groups of bubble feature information and K groups of bubble feature information according to the K types of processable bubbles; According to the mapping relationship between the K processable bubbles and the K crucible bubble processing units in the crucible bubble processing model, the K groups of bubble characteristic information are mapped and loaded into the K crucible bubble processing units to obtain K groups of real-time bubble processing parameters; Performing bubble processing in the bubble detection area according to the K groups of real-time bubble processing parameters; The real-time bubble distribution heat map is constructed in the real-time crucible model according to the MK group bubble characteristic information.

6. The method for detecting bubbles in a crucible based on nanorobots according to claim 2, characterized in that: Performing regional distribution analysis on the multiple bubble detection areas and generating a dynamic detection path based on the analysis results, the method comprising: Sorting the plurality of bubble detection areas according to the plurality of detection priority coefficients to obtain a detection area execution sequence; generating an initial detection path in the real-time crucible model according to the detection area execution sequence; Performing obstacle avoidance analysis on the initial detection path to generate the dynamic detection path; Loading the dynamic detection path into a path control analysis model to obtain a dynamic detection control sequence; The dynamic detection path and the dynamic detection control sequence are stored in association.

7. The method for detecting bubbles in a crucible based on nanorobots according to claim 2, characterized in that: Analyzing the detection priorities of the plurality of bubble detection areas to obtain a plurality of detection priority coefficients, the method comprising: Performing regional feature extraction on the multiple bubble detection areas to obtain multiple groups of bubble distribution features, wherein each group of bubble distribution features includes bubble density features, bubble type features, bubble size features, bubble shape features, bubble position features, bubble evolution features and bubble cluster features; Pre-built priority evaluation function; Substitute the multiple groups of bubble distribution features into the priority evaluation function to calculate and obtain the multiple detection priority coefficients.

8. The method for detecting bubbles in a crucible based on nanorobots according to claim 7, characterized in that: The priority evaluation function is as follows: Among them, P is the detection priority coefficient, D is the bubble density feature, T is the bubble type feature, S max is the maximum size of the bubble in the bubble size feature, S avg is the average size of bubbles in the bubble size feature, R is the bubble shape feature, L is the bubble position feature, V is the bubble evolution feature, and λ is the bubble clustering feature.

9. A crucible bubble detection system based on nanorobots, characterized in that: The steps for implementing the method according to any one of claims 1 to 8 include: A detection data calling unit, used to obtain historical crucible detection data by calling local detection data; A detection area positioning unit, used for performing bubble distribution analysis based on the historical crucible detection data, and positioning multiple bubble detection areas based on the analysis results; A detection path generating unit, configured to perform regional distribution analysis on the plurality of bubble detection areas and generate a dynamic detection path based on the analysis result; A device configuration processing unit is used to pre-embed an intelligent decision-making module and a motion drive module into the nanorobot; The bubble detection execution unit is used to synchronously run the intelligent decision-making module to perform bubble detection processing and output a real-time bubble distribution heat map when the motion driving module guides the nanorobot to move on the surface of the quartz crucible to be detected with the dynamic detection path as a constraint.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed, implements the steps of the nanorobot-based crucible bubble detection method described in any one of claims 1 to 8.

Images