Automatic ceramic glazing control system and method integrating machine vision and trajectory planning

Abstract

The invention discloses an automatic ceramic glazing control system and method fusing machine vision and track planning, and relates to the technical field of ceramic production automation, and the method comprises the steps: obtaining basic information and glazing demand information of to-be-glazed ceramic; according to the automatic ceramic glazing control system and method fusing machine vision and trajectory planning, three-dimensional data are obtained by shooting ceramic images from multiple angles and scanning, the three-dimensional features of surface profiles, textures and tiny defects can be accurately captured, even pits and protrusions can be recognized, and when a glazing path is planned, the glazing quality is greatly improved. The angle of the nozzle can be adjusted in real time according to the surface morphology of the ceramic to ensure that glaze is sprayed perpendicular to an arc-shaped surface and a special-shaped surface, and meanwhile, the moving speed of the nozzle and the glaze output quantity are matched to avoid material consumption deviation per unit area; the glazing thickness uniformity and the coverage area precision are improved through mutual cooperation, the problems that coating of special-shaped ceramics is missed, and the thickness fluctuation of curved-surface products is large are solved, and the percent of pass of complex ceramic products is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ceramic production automation, in particular to a ceramic automatic glazing control system and method combining machine vision and trajectory planning. BACKGROUND

[0002] In the ceramic production industry, the glazing process is a key link that determines the appearance and performance of the product. There are currently various non-manual glazing methods on the market, but there are still significant technical shortcomings in adapting to complex products, responding to environmental changes, and ensuring quality stability, which makes it difficult to meet the production needs of high precision and high efficiency. The specific problems are particularly prominent in the following mainstream glazing methods: Mechanical fixed trajectory glazing is commonly used in mass production lines of daily-use ceramics. The glazing nozzle moves along a fixed path by pre-setting the mechanical movement trajectory. However, this method has obvious limitations: first, the trajectory is completely fixed and cannot adapt to changes in product size. When switching product specifications on the production line, mechanical components need to be disassembled and trajectory parameters need to be adjusted, which takes a long time and causes the production line to stop, affecting order delivery efficiency. Second, it cannot recognize differences in product surface morphology. For areas with curved transitions, it still glazes according to a flat trajectory, which can cause uneven glazing thickness and reduce product pass rate. Third, it lacks the ability to adapt to surface defects. If the ceramic body has minor defects such as depressions, fixed trajectory glazing will cause defects to accumulate in the defect area, forming additional defects that need to be repaired manually, increasing production costs. Some high-end ceramic production lines have tried to introduce vision technology, using industrial cameras to capture two-dimensional images of ceramics to assist in adjusting glazing parameters. However, this method still has technical limitations: first, it can only obtain two-dimensional information and cannot build a three-dimensional model of the ceramic surface. For irregular products, it cannot recognize the three-dimensional morphology of protrusions and depressions, and during glazing, the nozzle may collide with protrusions or miss the depressions. Second, the vision function is single and only used to identify product location. It does not link with glaze characteristics and motion parameters for optimization. After identifying surface defects, it cannot automatically adjust the glaze output in the corresponding area, affecting product quality. Third, it has no historical data reuse capability. Each time a product type is changed, technicians need to re-adjust the glazing parameters based on new two-dimensional images, which takes a long time and cannot quickly respond to small batch, multi-variety production needs.

[0003] Although existing glazing methods have eliminated the complete dependence on manual labor, they still have technical shortcomings: traditional mechanical fixed trajectory glazing has poor adaptability, simple automatic glazing has low precision and weak anti-interference ability, and single vision-assisted glazing lacks three-dimensional recognition and linkage optimization capabilities. These problems result in low pass rate, low changeover efficiency, and poor environmental adaptability in ceramic glazing production, which cannot meet the needs of the current ceramic industry towards high-end, customization, and flexibility. Therefore, we propose a ceramic automatic glazing control system and method that combines machine vision and trajectory planning. SUMMARY

[0004] (I) Technical problems solved In view of the deficiencies in the prior art, the present application provides a ceramic automatic glazing control system and method fusing machine vision and trajectory planning, which realizes the full-process automation, precision and intelligence of ceramic glazing, improves the glazing quality stability, reduces the labor cost and shortens the production cycle.

[0005] (II) Technical solutions To achieve the above object, the present application is realized by the following technical solutions: The ceramic automatic glazing control method fusing machine vision and trajectory planning, the glazing control method comprising: Obtaining the basic information and glazing requirement information of the ceramic to be glazed, wherein the basic information is the information reflecting the appearance and surface condition of the ceramic, including the shape size, surface shape and surface state of the ceramic; the glazing requirement information is the glazing related standard determined according to the production requirements, including the target thickness and target coverage area of glazing; The ceramic is visually processed through image acquisition and analysis technology, which obtains ceramic images through special equipment and extracts useful information through calculation and analysis, and obtains the characteristic data and glazing parameters of the ceramic after processing, and establishes a three-dimensional model of the ceramic surface, which can stereoscopically display the surface morphology of the ceramic, and realizes the stereoscopic visualization of the ceramic characteristics through the model; According to the characteristic data, glazing parameters and three-dimensional model of the ceramic, the influence of real-time environmental temperature and humidity data on trajectory planning is analyzed, trajectory planning is a process of designing the motion route and related parameters of the glazing device, and the motion path and motion parameters of the glazing device are designed based on the analysis to form a glazing trajectory scheme which can be adjusted according to the actual situation; According to the glazing trajectory scheme, the glazing device is controlled to perform automatic glazing operation on the ceramic, and the temperature change data of the glazing area is collected in real time during the process, and the material output and motion speed of the glazing device are dynamically adjusted according to the temperature change data; The glazed ceramic is visually detected and the actual glazing thickness data is obtained through a laser thickness measuring device, which measures the glazing layer thickness by using laser technology, and forms the glazing quality result containing various evaluation contents in combination with the visual detection result; determine whether the glazing quality result meets the preset quality standard, if yes, store the current glazing parameters, track scheme and quality result to the database to form a case library, and the case library is classified and searched according to the ceramic type, glazing requirement and environmental condition; if not, intelligently match the current quality defect reason based on the successful glazing data in the case library similar to the current ceramic characteristics, generate parameter adjustment suggestions and optimize the glazing track scheme, re-execute the track scheme until the glazing quality result meets the preset quality standard.

[0006] Preferably, the step of obtaining the basic information and glazing requirement information of the ceramic to be glazed comprises: Place the ceramic to be glazed on a preset positioning platform specially used for fixing the ceramic and assisting in determining the position thereof, and determine the accurate position of the ceramic on the platform through double calibration of a positioning sensor and visual positioning, the positioning sensor can sense the position of the ceramic, the visual positioning determines the position of the ceramic through image analysis, and the accurate position of the ceramic on the platform is determined after double calibration, and the positioning error is controlled within a range corresponding to a high accuracy; Start an image acquisition device composed of a high-definition industrial camera and a structured light scanner, the high-definition industrial camera is used for industrial scenes and has high shooting accuracy, the structured light scanner obtains three-dimensional information of an object by emitting structured light and receiving reflected light, and the ceramic image is shot and the ceramic surface is scanned from three or more different angles, such as front, side and top, and two-dimensional image data reflecting the planar form of the ceramic and three-dimensional point cloud data composed of a large number of three-dimensional coordinate points are obtained at the same time; Fuse the two-dimensional image data and the three-dimensional point cloud data, which combines the useful information of the two kinds of data, and extract the shape size data of the ceramic after processing, the size data includes height, diameter, length and width; Analyze the form type of the ceramic surface through the three-dimensional model, accurately determine whether it is a plane, a curved surface, an arc surface or a special-shaped surface, and label the transition angle and the radius of curvature between different surfaces, the transition angle is the included angle of the connecting part of adjacent surfaces, and the radius of curvature reflects the bending degree of the curved surface; Detect the surface state of the ceramic through image gray scale analysis and three-dimensional point cloud comparison, the image gray scale analysis obtains information by analyzing the light and shade degree of different areas in the image, and the three-dimensional point cloud comparison compares the actual three-dimensional point cloud data of the ceramic with standard data, detects whether there are problems such as cracks, depressions, protrusions or stains on the surface of the ceramic through the two ways, and quantifies parameters such as the depth and width of the defects; According to the glazing requirement document, determine the specific value of the target thickness of glazing and the specific position range of the target coverage area on the ceramic surface, associate the range with the coordinate system of the three-dimensional model of the ceramic, the coordinate system is used to determine the position of each point in the three-dimensional model, and the accurate positioning of the coverage area is realized through the association.

[0007] Preferably, the step of visually processing the ceramic through the image acquisition and analysis technology to obtain the feature data of the ceramic and the glazing parameters comprises: The obtained ceramic image and three-dimensional point cloud data are preprocessed, Gaussian filtering is used to remove image noise points, Gaussian filtering can reduce useless interference points in the image, histogram equalization is used to enhance the contrast of the image, histogram equalization can make the light and dark difference of the image more obvious, and a point cloud denoising algorithm is used to optimize the three-dimensional data, the point cloud denoising algorithm can reduce useless points in the three-dimensional point cloud data, and high-quality image and three-dimensional model data are obtained after preprocessing; The feature data of the ceramic is extracted from the high-quality data, specifically including curve equation parameters describing the contour curve of the ceramic surface, texture gray value distribution law reflecting the distribution characteristics of the light and dark degree of the ceramic surface, three-dimensional coordinates and volume size determining the defect position and size; Based on the shape and surface state of the ceramic, a deep learning algorithm is used to identify the key areas for glazing, the deep learning algorithm can simulate the human learning process and obtain the rules from the data, and the identified key areas are specific range areas around surface defects and areas with large transition angles of special-shaped surfaces; Combined with the target thickness of glazing and the material characteristics of the ceramic surface, material data is obtained through a material detection module, the material detection module is specially used for detecting the material properties related to the ceramic surface, an association model reflecting the relationship between the output of glazing material and the final glazing thickness is established after obtaining the data, the output of glazing material of the glazing equipment in different areas is calculated through the model, and the distance parameter between the glazing equipment and the ceramic surface is dynamically determined according to the curvature change of the ceramic surface, so that the glazing thickness meets the target requirements; The feature data, glazing material output and distance parameter are summarized to form a glazing parameter set containing all necessary information, and a mapping relationship table clearly showing the correspondence between each parameter and the three-dimensional model is generated.

[0008] Preferably, the step of designing the motion path and motion parameters of the glazing equipment according to the feature data of the ceramic and the glazing parameters to form a glazing trajectory scheme comprises: According to the surface shape of the ceramic, the target coverage area of glazing and the three-dimensional model, an A* path planning algorithm is used to plan the motion path of the glazing equipment, the A* path planning algorithm can efficiently find the optimal motion route, the path needs to completely cover all target coverage areas, and the path overlap rate is controlled at a low level to avoid uneven glazing thickness; According to the different shape regions of the ceramic surface, the motion speed of the glazing equipment is calculated combined with the fluidity parameters of the glazing material, the fluidity parameters are obtained from a material characteristics database, the material characteristics database stores attribute data of various glazing materials, the motion speed of the planar region is set to a high level, the motion speed of the curved surface region is set to a medium level, and the motion speed of the special-shaped surface region is set to a low level, so as to ensure uniform glazing; The kinematics inverse solution algorithm is used to determine the angle adjustment mode of the glazing equipment, and the kinematics inverse solution algorithm can calculate the movement angle of each component of the equipment according to the target position. When the ceramic surface is an arc surface or a special-shaped surface, the spraying angle of the glazing equipment is adjusted in real time to make the glazing material perpendicular to the ceramic surface, and the angle adjustment reaches a high precision corresponding to the standard. According to the output of the glazing material in the glazing parameters, a flow and speed matching model is established in combination with the movement speed of the glazing equipment. The model reflects the matching relationship between the glazing material flow and the equipment movement speed, ensures that the glazing material usage per unit area meets the target thickness requirement, and the thickness error is controlled within a small range. The movement path, movement speed and angle adjustment mode are integrated in time sequence, the influence of real-time environmental temperature and humidity on the material solidification speed is combined, the material solidification speed is the speed of the glazing material changing from liquid to solid, the trajectory scheme is dynamically corrected according to the influence, and the glazing trajectory scheme of the glazing equipment position, speed and angle at each time node is formed.

[0009] Preferably, the step of controlling the glazing equipment to perform automatic glazing operation on the ceramic according to the glazing trajectory scheme specifically includes: The movement path, movement speed and angle adjustment mode in the glazing trajectory scheme are converted into control instructions recognizable by the glazing equipment, the control instructions can make the glazing equipment perform corresponding actions, and the instruction transmission delay is controlled at a low level. According to the output of the glazing material in the glazing parameters, the material flow and spraying pressure of the glazing equipment are set through the flow control valve, the flow control valve is used to adjust the output of the glazing material, and after the setting is completed, a pre-spraying test is performed. The pre-spraying test is a small-scale spraying operation before formal glazing, and the material output is ensured to be stable through the test. Start the glazing equipment and perform movement and glazing operation according to the control instructions, and start the real-time monitoring device composed of a high-speed camera and a temperature sensor. The high-speed camera has a high shooting speed and can capture dynamic processes, and the temperature sensor can sense temperature changes. Image data and equipment operation data in the glazing process are obtained through the device at a high frequency. According to the real-time monitored image data, an image segmentation algorithm is used to judge whether there is a missed coating or over-thickness in the glazing area. The image segmentation algorithm can divide the image into different regions and identify specific regions. If there is a missed coating or over-thickness, the material flow and movement speed of the glazing equipment are adjusted in real time through a PID control algorithm. The PID control algorithm adjusts according to the deviation between the actual situation and the target, and the adjustment response time is controlled within a short time. The real-time monitored equipment operation data is compared with preset data in the trajectory scheme, the equipment operation data including equipment joint angle, motor speed, etc., if there is deviation, the motion parameters of the equipment are adjusted through the servo control system, the servo control system can accurately control the motion of the equipment, and the motion accuracy is ensured through adjustment.

[0010] Preferably, the step of visually inspecting the glazed ceramic to obtain glazing quality results comprises: A detection device composed of a high-definition camera and a laser thickness gauge is started, the high-definition camera has high shooting accuracy, and the laser thickness gauge measures thickness using laser technology, full-range images of the glazed ceramic are shot through the device, and actual thickness data of different areas of the ceramic surface are obtained, obtaining image data and thickness data after glazing; The image data after glazing is compared with the ceramic image data before glazing, and an image registration algorithm is used to analyze whether the glazing coverage area completely covers the target coverage area, the image registration algorithm can align different images for easy comparison and analysis, identify the location and range of the missed coating area, and reach a higher precision identification standard; The thickness distribution uniformity of the glazing layer is calculated by comparing the laser thickness data with the target thickness, the uniformity is the consistency degree of the thickness of each area of the glazing layer, whether the glazing thickness of different areas meets the target thickness requirement is judged, and the location and degree of the over-thick or over-thin area are identified; A defect detection algorithm is used to detect whether there are bubbles, sagging, pinholes and other glazing defects on the surface of the glazing layer, the defect detection algorithm can identify the surface problems of the glazing layer, and record the type, location and size of the defects; The coverage integrity, thickness uniformity and surface defect conditions are summarized, and a fuzzy comprehensive evaluation method is used to score the overall glazing quality, the fuzzy comprehensive evaluation method can evaluate multiple factors comprehensively, and form a glazing quality result containing the score result and defect details.

[0011] An automatic glazing control system for ceramics that combines machine vision and trajectory planning, comprising the following modules: An information acquisition module is used to collect ceramic-related information, obtain basic information and glazing requirement information of the ceramic to be glazed, the basic information reflects the appearance and surface condition of the ceramic and includes shape size, surface shape and surface state, the glazing requirement information is determined according to production requirements and includes target thickness and target coverage area of glazing; The module also integrates a material detection unit, which is specially used to detect the material properties of the ceramic surface, and can detect the material properties of the ceramic surface and output material data; a visual analysis module for visual processing and analysis of the ceramic, obtaining characteristic data and glazing parameters of the ceramic through image acquisition and analysis technology, which acquires ceramic images through special equipment and extracts useful information through calculation and analysis, and establishes a three-dimensional model of the ceramic surface, which can stereoscopically display the surface morphology of the ceramic, and realizes feature stereoscopic visualization through the model; a trajectory planning module for designing a glazing device motion related scheme, which is based on the characteristic data, glazing parameters and three-dimensional model of the ceramic, combined with real-time temperature and humidity data obtained by the environment perception module for obtaining surrounding environment temperature and humidity information, and designs the motion path and motion parameters of the glazing device based on these data, forming a glazing trajectory scheme that can be adjusted according to actual conditions; a glazing control module for controlling the glazing device to perform operations, which controls the glazing device to perform automatic glazing operation on the ceramic according to the glazing trajectory scheme, integrates a real-time monitoring unit and a PID control unit, the real-time monitoring unit can obtain real-time glazing process data, and the PID control unit adjusts by using a PID control algorithm, and through the two units, the device operation parameters are adjusted in real time to ensure the glazing accuracy; a quality detection module for evaluating the glazing quality, which performs visual detection and thickness measurement on the glazed ceramic, obtains actual thickness data through a laser thickness measurement device, which measures the glazing layer thickness by using laser technology, and combines the visual detection results to form a glazing quality result containing multiple evaluation contents; a judgment feedback module for judging the glazing quality and giving an adjustment scheme, which judges whether the glazing quality result meets the preset quality standard, if it meets, stores the related data according to the classification retrieval identifier to the case database; if it does not meet, performs similar case intelligent retrieval and parameter matching based on the case database, generates an optimization scheme, and restarts the glazing and detection process; a case database module for storing glazing related data, storing the glazing parameters, trajectory scheme and quality result of the previous successful glazing, supporting multi-dimensional classification retrieval and similar case intelligent recommendation, and providing data support for parameter adjustment.

[0012] Preferably, the information acquisition module specifically includes the following units: a ceramic positioning unit for placing the to-be-glazed ceramic on a preset positioning platform, which is specially used for fixing the ceramic and assisting in determining its position, and through double calibration of positioning sensors and visual positioning, the positioning sensors can sense the position of the ceramic, and the visual positioning determines the position of the ceramic through image analysis, and the accurate position of the ceramic on the platform is determined after double calibration; A multi-dimensional data acquisition unit is configured to start an image acquisition device composed of a high-definition industrial camera and a structured light scanner. The high-definition industrial camera is used for industrial scenes and has high shooting accuracy. The structured light scanner obtains three-dimensional information of an object by emitting structured light and receiving reflected light. The ceramic image is shot from multiple angles and the surface is scanned to obtain two-dimensional image data reflecting the ceramic plane morphology and three-dimensional point cloud data composed of a large number of three-dimensional coordinate points. A data fusion processing unit is configured to fuse the two-dimensional image data and the three-dimensional point cloud data. The processing combines the useful information of the two types of data. After processing, the size data of the ceramic is extracted, and the size data reaches a high-precision corresponding standard. A surface morphology analysis unit is configured to analyze the morphology type of the ceramic surface through a three-dimensional model, label the transition angle and the radius of curvature between different surfaces, and the transition angle is the included angle of the connecting part of adjacent surfaces, and the radius of curvature reflects the bending degree of the curved surface. A surface defect quantification unit is configured to detect and quantify the ceramic surface defects by combining image gray scale analysis and three-dimensional point cloud comparison. The image gray scale analysis obtains information by analyzing the light and dark degree of different regions in the image. The three-dimensional point cloud comparison compares the actual three-dimensional point cloud data of the ceramic with the standard data. Defects are detected and quantified by the two methods. A requirement analysis and mapping unit is configured to determine target parameters according to the glazing requirement document, associate the target coverage area with the coordinate system of the ceramic three-dimensional model, and determine the position of each point in the three-dimensional model through the association to realize accurate positioning of the coverage area.

[0013] Preferably, the visual analysis module specifically includes the following units: A multi-source data preprocessing unit is configured to preprocess the ceramic image and three-dimensional point cloud data, optimize data quality by using multiple algorithms, reduce useless interference points in the image, enhance the light and dark difference of the image, and optimize the three-dimensional point cloud data to obtain high-quality image and three-dimensional model data. A deep learning feature extraction unit is configured to extract feature data of the ceramic from high-quality data, including mathematical parameters describing the contour curve of the ceramic surface, texture gray value rules reflecting the distribution characteristics of the light and dark degree of the ceramic surface, three-dimensional coordinates and volume parameters determining the defect position and size, and providing data support for subsequent glazing parameter calculation. A key area intelligent identification unit is configured to identify the glazing focus area of a specific range around the surface defect, the large transition angle of the special-shaped surface, based on the shape and state of the ceramic surface, using a calculation method that can simulate the human learning process, to ensure the pertinence of the key area in the glazing process. The material and parameter correlation calculation unit is used for combining the glazing target thickness and the ceramic surface material characteristics, acquiring material data through a component specially detecting the ceramic surface material, establishing a correlation model reflecting the relationship between the glazing material output and the final glazing thickness, calculating the glazing material output of the glazing equipment in different regions through the model, and dynamically determining the distance parameter between the glazing equipment and the ceramic surface according to the ceramic surface curvature change, so as to ensure that the glazing thickness of each region reaches the target requirement. The parameter and model mapping unit is used for summarizing the feature data, the glazing material output and the distance parameter, forming a glazing parameter set containing all necessary information, and generating a mapping relationship table clearly showing the corresponding relationship between each parameter and the three-dimensional model, so as to provide intuitive data basis for the parameter calling of the trajectory planning module.

[0014] Preferably, the trajectory planning module specifically includes the following units: The intelligent path design unit is used for planning the movement path of the glazing equipment according to the surface shape of the ceramic, the glazing target coverage area and the three-dimensional model, using a calculation method capable of efficiently finding the optimal movement route, so that the path completely covers all target coverage areas and the overlap rate is controlled at a low level, thereby avoiding uneven glazing thickness caused by path repetition. The material characteristic and speed matching unit is used for acquiring the flowability parameter from a database storing the glazing material attribute data, and calculating the movement speed of the glazing equipment in combination with the parameter, wherein the speed of the planar region is set to a high level, the speed of the curved surface region is set to a medium level, and the speed of the irregular surface region is set to a low level, so as to ensure uniform glazing. The adjustment unit is used for determining the angle adjustment mode of the glazing equipment by using a calculation method capable of calculating the movement angle of the equipment component according to the target position, so that when the ceramic surface is an arc surface or an irregular surface, the spraying angle is adjusted in real time to make the glazing material spray perpendicular to the ceramic surface, and the angle adjustment reaches a high precision standard. The flow and speed dynamic matching unit is used for establishing a flow and speed matching model in combination with the movement speed of the glazing equipment according to the material output in the glazing parameter, so as to ensure that the amount of glazing material in a unit area meets the target thickness requirement, and the thickness error is controlled in a small range. The environment adaptive trajectory correction unit is used for integrating the movement path, the movement speed and the angle adjustment mode in time sequence, and combining the influence of the real-time environment temperature and humidity data on the material solidification speed, which is the speed of the glazing material changing from liquid to solid, to dynamically correct the trajectory scheme formed by the preliminary integration, so as to finally form a glazing trajectory scheme, which needs to clearly show the position, speed and angle of the glazing equipment at each time node, so as to ensure that the trajectory scheme can adapt to the environment change and guarantee the glazing quality stability under different temperature and humidity conditions.

[0015] (Three) beneficial effects 1. Through multi-angle shooting of ceramic images and scanning to obtain three-dimensional data, the three-dimensional features of surface profile, texture and micro-defects can be accurately captured, and even millimeter-level depressions and protrusions can be identified. When planning the glazing path, the nozzle angle is adjusted in real time according to the ceramic surface morphology to ensure that the glaze is sprayed vertically to the curved surface and the shaped surface, and the nozzle moving speed and the glaze output are matched to avoid unit area material usage deviation. The cooperation improves the uniformity of glazing thickness and the accuracy of the coverage area, completely solves the problem of uneven thickness of curved surface products, and significantly improves the pass rate of complex ceramic products. In addition, when planning the glazing scheme, the environmental temperature and humidity data are collected in real time to analyze their influence on the glaze solidification speed. When the temperature is too low, the nozzle moving speed is slowed down to prevent the glaze from flowing, and when the humidity is too high, the spraying pressure is adjusted to ensure stable glaze adhesion. And for different shapes and different surface states of ceramics, there is no need to adjust the hardware, and the product three-dimensional data can be quickly adapted by analyzing the product three-dimensional data, realizing no downtime during production change. This dual adaptation capability of environment and product greatly reduces the unqualified product rate caused by production fluctuation, and the adaptability is much higher than that of the existing glazing method.

[0016] 2. The successful glazing parameters and path schemes each time are stored according to product type and environmental conditions. When the product is replaced, similar case parameters can be directly called, adjusted according to the current product characteristics, and there is no need to start from zero. In addition, the product state and equipment deviation are monitored in real time during the glazing process, and problems such as missing coating and uneven thickness are corrected immediately to avoid batch defects. These designs shorten the production change debugging time from several days to less than half an hour, and the problems can be corrected without waiting for post-detection during production. The daily production capacity of a single production line is significantly improved. Secondly, the planned path realizes the automation of the entire glazing process without manual operation, which can greatly reduce labor input. At the same time, real-time monitoring and immediate correction function can solve the defect at the initial stage to avoid batch rework and significantly reduce glaze waste. In addition, the historical case reuse capability avoids the trial and error cost of each production change, and accurate parameters can be determined without repeated coating. BRIEF DESCRIPTION OF DRAWINGS

[0017] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application and can be implemented according to the content of the description, the following will be described in detail with the preferred embodiments of the present application and with the help of the drawings.

[0018] Figure 1 The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application and can be implemented according to the content of the description, the following will be described in detail with the preferred embodiments of the present application and with the help of the drawings. DETAILED DESCRIPTION

[0019] The embodiment of the application provides a ceramic automatic glazing control system and method fusing machine vision and trajectory planning, realizes full-process automation, precision and intelligentization of ceramic glazing, improves glazing quality stability, reduces labor cost and shortens production cycle, can accurately capture the three-dimensional features of surface profile, texture and tiny defects through multi-angle shooting of ceramic images and scanning to obtain three-dimensional data, and can even identify millimeter-level concave and convex, adjusts the nozzle angle in real time according to the ceramic surface morphology when planning the glazing path, ensures that the glaze is perpendicular to the arc surface and the special-shaped surface, matches the nozzle moving speed and the glaze output, and avoids the deviation of the unit area material consumption; the glazing thickness uniformity and the accuracy of the coverage area are improved, the problems of missing coating of special-shaped ceramics and large thickness fluctuation of curved surface products are solved, and the qualified rate of complex ceramic products is significantly improved.

[0020] Embodiment: The technical scheme in the embodiment of the application realizes full-process automation, precision and intelligentization of ceramic glazing, improves glazing quality stability, reduces labor cost and shortens production cycle, and the general idea is as follows: Ceramic glazing is a core process that determines the appearance quality, mechanical strength and corrosion resistance of ceramic products, and its quality directly affects the market competitiveness of products; current ceramic glazing production mainly relies on manual glazing and semi-automatic equipment, but both have obvious limitations: manual glazing requires operators to control the nozzle angle, distance and moving speed by experience, which leads to glazing thickness deviation, coverage area misplacement rate and single production cycle, and there is a certain uncontrollability; although the semi-automatic equipment can reduce part of manual intervention, it mainly supports regular-shaped ceramics and cannot adapt to special-shaped artistic ceramics or surface defect ceramics, and the trajectory planning scheme is fixed and does not consider the influence of environmental temperature and humidity on the solidification speed of glaze, for example, when the temperature fluctuates ±5℃ or the humidity fluctuates ±10%, the glazing qualified rate will decrease by 10% to 15%; in addition, both types lack a historical production data reuse mechanism, and the same type of ceramic needs to be re-adjusted parameters each time, which causes a certain degree of glaze waste and time loss.

[0021] In recent years, machine vision and trajectory planning technology have gradually increased in the application of ceramic glazing field, but the existing research still has technical gaps: some researches use monocular cameras to obtain two-dimensional images of ceramics, which can only identify the plane size and surface texture, and cannot capture three-dimensional features such as curved surface curvature and special-shaped transition angle, resulting in uneven glazing thickness of curved surface ceramics; some researches plan glazing paths based on A* algorithm, which improves the coverage integrity, but does not adjust the motion speed combined with the viscosity of glaze, and does not consider the influence of temperature and humidity on the solidification speed, and is prone to sagging defects in high temperature and high humidity environment; some researches design a quality detection system based on machine vision, which can identify missing coating, bubbles and other defects, but lacks real-time feedback with the glazing execution module, cannot dynamically correct parameters, and needs to be re-adjusted manually, which is low in efficiency.

[0022] In response to the problems existing in the prior art, the present invention provides a ceramic automatic glazing control method integrating machine vision and trajectory planning, the glazing control method comprising: 1. Architecture (1) Data interaction: The system architecture is built in a modular and distributed way. It mainly consists of 7 core modules. The modules interact with each other in real time through industrial Ethernet based on Profinet protocol. The overall architecture is as follows Figure 1 As shown; the functions of each module and the data flow are as follows: The information acquisition module serves as the data input terminal, collecting basic ceramic information, glazing requirements, and environmental data and transmitting them to the visual analysis module. Basic ceramic information includes size, shape, and defects. Glazing requirements include target thickness, coverage area, and other key data. Environmental data mainly includes temperature and humidity. The visual analysis module pre-processes the collected data, extracts features from it, calculates parameters such as glaze flow and equipment distance, and transmits them to the trajectory planning module. The trajectory planning module optimizes the path, speed, and angle based on the environmental data, generates a trajectory plan, and transmits it to the glazing control module. The glazing control module converts the trajectory plan into equipment instructions, executes the glazing operation, and transmits real-time monitoring data to the quality inspection module. The quality inspection module evaluates the glazing quality, including coverage integrity, thickness uniformity, and surface defects, generates quality results, and transmits them to the judgment and feedback module. The judgment and feedback module transmits qualified data to the case database for storage. If it is unqualified, parameter adjustment suggestions are retrieved from the case library and fed back to the glazing control module. The case database module stores historical production data, provides similar case retrieval services for the judgment and feedback module, and provides parameter recommendations for new production requirements. The specific contents are as follows: (2) Information collection module: Responsible for obtaining accurate raw data, including three key links: ceramic positioning, multi-dimensional data acquisition, and data fusion processing; ceramic positioning adopts dual calibration of laser displacement sensor and high-definition industrial camera. During operation, the ceramic is placed on a platform with positioning marks. The laser sensor detects the distance deviation between the bottom of the ceramic and the platform, and the camera shoots the relative position of the positioning mark and the edge of the ceramic. The coordinate calibration algorithm based on the least squares method is used to control the ceramic coordinate error within ≤0.1mm to facilitate the benchmark unification of subsequent image acquisition and glazing operations; in the multi-dimensional data acquisition link, the high-definition industrial camera shoots from 5 angles: front, side, top, 45° oblique and 135° oblique. For special-shaped ceramics, the shooting angle is set and added according to the shape of the ceramic to obtain a two-dimensional image of the ceramic surface texture and defects; the point cloud density scanned by the structured light scanner is 100 points / mm 2, a three-dimensional point cloud data is generated by scanning the ceramic surface to reflect the three-dimensional shape of the ceramic, such as the curvature radius of the curved surface and the transition angle of the special-shaped surface; a coordinate mapping algorithm is used in the data fusion processing link to associate the two-dimensional image with the three-dimensional point cloud, a coordinate system is established with the center point of the positioning platform as the origin, the pixel coordinates of the two-dimensional image are converted into three-dimensional space coordinates, then a feature matching algorithm is used to extract the ceramic size, i.e. height, diameter, length and width, to identify the surface shape as a plane, a curved surface or a special-shaped surface, the curved surface curvature radius identification range is 50 to 500 mm, the defect parameters such as crack length identification range is up to 100 mm, and the recess depth identification range is 0.1 to 1 mm, while the glazing requirement document is analyzed, such as the target thickness of 0.1 to 0.2 mm, the coverage area is the full surface or the local area, the coverage area is mapped to the three-dimensional coordinate system, and a data table corresponding to the ceramic features and the glazing requirements is generated.

[0023] (III) Visual analysis module: The original data is converted into executable glazing parameters, and the implementation steps are as follows: in the data preprocessing link, a 3x3 to 5x5 kernel Gaussian filter is used for two-dimensional images to remove random noise, the filter standard deviation is set to 0.5 to 1.0 according to the image noise intensity, the contrast of surface texture and defects is enhanced through histogram equalization in the gray level range of 0 to 255; for three-dimensional point cloud data, statistical filtering is used to remove outliers, the number of neighborhood points is set to 50, and the distance threshold is set to 1.5 times the standard deviation to ensure data quality; in the feature extraction link, the ceramic surface contour curve parameters are extracted based on a convolutional neural network, a ResNet-50 network structure is used, the training data set contains more than 1000 image data of different ceramics, a circular arc equation is fitted for circular ceramics; a segmented cubic B-spline curve is fitted for special-shaped ceramics, the node spacing is set to 5 mm; the three-dimensional coordinates of defects are extracted based on the PointNet algorithm, the input point cloud number is 10000 to 20000 points.

[0024] The glazing parameter calculation link is the core of the visual analysis module, and needs to be based on the ceramic material characteristics, glazing target thickness and equipment movement speed to establish a quantitative correlation. Through experimental data fitting and derivation, it aims to ensure that the amount of glaze per unit area accurately corresponds to the target thickness, avoiding excessive glaze leading to dripping and insufficient glaze leading to missed coating. The core calculation formula is: Q=kxdv, where Q is the glaze output, unit mL / min, indicating the volume of glaze sprayed by the glazing spray gun per unit time, directly determining the glaze accumulation per unit area, with a value range of 0.5 to 5.0 mL / min, which needs to match the adjustment range of the high-precision flow valve; its value needs to balance the thickness requirement and equipment capacity, for example, when the target thickness of a daily-use ceramic bowl is 0.1 mm and the speed is 10 mm / s, Q needs to be controlled at about 1.1 mL / min, too high is easy to lead to dripping, too low is easy to lead to missed coating; k is the material coefficient, reflecting the adsorption capacity and penetration characteristics of the ceramic surface to the glaze, with a value range of 1.0 to 1.5, which needs to be determined after measuring the water absorption and porosity of the ceramic by near-infrared spectrometer: when the water absorption is 2% to 5%, k=1.0 to 1.1, when the water absorption is 5% to 8%, k=1.2 to 1.3; if there are defects such as cracks and depressions on the ceramic surface, the defect area needs to be filled with additional glaze, and the k value needs to be increased by 0.2, such as k=1.1 for normal areas and k=1.3 for defect areas, to ensure that the glazing thickness of the defect area meets the standard; d is the glazing target thickness, unit mm, determined according to the ceramic product design standard and use scenario, with a value range of 0.1 to 0.2 mm: daily-use ceramics, such as bowls and plates, d=0.1 to 0.12 mm, too thick will lead to a decrease in glaze gloss, too thin will affect the surface smoothness; artistic ceramics d=0.12 to 0.15 mm, taking into account the decoration effect and glaze flatness; industrial ceramics d=0.15 to 0.2 mm, to improve wear resistance and corrosion resistance, the setting accuracy of d needs to reach 0.01 mm, matching the measurement accuracy of the laser thickness gauge, to ensure that the deviation between the target value and the measured value can be quantitatively evaluated; v is the equipment movement speed, unit mm / s, indicating the linear speed of the spray gun moving with the multi-axis robot arm, determining the residence time of the spray gun on the ceramic surface per unit area, with a value range of 3.0 to 15.0 mm / s, which needs to be adjusted in combination with the ceramic surface morphology and glaze viscosity: flat surface smooth, v=10 to 15 mm / s; curved surface with large curvature change, v=5 to 10 mm / s; complex-shaped surface, v=3 to 5 mm / s; glaze viscosity 500 to 600 mPa・s, i.e. low viscosity, at this time v can be appropriately increased, 600 to 800 mPa・s, i.e. high viscosity, at this time v needs to be reduced to prevent glaze accumulation; at the same time, v needs to match the path in the trajectory planning, the spiral path v remains stable, the segmented path v smoothly transitions between segments, the acceleration control is ≤5 mm / s 2 , avoiding speed fluctuations caused by mechanical arm impact.

[0025] The glazing parameters are calculated according to the ceramic material properties and target thickness, for example, the water absorption of a daily-use ceramic bowl is 5%, i.e. k = 1.1, the target thickness is 0.1 mm, and the planar area speed is 10 mm / s, Q = 1.1 x 0.1 x 10 = 1.1 mL / min; the water absorption of a crack area of an industrial ceramic pipe is 7%, i.e. k = 1.5, the target thickness is 0.15 mm, and the speed is 5 mm / s, Q = 1.5 x 0.15 x 5 = 1.125 mL / min; at the same time, the distance between the equipment and the ceramic is adjusted according to the ceramic surface curvature, the planar area is 9-10 mm, the curved surface area is 7-8 mm, and the special-shaped surface area is 5-6 mm, and the distance is adjusted by the mechanical arm z-axis movement; finally, a dictionary format parameter and three-dimensional coordinate mapping table is generated, the key is the coordinate range, such as x ∈ [0, 10] mm, y ∈ [0, 10] mm, z ∈ [50, 60] mm, and the value is the corresponding flow Q and distance h, which is convenient for the trajectory planning module to call parameters according to coordinates; in case of fault handling, if the image contrast after pretreatment is still insufficient, check the brightness of the ring light, which needs to be 500-1000 lux; if the feature extraction error exceeds the threshold, increase 50-100 samples of the same type of ceramic to retrain the neural network or adjust the point cloud sampling number to 25000 points; if the actual thickness after parameter calculation deviates from the target by ±0.02 mm, correct the material coefficient k, adjust by 0.05 each time and recalculate.

[0026] (Four) Trajectory planning and glazing control module: Trajectory optimization and precise execution are achieved cooperatively; the A* path planning algorithm is used for trajectory planning, the surface coordinates of the ceramic three-dimensional model are used as nodes, the target function is set as the shortest path length and the lowest overlap rate, the Euclidean distance is used for the heuristic function, and the path is ensured to completely cover the target area during planning, the coverage integrity is ≥99%, and the overlap rate is ≤5% to avoid uneven thickness caused by repeated glazing; the special-shaped ceramic adopts segmented path design, each segment corresponds to a surface form such as a planar segment, a curved surface segment, and a transition segment, the speed is smoothly transitioned at the connection between segments, and the acceleration is set to 5 mm / s 2Avoid impact; combine environmental data to correct trajectory, based on temperature 25℃, humidity 60%, temperature increases by 5℃, speed increases by 10%, such as 10mm / s to 11mm / s; humidity increases by 10%, spray pressure increases by 0.05MPa, such as 0.3MPa to 0.35MPa, finally generate XML format final trajectory plan, including each 0.1s time node mechanical arm position (x, y, z), motion speed v, spray angle θ, spray pressure P, ensure that the plan can be directly parsed by the glazing control module; glazing control selects a 6-axis mechanical arm equipped with a high-precision flow valve with a flow adjustment range of 0 to 5mL / min, first convert the trajectory plan into a command recognizable by the mechanical arm, transmit based on the Profinet protocol, then set the flow and distance according to the mapping table, start the pre-spray test, spray time 1 to 2s, collect the thickness of the test glaze, if the deviation is more than ±0.01mm, adjust again; during execution, a high-speed camera takes real-time photos of the glazing area, and a semantic segmentation algorithm based on U-Net is used to determine whether there is missed coating or over-thickness, the missed coating determination standard is area >0.5mm 2 , the over-thickness determination standard is 10% more than the target thickness; a temperature sensor monitors the surface temperature, if the temperature rises causing the solidification to speed up, a PID control algorithm is used to adjust the speed or flow in real time, the proportional coefficient Kp=2.0, the integral coefficient Ki=0.5, and the differential coefficient Kd=0.1; at the same time, the joint angle and motor speed of the mechanical arm are monitored, and when the deviation exceeds the threshold, the servo control system is corrected to ensure the motion accuracy.

[0027] (Five) Quality detection and case database module: Realize quality evaluation and data reuse; the quality detection device consists of a high-definition camera and a laser thickness gauge, the high-definition camera is the same as the equipment used in the information collection module, after glazing is completed, the camera takes images from 5 angles, the laser thickness gauge sets detection points on the ceramic surface at 2mm intervals, regular ceramics ≥50 points, special-shaped ceramics ≥100 points; fuzzy comprehensive evaluation method is used for quality analysis, a three-level index system is established: the first-level index is glazing quality and the weight is 1.0, the second-level indexes are coverage integrity, thickness uniformity and surface defects, the weights are 0.4, 0.35 and 0.25 respectively, in the third-level index, the missed coating area gets 100 points without missed coating, each 0.5mm 2 of missed coating deducts 10 points, the thickness standard deviation gets 100 points if it is ≤5% of the target thickness, each 1% over deducts 20 points, the number of bubbles gets 100 points without bubbles, each 1 bubble deducts 5 points, the comprehensive score ≥80 points is qualified, a quality report containing the score result, defect position map and thickness distribution table is generated.

[0028] The case database stores data in a MySQL database, and the data structure includes a ceramic information table, a glazing requirement table, a parameter table, an environment table, and a quality table. The ceramic information table records the type, size, and defects of the ceramic. The glazing requirement table records the target thickness and coverage area. The parameter table records the flow rate, speed, and pressure, with the flow rate parameter calculated based on the core formula Q=kxdv. The specific values of k, d, and v are stored in association for subsequent retrieval and reuse. The environment table records the temperature and humidity, and the quality table records the score and defects. A three-level index is established based on the ceramic type, glazing requirement, and environmental conditions, such as daily-use ceramics, full-surface, and 25°C 60% RH. During case retrieval, the type, requirement, and environmental parameters of the current ceramic are input, and the similarity between the current parameters and historical parameters is calculated using the cosine similarity algorithm. The similarity = (current parameter vector x historical parameter vector) / (||current parameter vector|| x ||historical parameter vector||), where the current parameters include the Q value calculated based on the formula and the corresponding k, d, and v, ensuring that the retrieved cases are highly matched in terms of parameter logic. Cases with a retrieval similarity of ≥85% are sorted and recommended in the top 3, with a parameter reuse accuracy of ≥90%. The database supports the storage of 100,000+ cases, with a retrieval response time of ≤1s. Invalid cases such as those using obsolete glazes are deleted regularly every month to ensure data effectiveness.

[0029] II. Complete implementation process of control method The ceramic automatic glazing control method combining machine vision and trajectory planning adopts a six-step process, with each step clearly defining operation details, parameter standards, and fault handling to ensure stable implementation. The specific process is as follows: Step 1: Information acquisition Information acquisition is the basis for all subsequent operations, and it is necessary to ensure data accuracy and completeness. When positioning the ceramic, place the ceramic to be glazed in the center of the platform with positioning marks. The positioning marks are cross-shaped engraved lines with a depth of 0.1 mm, a width of 0.2 mm, a spacing of 100 mm, and an accuracy of ±0.02 mm. Start the laser displacement sensor to detect the height deviation from the four symmetric points at the bottom of the ceramic. The detection points are 20 mm away from the edge. If the deviation exceeds 0.1 mm, adjust it by the platform fine tuning knob. At the same time, start the high-definition industrial camera to capture the relative image of the positioning marks and the ceramic edge. Use the least squares method to calculate the coordinate deviation of the ceramic center and the platform origin. If the deviation is >0.1 mm, control the platform electric slide to correct it. The electric slide has a travel of 50 mm and an accuracy of 0.005 mm, finally ensuring that the ceramic coordinate error is ≤0.1 mm.

[0030] Multi-dimensional data acquisition determines the collection angle according to the type of ceramic. Regular ceramic is collected from the front, side and top, 3 angles, and special-shaped ceramic adds 2 angles of 45° and 135°, a total of 5 angles. The high-definition industrial camera parameter setting is resolution 2592x1944, frame rate 30fps, exposure time 10-20ms, white ceramic takes 10ms to avoid overexposure, dark ceramic takes 20ms to ensure brightness, gain 1.0-1.5dB is adjusted according to the intensity of environmental light, and 1.2dB is taken under natural light. Each angle takes 5 images, removes the blurred images with a clarity lower than 0.8, and keeps 3 for subsequent processing; the structured light scanner is set to fine mode, and the point cloud density is 100 points / mm 2 . The scanning range covers the whole ceramic. If there is a point cloud missing area >5mm 2 , adjust the scanner angle to re-scan to ensure the integrity of the three-dimensional point cloud data.

[0031] Data fusion and demand analysis uses Python to write a coordinate mapping program, establishes a three-dimensional coordinate system with the platform origin as (0, 0, 0), converts the two-dimensional image pixel coordinates to three-dimensional space coordinates through the camera intrinsic matrix, sets the focal length in the camera intrinsic matrix to 8mm, and the principal point coordinates to (1296, 972). The conversion formula is x=(u-1296)×z / f, y=(v-972)×z / f, where x, y are three-dimensional space coordinates, z is the height value detected by the laser sensor, f is the distance from the optical center of the camera lens to the image sensor, which determines the magnification of the image, and the conversion error is controlled to be ≤0.03mm, and the pixel coordinates are (u, v); extract the ceramic size through the Canny edge detection algorithm of OpenCV, set the edge detection threshold to 100-200, the height is the z coordinate difference between the top and bottom of the ceramic, the diameter is the maximum x coordinate difference in the horizontal direction of the circular ceramic, and the length and width are the maximum coordinate differences in the x and y directions of the special-shaped ceramic; identify the surface morphology through the PointNet network, and determine that the surface curvature radius changes <5% as a plane, 5%-30% as a curved surface, and >30% as a special-shaped surface; detect surface defects through gray difference analysis, and if the gray difference between the pre-glazing image and the standard non-defect image is >30, it is determined to be a defect, and parameters such as crack length and depression depth are recorded; finally, parse the glazing demand document stored in XML format, extract the target thickness and target coverage area, mark the boundary coordinates of the coverage area in the three-dimensional coordinate system, and generate a CSV format ceramic feature and glazing demand data table for subsequent module calling; in terms of fault handling, if the positioning deviation exceeds the threshold repeatedly ≥3 times, check the levelness of the positioning platform and the calibration state of the laser sensor; if there is a missing data collection, check the cleanliness of the camera lens and the working distance of the scanner.

[0032] Step 2, visual processing: Visual processing needs to convert raw data into executable glazing parameters, ensuring that the parameters are accurately matched with ceramic features and glazing requirements. In the data preprocessing stage, MATLAB is used for Gaussian filtering and histogram equalization for two-dimensional images: Gaussian filtering selects a 3x3 kernel based on noise intensity, with σ=0.8 or a 5x5 kernel, σ=1.0, to remove random noise such as dust highlights, uneven light spots, and dark spots. Histogram equalization adjusts the distribution of gray levels from 0 to 255 to a uniform distribution, enhancing the contrast of surface texture and defects. The standard deviation of the processed image gray scale should be ≥50, and if the original image standard deviation is <30, repeat the processing once. For three-dimensional point cloud data, PCL is used to implement statistical filtering, setting the number of neighboring points to 50, calculating the average distance of each point and its neighboring points, and removing points with a distance greater than 1.5 times the standard deviation. The processed point cloud density remains 100 points / mm 2 , ensuring the accuracy of subsequent feature extraction.

[0033] In the fine feature extraction stage, a ResNet-50 network is built based on the TensorFlow framework. The training data set contains more than 1000 different types of ceramic images, including 300 types of daily-use ceramics, 400 types of artistic ceramics, and 300 types of industrial ceramics. Each sample contains 5 preprocessed images at different angles, and the label is the contour parameter and defect position. The network input is a 256x256 region of interest image, and the output is the contour curve parameter: for circular ceramics, the output is the center (a, b) and radius r of the circle equation (x-a) 2 + (y-b) 2 = r 2 , with a center error ≤0.02mm and a radius error ≤0.03mm. For irregular-shaped ceramics, the output is the node coordinates of the segmented cubic B-spline curve, with a node spacing of 5mm and a curve fitting error ≤0.04mm, and the maximum distance difference from the actual point cloud data. PointNet network is used to process three-dimensional point cloud data, with an input of 10000 to 20000 uniformly sampled point coordinates and an output of defect three-dimensional coordinates: for cracks, the output is the starting point (x1, y1, z1) and the ending point (x2, y2, z2); for depressions, the output is ≥8 boundary point coordinates and depth, with a depth error ≤0.05mm.

[0034] In the glazing parameter calculation stage, first, the Thermo Scientific Nicolet iS50 near-infrared spectrometer is used to detect the material properties of the ceramic surface, with a wavelength range of 4000-400cm -1 , 32 scans, and a resolution of 4cm -1, the water absorption rate is 3% to 8%, the porosity is 1% to 3%, the water absorption rate calculation formula is porosity = (mass after water absorption - dry mass) / dry mass x 100%, and the porosity calculation formula is porosity = (water absorption volume / total volume of ceramic) x 100%; Then, based on the core formula Q = k x d x v, the glaze flow is calculated, wherein the k value is determined according to the water absorption rate, the water absorption rate is 3% to 5%, and the k value is 1.0 to 1.1, the water absorption rate is 5% to 8%, and the k value is 1.2 to 1.3, d is the target thickness in the glaze requirement document, 0.1 to 0.2 mm, v is the equipment motion speed determined in the subsequent trajectory planning link, 3 to 15 mm / s; For example, the water absorption rate of a daily-use ceramic bowl is 5%, k = 1.1; the target thickness is 0.1 mm; and the plane area speed is 10 mm / s, Q = 1.1 x 0.1 x 10 = 1.1 mL / min; If the ceramic has a crack defect, the k value of the defect area is increased by 0.2, k = 1.3, and the speed is reduced to 5 mm / s, Q = 1.3 x 0.1 x 5 = 0.65 mL / min, which ensures that the defect area is filled with sufficient glaze; At the same time, the distance between the equipment and the ceramic is adjusted according to the surface curvature: 9 to 10 mm for a plane area, 7 to 8 mm for a curved surface area, and 5 to 6 mm for a special-shaped surface area, and the distance is adjusted by the z-axis motion of the mechanical arm; Finally, a dictionary format parameter and three-dimensional coordinate mapping table is generated, the key is the coordinate range, such as x ∈ [0, 10] mm, y ∈ [0, 10] mm, z ∈ [50, 60] mm, and the value is the corresponding flow Q and distance h, which is convenient for the trajectory planning module to call parameters according to coordinates; When handling faults, if the image contrast after preprocessing is still insufficient, check the brightness of the ring light source, which needs to be 500 to 1000 lux; If the feature extraction error exceeds the threshold value, increase 50 to 100 similar ceramic samples to retrain the neural network or adjust the point cloud sampling number to 25000 points; If the actual thickness deviates from the target by ±0.02 mm after parameter calculation, correct the material coefficient k and recalculate the Q value, and adjust by 0.05 each time.

[0035] Step 3, trajectory planning: Trajectory planning needs to combine ceramic features, glazing parameters and environmental data to generate a dynamic trajectory that can be directly executed; In the basic path design link, C++ is used to realize A* path planning algorithm on ROS platform, the map model is the three-dimensional point cloud data of the ceramic surface after preprocessing, the node is the point cloud coordinate point selected with an interval of 0.5 mm to ensure smooth path, and the cost function is the weighted sum of path length and overlap rate, and the weights are 0.6 and 0.4 respectively, and the heuristic function is the Euclidean distance from the current node to the target node; When planning, start from the starting point of the ceramic surface, adopt spiral path for circular ceramic, and increase the spiral radius by 2 mm per circle to match the glaze spraying diameter; The special-shaped ceramic is segmented according to the surface form, the linear path is adopted for the plane segment, the circular arc path is adopted for the curved surface segment, and the spline curve path is adopted for the transition segment, and the speed change rate at the connection between segments is ≤5 mm / s 2, avoid mechanical arm impact; after path generation, check coverage completeness and overlap rate, increase path density if coverage is insufficient, node interval is 0.3mm, adjust path direction if overlap rate is too high, such as changing clockwise to counterclockwise.

[0036] Motion parameter calculation link, according to the glaze viscosity detected by NDJ-8S rotary viscometer, 500 to 800 mPa・s, detection temperature 25℃, determine the basic motion speed v, which will be the key parameter of the formula Q=k×d×v: when viscosity is 500 to 600 mPa・s, i.e. low viscosity, flat area 12 to 15 mm / s, curved surface area 8 to 10 mm / s, special-shaped surface area 4 to 5 mm / s; when viscosity is 600 to 800 mPa・s, i.e. high viscosity, flat area 10 to 12 mm / s, curved surface area 5 to 8 mm / s, special-shaped surface area 3 to 4 mm / s; calculate the spray angle θ through the inverse kinematics algorithm based on the mechanical arm DH parameter table, θ=90°-α, where α is the included angle between the normal of the ceramic surface at this point and the vertical direction, ensure that the glaze spray direction is perpendicular to the ceramic surface through point cloud data normal vector calculation, angle adjustment accuracy ≤0.5°, control each joint angle through mechanical arm joint angle, each joint angle error ≤0.05°.

[0037] Environment adaptive correction link, obtain real-time environment data through Sick TH300 temperature and humidity sensor, correct based on 25℃, 60%RH: when temperature rises by 5℃, motion speed v increases by 10%, such as 10mm / s→11mm / s, at this time need to recalculate Q value according to formula Q=k×d×v simultaneously, such as original Q=1.1mL / min, v increases Q=1.1×1.1=1.21mL / min, respond to the faster glaze solidification speed; when temperature drops by 5℃, v decreases by 8%, such as 10mm / s→9.2mm / s, Q value decreases to 1.1×0.92≈1.01mL / min simultaneously, prevent glaze flowability decline leading to missed coating; when humidity rises by 10%, spray pressure increases by 0.05MPa, such as 0.3MPa→0.35MPa, realize through flow valve pressure adjustment, overcome humidity leading to glaze viscosity increase; when humidity drops by 10%, pressure decreases by 0.04MPa, such as 0.3MPa→0.26MPa, avoid flowability too strong leading to sagging; after correction, generate final trajectory scheme in XML format, including mechanical arm position (x, y, z), motion speed v, spray angle θ, spray pressure P and corresponding glaze flow Q of each 0.1s time node, ensure that the scheme can be directly parsed by the glazing control module; if path coverage is insufficient during fault handling, check target area coordinate marking accuracy; if mechanical arm motion exceeds joint range after motion parameter calculation, adjust starting point position; if sagging / missed coating still occurs after environment correction, increase correction coefficient and recalculate Q value and trajectory.

[0038] Step 4, glazing execution: Glazing execution is the core link of converting track scheme into actual operation, which needs to ensure the precise action and real-time adjustment of the equipment; the equipment initialization and instruction conversion link, the glazing equipment adopts ABB IRB12006 shaft mechanical arm, the end of which is installed with a glaze spray gun with a nozzle diameter of 0.5 mm, the spray range is 2 to 3 mm, and the flow valve is Festo MPPES, the flow adjustment range is 0 to 5 mL / min; the communication connection between the mechanical arm and the track scheme is established through ABB Robot Studio software, the XML format track scheme is converted into joint angle instructions recognizable by the mechanical arm based on Profinet protocol, each time node corresponds to 6 joint angle values, the precision is ±0.01°, according to the Q value calculated by the parameter and three-dimensional coordinate mapping table and formula, the initial flow of the flow valve is set, such as 1.1 mL / min, the initial pressure P of the pressure regulating valve is set, such as 0.3 MPa, the pre-spraying test is started: the mechanical arm moves to the test area beside the platform, sprays for 1 to 2 s according to the initial parameters, and detects the glaze thickness of the test area with a laser thickness gauge, if the deviation of the thickness from the target is more than ±0.01 mm, adjust the flow by ±0.1 mL / min or the pressure by ±0.01 MPa each time, until the deviation is ≤±0.01 mm, for example, when Q=1.1 mL / min, the measured thickness is 0.09 mm, the deviation is -0.01 mm, Q needs to be adjusted to 1.2 mL / min, and the measured thickness is 0.10 mm, which meets the standard.

[0039] Real-time monitoring and dynamic adjustment link, start the mechanical arm to execute the glazing operation, at the same time, start the Optronis CP80 high-speed camera and Sick TMT8 temperature sensor: the high-speed camera shoots dynamic images of the glazing area, through the semantic segmentation algorithm based on U-Net, the data set contains 5000+ glazing process images, the labels are glazed area, unglazed area and over-thick area, and the area of the missed coating unglazed area is identified to be >0.5 mm 2, or too thick, the thickness of the over-thick area exceeds the target by 10%, and there is a leak, then the PID control algorithm is used, Kp=2.0, Ki=0.5, Kd=0.1, increase the flow rate by +0.1mL / min each time, or reduce the speed by -0.5mm / s each time. After adjustment, the matching of the formula Q=k×d×v needs to be verified synchronously. For example, when the speed drops from 10mm / s to 9.5mm / s, Q needs to be adjusted from 1.1mL / min to 1.1×(9.5 / 10)=1.045mL / min to ensure that the glaze per unit area is The material dosage is stable; if the material is too thick, reduce the flow rate by -0.1mL / min each time, or increase the speed by +0.5mm / s each time, and the Q value also needs to be corrected synchronously; the temperature sensor monitors the surface temperature of the ceramic. If the temperature rises by ≥3℃ compared to the ambient temperature, increase the speed by 5% or reduce the flow rate by 5%. At this time, the Q value is adjusted synchronously according to the formula. For example, if the speed is increased from 10mm / s to 10.5mm / s, Q is adjusted from 1.1mL / min to 1.1×1.05=1.155mL / min to prevent local overheating and uneven curing.

[0040] At the same time, the joint angle and motor speed are monitored through the built-in encoder of the robotic arm. When the joint angle deviation exceeds ±0.1° or the speed deviation exceeds ±1r / min, the servo control system is started to correct it to ensure motion accuracy and speed stability, and avoid the imbalance of Q value and v matching due to speed fluctuations; in the glazing process recording link, the NIUSB-6211 data acquisition card is used to record key data such as time, robotic arm position, real-time Q value, pressure, temperature, image recognition results, etc., and store them in CSV format. The data must include the Q value and v value before and after each parameter adjustment to facilitate the subsequent tracing of the accuracy of the formula application.

[0041] During the interrupt processing phase, if there is insufficient glaze, a sudden drop in the flow valve pressure of more than 0.05MPa, an equipment abnormality alarm, or an arm joint temperature exceeding 45°C, the interrupt program is triggered: the arm immediately stops and moves to a safe position, ≥100mm away from the ceramic, closes the flow valve, and records the current glazing progress and the last valid Q and v values. After the fault is eliminated, execution continues from the interruption point, calling the coordinates and parameters in the historical trajectory data and recalculating the flow of the subsequent path based on the formula Q=k×d×v to avoid material waste caused by re-glazing. During troubleshooting, if the thickness deviation of the pre-spray test repeatedly exceeds the threshold, check the blockage of the flow valve nozzle, clean it with 0.3MPa compressed air, and calibrate it with the laser thickness gauge, and calibrate it once every two weeks; if the coating omission / overthickness still exists after real-time adjustment, check the stains on the high-speed camera lens, clean it with a dust-free cloth dipped in anhydrous alcohol, or retrain the image segmentation model, adding 100 to 200 similar glazing abnormal images; if it is impossible to continue from the interruption position after interruption, check the integrity of the historical trajectory data storage path, re-import the trajectory plan and restart from the end point of the completed area.

[0042] Step 5, quality detection: Quality detection needs to comprehensively evaluate the glazing effect and verify the application accuracy of the formula Q=kxdv to provide quantitative basis for subsequent feedback adjustment. In the data collection link, after glazing is completed, the ceramic is transferred to the detection platform with the same precision as the glazing platform, and the horizontal error is ≤0.02mm / m. A high-definition camera is started, which is consistent with the information collection module equipment, and the Keyence LK-G80 laser thickness gauge: the high-definition camera shoots the glazing ceramic image from the front, side, top, 45° oblique, and 135° oblique five angles, and shoots three at each angle to ensure that all glazing areas are covered. The laser thickness gauge sets detection points in a grid shape, regular ceramics in a 5mmx5mm grid, and special-shaped ceramics in a 3mmx3mm grid. The defect area is encrypted to a 1mmx1mm grid. Each detection point is measured three times to take the average value, and the coordinates (x, y, z) and the actual glazing thickness dreal of each detection point are recorded. At the same time, the Q value and the v value corresponding to the coordinates are associated to verify the deviation between the calculated value of the formula and the actual effect.

[0043] In the quality index analysis link, the glazing image after glazing and the target coverage area marking graph before glazing are matched and registered based on feature points using the OpenCV library of Python. Through the gray difference method, the gray value of the glazing area is 50 to 100 higher than that of the non-glazing area, the uncovered area is identified, the missing coating area is calculated, and the missing coating area is ≤0.5mm 2 determination as qualified, every 0.5mm 2 deduct 10 points, full score 40 points, missing coating area needs to trace corresponding Q value and v value, if Q value is more than 10% lower than the calculated value of the formula, k value needs to be corrected in subsequent production; thickness uniformity analysis calculates the thickness average value dflat of all detection points and the standard deviation σ, σ≤dtargetx5%, dtarget is the target thickness, and the determination is uniform, full score 35 points, σ>dtargetx5% every 1% deduction 7 points, at the same time, the formula calculates the thickness dcal=(Qxt) / (Sxr), t is the time of the spray gun staying in the area, S is the area, and r is the density of glaze. The deviation of dcal and dreal is ≤±0.01mm, which is qualified, and the over-difference area needs to analyze whether the matching of Q value and v value is reasonable; surface defect analysis identifies surface defects of the glazing layer through a defect detection model based on YOLOv5, and the training data set contains 500 images of defects such as bubbles, sagging, pinholes, etc. Bubble diameter >0.5mm, sagging length >2mm, and pinhole diameter >0.3mm are determined as unqualified defects. Every 1 defect deducts 5 points, full score 25 points, sagging defects need to trace whether it is caused by too high Q value or too low v value, and bubble defects need to check whether the spray pressure matches the Q value.

[0044] Quality result generation link, comprehensive three indicators score, coverage integrity 40 points + thickness uniformity 35 points + surface defects 25 points, total score ≥ 80 points for qualified, < 80 points for unqualified, generate PDF format quality report containing total score, index score, defect details, thickness distribution table, formula calculation value and actual value deviation analysis, the report must be clearly marked Q value, v value and k value of the overage area, providing direction for subsequent parameter optimization; When troubleshooting, if the detection data collection is not complete, the working distance of the laser thickness gauge must be within 20 to 100 mm, and the camera detection area must be aligned; If the defect detection accuracy is < 90%, increase 100 images of each defect to retrain the model; If the thickness distribution heat map shows local thickness abnormalities, backtrack the real-time data of the glazing execution step to check if Q value and v value are adjusted synchronously according to the formula.

[0045] Step 6, intelligent feedback: Intelligent feedback is the key to realizing the closed-loop optimization of glazing process and formula parameter reuse; qualified case storage link, if the quality report is qualified, store the complete data of this production according to ceramic information, glazing parameters, environmental data and quality results into the MySQL case database: ceramic information includes type, size, surface morphology, defect condition; glazing parameters include target thickness d, coverage area coordinates, Q value calculated based on the formula and corresponding k value, v value, device distance, spray pressure; environmental data includes temperature and humidity during glazing; quality results include total score, index score, thickness standard deviation, average deviation of formula calculation value and actual value, and four-level index is established according to ceramic type, d, environmental temperature and humidity, such as household ceramics, 0.1 mm, 25℃, 60%RH, index uses B+ tree structure, ensures retrieval time ≤ 1s, data storage capacity supports 100,000+ cases, and each case needs to be associated with the calculation process of formula parameters for subsequent reuse reference.

[0046] Unqualified parameter optimization link, if the quality report is unqualified, start similar case retrieval: first input the search keywords, including current ceramic type, d, environmental temperature and humidity and main defect type, such as missing coating, excessive thickness, sagging; then calculate the similarity with the cases in the database using cosine similarity algorithm, standardize the current parameters and historical case parameters, calculate the cosine value of the vector angle, cases with similarity ≥ 85% are included in the recommended range; finally generate optimization suggestions, extract the parameter adjustment rules related to the current defects from the recommended cases, for example, missing coating, defect in historical cases is solved by reducing v5% + increasing Q8%, and the adjusted parameters meet Q=k×d×v, this time the same adjustment direction is recommended, if the current missing coating area is more than 1mm 2The adjustment range can be fine-tuned to reduce v by 8% and increase Q by 10%, and it is verified whether the adjusted Q value meets the formula logic. For example, if v is reduced from 10 mm / s to 9.2 mm / s, k = 1.1, and d = 0.1 mm, then Q should be adjusted to 1.1 x 0.1 x 9.2 = 1.012 mL / min. Ensure that parameter adjustment is always based on formula correlation.

[0047] Re-execute the human pre-warning link, update the optimized k, v, and Q values to the parameter and three-dimensional coordinate mapping table, regenerate the trajectory scheme, and return to step 4 to re-operate; if the same ceramic is still unqualified after three consecutive optimizations, the total score is less than 80, the system triggers a manual pre-warning: a pre-warning window pops up, showing three consecutive unqualified, suggesting checking: 1. Is the glaze viscosity abnormal? The current viscosity is mPa・s, the standard range is 500 to 800 mPa・s, and abnormal viscosity will cause v value adaptation deviation, affecting the calculation accuracy of Q = k x d x v; 2. Is the joint precision of the mechanical arm up to standard? The current joint 1 error is °, and the standard is ≤0.05°. Insufficient precision will cause v value fluctuation; 3. Is the laser thickness gauge calibrated? It affects the d measurement, causing formula verification deviation, and sends a pre-warning SMS to the device administrator's mobile phone to avoid invalid repeated production; when troubleshooting, if the qualified case storage fails, check the database connection parameters, IP address, port number, username and password, data format, to ensure that numerical data such as k, v, and Q have no character garbled code; if the number of similar case search results is less than 3, reduce the similarity threshold to 80% or manually enter historical experience parameters, and the same type of ceramic debugging parameters provided by the device administrator need to meet the formula Q = k x d x v; if the problem is still not solved after manual pre-warning, contact the equipment manufacturer to calibrate the mechanical arm or replace the glaze batch, and re-determine the glaze viscosity to adjust the matching relationship between v and Q values.

[0048] Finally, it should be noted that the above embodiments are merely examples for the sake of clarity and do not limit the implementation. Based on the above description, those skilled in the art can make other different forms of changes or variations. Here, it is not necessary or possible to exhaust all implementation methods. The obvious changes or variations derived therefrom are still within the scope of protection of the present application.

Claims

1. A method of automatic glazing control of ceramics fusing machine vision with trajectory planning, characterized in that: The glazing control method comprises the following steps: Obtain the base information and glazing requirement information of the ceramic to be glazed, the base information including the shape size, surface shape and surface state of the ceramic, and the glazing requirement information including the target thickness and target coverage area of glazing; Perform visual processing on the ceramic through image acquisition and analysis technology, which obtains characteristic data and glazing parameters by acquiring ceramic images and analyzing and extracting the characteristic data and glazing parameters, and establishes a three-dimensional model of the ceramic surface; According to the characteristic data, glazing parameters and three-dimensional model of the ceramic, and in combination with the influence of real-time environmental temperature and humidity on trajectory planning, trajectory planning is a process of setting the movement route and parameters of the glazing equipment, forming a dynamically adjustable glazing trajectory scheme; Control the glazing equipment to perform glazing operation according to the glazing trajectory scheme, and in the process, real-time acquisition of temperature change data of the glazing area is performed, and the material output quantity and movement speed of the glazing equipment are dynamically adjusted according to the temperature change data; Perform visual detection and thickness measurement on the glazed ceramic to form a glazing quality result; Determine whether the glazing quality result meets the preset standard: if yes, store the data of this glazing into a database to form a case library; otherwise, collect the successful data of the ceramic in the case library, adjust the parameters and trajectory scheme, and perform the glazing operation again until the standard is met.

2. The method of claim 1, wherein the method further comprises: The process of obtaining the base information and glazing requirement information of the ceramic to be glazed comprises: Place the ceramic to be glazed on a detection platform, perform double calibration through a positioning sensor sensing the position of the ceramic and visual positioning determining the position through image analysis, determine the position of the ceramic on the platform, take ceramic images from multiple angles, acquire two-dimensional image data reflecting the planar morphology of the ceramic and three-dimensional point cloud data reflecting the three-dimensional morphology of the ceramic from the ceramic images, and fuse the two-dimensional image data and three-dimensional point cloud data, and extract shape size data including the height, diameter, length and width of the ceramic from the fused data; Analyze the ceramic surface morphology through the three-dimensional model, and mark the transition angle of adjacent surfaces and the curvature radius reflecting the bending degree of the curved surface, wherein the ceramic surface morphology is a plane, a curved surface, an arc surface or a special-shaped surface; combine image grayscale analysis and three-dimensional point cloud comparison; determine the target thickness value and target coverage area range according to the glazing requirement, and associate the area range with the coordinates of the ceramic three-dimensional model.

3. The method of claim 1, wherein the method further comprises: The process of performing visual processing on the ceramic through image acquisition and analysis technology to obtain characteristic data and glazing parameters comprises: Preprocess the ceramic images and three-dimensional point cloud data, the preprocessing including Gaussian filter denoising, histogram equalization to deepen contrast, and point cloud denoising algorithm to optimize data; extract ceramic characteristic data including surface contour curve equation parameters, surface texture grayscale value distribution law, surface defect three-dimensional coordinates and volume from the preprocessed data; identify the glazing key area based on the ceramic surface shape and state by using a deep learning algorithm; establish a digital correlation model of the glazing material output quantity and thickness based on the target glazing thickness and the ceramic surface material characteristics, calculate the material output quantity of different areas through the model, and dynamically adjust the distance parameter between the glazing equipment and the ceramic surface according to the change of the ceramic surface curvature. The glazing control system comprises:

4. The method of claim 1, wherein the method further comprises: An information collection module collects basic information and glazing requirement information of the ceramic to be glazed, the basic information including the shape size, the surface shape and the surface state, and the glazing requirement information including the target thickness and the target coverage area; A visual analysis module processes the ceramic through image acquisition and analysis technology of obtaining ceramic images and analyzing and extracting information to obtain characteristic data and glazing parameters, and establishes a three-dimensional model for stereoscopic display of the ceramic surface morphology; 5. The method of claim 1, wherein the method further comprises: A glazing parameter design module designs glazing equipment motion paths and parameters according to the ceramic characteristic data, and forms a glazing track scheme, the process including: According to the ceramic surface shape, the glazing target coverage area and the three-dimensional model, the A* path planning algorithm is used to plan the glazing equipment motion path; the glazing material fluidity parameters are obtained, and the glazing equipment motion speed is determined in combination with the ceramic surface shape, wherein the high speed is adopted for the planar area, the medium speed is adopted for the curved surface area, and the low speed is adopted for the special-shaped surface area; the kinematics inverse solution algorithm is used to determine the glazing equipment angle adjustment mode to keep the glazing material perpendicular to the ceramic surface; the flow rate and speed matching model reflecting the matching relationship between the material flow rate and the glazing equipment motion speed is established according to the glazing material output and the glazing equipment motion speed; the motion path, the speed and the angle adjustment mode are integrated in time sequence, and the track scheme is corrected in combination with the influence of the real-time environmental temperature and humidity on the material solidification speed. A glazing equipment control module controls the glazing equipment to execute automatic glazing according to the glazing track scheme, the process including:

6. The method of claim 1, wherein the method further comprises: The motion path, the speed and the angle adjustment mode in the glazing track scheme are converted into control instructions recognizable by the glazing equipment; the material flow rate and the spraying pressure of the glazing equipment are set through the flow control valve, and the pre-spraying test is performed after the setting; the glazing equipment is started to execute the motion and the glazing operation according to the control instructions, and the real-time monitoring device is started to obtain the images and the glazing equipment operation data in the glazing process; The real-time images are analyzed by using the image segmentation algorithm to judge whether there is a missed coating or an over-thick condition; if yes, the material flow rate and the motion speed are adjusted in real time according to the actual and target deviation through the PID control algorithm, otherwise, no response is made; the glazing equipment operation data including the joint angle and the motor speed are compared with the preset data of the track scheme, and the motion parameters are adjusted if there is a deviation.

7. A ceramic auto-enameling control system fusing machine vision and trajectory planning, characterized in that, A quality result obtaining module obtains the quality result of the glazed ceramic through visual detection, the process including: The full-range images of the glazed ceramic are shot, and the actual thickness data of different areas of the surface are obtained; the ceramic images before and after glazing are compared to analyze the completeness of the glazing coverage, identify the position and range of the missed coating area; the laser thickness measurement data and the target thickness are compared to calculate whether the glazing layer thickness of each area meets the target requirement, and identify the over-thick or over-thin area; then, the glazing layer surface is detected by using the defect detection algorithm to judge whether there is a defect including a bubble, a sagging and a pinhole, and record the type, position and size of the defect; the coverage completeness, the thickness and the surface defect condition are summarized to score the overall glazing quality by using the fuzzy comprehensive evaluation method to form the quality result including the scoring result and the defect details. The glazing control system comprises: An information collection module collects basic information and glazing requirement information of the ceramic to be glazed, the basic information including the shape size, the surface shape and the surface state, and the glazing requirement information including the target thickness and the target coverage area; A visual analysis module processes the ceramic through image acquisition and analysis technology of obtaining ceramic images and analyzing and extracting information to obtain characteristic data and glazing parameters, and establishes a three-dimensional model for stereoscopic display of the ceramic surface morphology; A glazing parameter design module designs glazing equipment motion paths and parameters according to the ceramic characteristic data, and forms a glazing track scheme, the process including: According to the ceramic surface shape, the glazing target coverage area and the three-dimensional model, the A* path planning algorithm is used to plan the glazing equipment motion path; the glazing material fluidity parameters are obtained, and the glazing equipment motion speed is determined in combination with the ceramic surface shape, wherein the high speed is adopted for the planar area, the medium speed is adopted for the curved surface area, and the low speed is adopted for the special-shaped surface area; the kinematics inverse solution algorithm is used to determine the glazing equipment angle adjustment mode to keep the glazing material perpendicular to the ceramic surface; the flow rate and speed matching model reflecting the matching relationship between the material flow rate and the glazing equipment motion speed is established according to the glazing material output and the glazing equipment motion speed; the motion path, the speed and the angle adjustment mode are integrated in time sequence, and the track scheme is corrected in combination with the influence of the real-time environmental temperature and humidity on the material solidification speed. A glazing equipment control module controls the glazing equipment to execute automatic glazing according to the glazing track scheme, the process including: The motion path, the speed and the angle adjustment mode in the glazing track scheme are converted into control instructions recognizable by the glazing equipment; the material flow rate and the spraying pressure of the glazing equipment are set through the flow control valve, and the pre-spraying test is performed after the setting; the glazing equipment is started to execute the motion and the glazing operation according to the control instructions, and the real-time monitoring device is started to obtain the images and the glazing equipment operation data in the glazing process; The real-time images are analyzed by using the image segmentation algorithm to judge whether there is a missed coating or an over-thick condition; if yes, the material flow rate and the motion speed are adjusted in real time according to the actual and target deviation through the PID control algorithm, otherwise, no response is made; the glazing equipment operation data including the joint angle and the motor speed are compared with the preset data of the track scheme, and the motion parameters are adjusted if there is a deviation. A quality result obtaining module obtains the quality result of the glazed ceramic through visual detection, the process including: The full-range images of the glazed ceramic are shot, and the actual thickness data of different areas of the surface are obtained; the ceramic images before and after glazing are compared to analyze the completeness of the glazing coverage, identify the position and range of the missed coating area; the laser thickness measurement data and the target thickness are compared to calculate whether the glazing layer thickness of each area meets the target requirement, and identify the over-thick or over-thin area; then, the glazing layer surface is detected by using the defect detection algorithm to judge whether there is a defect including a bubble, a sagging and a pinhole, and record the type, position and size of the defect; the coverage completeness, the thickness and the surface defect condition are summarized to score the overall glazing quality by using the fuzzy comprehensive evaluation method to form the quality result including the scoring result and the defect details. A trajectory planning module, according to the ceramic data, the glazing parameters and the three-dimensional model, and obtaining real-time temperature and humidity data, designs a glazing equipment motion path and parameters, and forms a dynamically adjustable glazing trajectory scheme; A glazing control module controls the glazing equipment to perform automatic glazing according to the glazing trajectory scheme, integrates real-time acquisition of glazing process data, and adjusts the glazing equipment parameters by using a PID algorithm; A quality detection module performs visual detection and thickness measurement on the glazed ceramic, obtains actual thickness data through a laser thickness measurement device, and forms a comprehensive quality result in combination with the visual detection result; A judgment feedback module judges whether the quality result meets the preset standard: if yes, the data is stored in a case database according to a classification identifier; otherwise, similar cases are searched based on the case database, and the parameters and the scheme are matched and adjusted, and the glazing and detection processes are restarted; A case database module stores the glazing parameters, the trajectory scheme and the quality result of each successful glazing.

8. The ceramic auto-enameling control system that fuses machine vision with trajectory planning of claim 7, wherein, The information acquisition module includes: A ceramic positioning unit places the to-be-glazed ceramic on a positioning platform, and performs double calibration through a positioning sensor sensing the ceramic position and a visual positioning determining the position through image analysis; A multi-dimensional data acquisition unit starts an industrial camera and a structured light scanner, takes ceramic images from multiple angles and scans the surface, and obtains two-dimensional images reflecting the ceramic plane shape and three-dimensional point cloud data reflecting the ceramic three-dimensional shape; A data fusion processing unit fuses the two-dimensional images and the three-dimensional point cloud data, and extracts ceramic outline size data therefrom; A surface morphology analysis unit analyzes the ceramic surface morphology through a three-dimensional model, labels the transition angle of adjacent surfaces and the curvature radius reflecting the curvature degree of the curved surface; A surface defect quantification unit detects ceramic surface defects and quantifies defect parameters in combination with image grayscale analysis and three-dimensional point cloud comparison; A requirement analysis and mapping unit determines target parameters according to a glazing requirement document, and associates the target coverage area with a three-dimensional model coordinate system determining the position of each point of the model.

9. The ceramic auto-enameling control system fusing machine vision with trajectory planning of claim 7, wherein, The visual analysis module includes: A preprocessing unit pre-processes the ceramic image and the three-dimensional point cloud data, and reduces image useless interference points, enhances image light and dark differences, and optimizes the three-dimensional point cloud data; A feature extraction unit extracts ceramic features including surface contour parameters, texture distribution rules and three-dimensional defect parameters from the data; An identification unit identifies glazing key areas by using a deep learning algorithm based on the ceramic surface shape and state; An association calculation unit establishes an association model of material output and thickness by combining the ceramic material characteristics obtained from the target thickness, calculates the material output of the equipment in different areas through the model, and dynamically determines the distance between the equipment and the ceramic surface; A model mapping unit forms a complete glazing parameter set by summarizing the feature data, the material output and the distance parameters, generates a mapping table of the parameters and the three-dimensional model, and provides data association basis for the trajectory planning.

10. The ceramic auto-enameling control system that fuses machine vision with trajectory planning of claim 7, wherein, The trajectory planning module includes: A path design unit plans an equipment motion path by using an A* algorithm according to the ceramic surface shape, the target coverage area and the three-dimensional model; Material property speed matching unit, obtaining the flowability parameters of the glazing material from the database, determining the equipment movement speed according to the ceramic surface shape, wherein the plane area is high speed, the curved surface area is medium speed, and the special-shaped surface area is low speed; Adjustment unit, adjusting the angle of the glazing equipment by using the inverse kinematics algorithm, so that the material is perpendicular to the ceramic surface; Flow speed dynamic matching unit, establishing a flow and speed matching model according to the material output and the equipment movement speed, and adjusting the material consumption per unit area to meet the target thickness; Trajectory correction unit, integrating the path, speed and angle adjustment mode in time sequence, and correcting the trajectory scheme in combination with the influence of the environment temperature and humidity on the material solidification speed.

Images