Broken bridge aluminum door and window frame intelligent trepanning system and method based on machine vision

Abstract

The invention discloses a broken bridge aluminum door and window frame intelligent trepanning system and method based on machine vision, and relates to the technical field of door and window processing equipment, and the system comprises a machine vision collection module which is composed of six industrial CCD cameras and an annular light source, and is used for collecting the three-dimensional contour, wall thickness, reinforcing rib position and surface defects of a frame; the frame positioning calibration module is used for extracting feature points to establish coordinate mapping and calibrating deviation through laser ranging; the trepanning parameter planning module is used for receiving requirements and profile parameters and planning a path by avoiding reinforcing ribs; the intelligent execution module comprises a three-axis numerical control platform, an automatic tool changer and a torque sensor; the quality detection module is used for judging the trepanning quality grade by combining secondary visual acquisition with laser diameter measurement; and the data storage and interaction module supports viewing of a touch screen and an APP and can export a production report. According to the system, broken bridge aluminum frame trepanning positioning is improved, the system is suitable for frames of multiple specifications, equipment safety and operation and maintenance convenience are enhanced, and intelligent machining is achieved.

Description

Technical Field

[0001] This invention relates to the field of door and window processing equipment technology, and in particular to a machine vision-based intelligent hole-opening system and method for thermally broken aluminum door and window frames. Background Technology

[0002] Due to their thermal insulation and high strength, thermally broken aluminum windows and doors are widely used in the construction industry. The opening of the frame (such as screw holes, drainage holes, and lock holes) is a critical production process that directly affects the assembly accuracy and performance of the windows and doors. Currently, frame opening in the industry is mostly completed manually with the assistance of semi-automatic equipment, which has significant technical limitations: manual positioning relies on measuring tapes and pens to determine hole positions, which is greatly affected by the operator's experience, and the positioning error often reaches 1-2mm. If the opening position deviates and touches the internal reinforcing ribs or weak walls of the cavity, it will cause the frame to crack and be scrapped, especially for thin-walled frames such as the 55 series and 60 series, where the scrap rate is as high as 8%-12%. At the same time, it is difficult for humans to accurately identify surface defects (such as scratches and dents) and internal cavity structures, making it easy to open holes in weak areas, affecting the overall strength of the frame.

[0003] With the application of machine vision technology in the processing field, some companies have introduced single-camera or dual-camera vision systems to assist in hole drilling, but the core problems have not yet been solved: insufficient visual acquisition dimensions, only able to acquire front or side images of the frame, unable to fully capture the three-dimensional contour and internal reinforcing rib distribution, resulting in the need for manual intervention in hole drilling path planning; lack of thermal deformation compensation mechanism, the thermally broken aluminum profile will expand due to the heat generated by tool friction during hole drilling, and the hole diameter is easily smaller than the design value after cooling, requiring secondary hole enlargement correction, increasing process time and material loss; poor adaptability to multiple frame specifications, when changing different models of frames such as 55 series to 70 series, it is necessary to manually adjust equipment parameters (such as tool speed and feed rate), the debugging process takes 1-2 hours, which cannot meet the needs of multi-batch small-volume production; in addition, tool wear monitoring relies on manual observation of the cutting edge condition, if the tool is dulled and not replaced in time, it will lead to excessive hole wall roughness, and the screws cannot be installed smoothly during subsequent assembly, requiring rework.

[0004] The quality inspection and equipment maintenance processes also have shortcomings: traditional quality inspection uses manual handheld calipers to measure hole diameter and position, taking more than 30 seconds per hole, which is inefficient and prone to omissions, especially for openings on the back of the frame or in hidden locations, where inspection is even more difficult; equipment failure handling relies on on-site engineers, and if problems such as CNC platform deviation or sensor failure occur, it is necessary to wait for engineers to arrive for repairs, with downtime often exceeding half a day, affecting production progress; safety protection measures are simplistic, with only an emergency stop button and a lack of active protection devices such as infrared gratings, which can easily lead to safety accidents if operators accidentally touch dangerous areas; at the same time, production data is not systematically managed, and data such as hole parameters and quality results are only recorded on paper, making it impossible to trace the production information of each batch of frames, making it difficult to troubleshoot the causes of subsequent quality problems, and hindering the optimization of production processes. Summary of the Invention

[0005] The present invention proposes a machine vision-based intelligent opening system and method for thermally broken aluminum window and door frames to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent opening system for thermally broken aluminum window and door frames based on machine vision, comprising the following modules: The machine vision acquisition module consists of 6 industrial CCD cameras and a 360° ring light source. The cameras are arranged in six directions: up, down, left, right, front, and back of the frame. The light source intensity can be automatically switched according to the reflectivity of the frame surface to acquire information on the three-dimensional contour of the frame, the thickness of the profile wall, the position of the reinforcing ribs, and surface defects. The frame positioning and calibration module is equipped with a feature point recognition algorithm. It automatically extracts feature points such as frame edge corners, stiffener endpoints, and preset reference holes, establishes a mapping relationship between the image coordinate system and the world coordinate system, obtains the actual size of the frame through a laser range sensor, and performs deviation calibration by comparing the size with the design drawings. The hole opening parameter planning module receives customer requirement parameters and frame profile parameters, combines them with the actual frame structure acquired by machine vision, automatically avoids reinforcing ribs and weak areas of the cavity, plans the hole opening path, and generates CNC code containing hole opening coordinates, tool speed, and feed speed. The intelligent execution module consists of a three-axis CNC platform, an automatic tool changer, and a torque sensor. The CNC platform drives the tool to move according to the CNC code generated by the hole opening parameter planning module, and the torque sensor monitors the cutting torque in real time during the hole opening process. The quality inspection module, after the hole is opened, acquires images of the hole area a second time through the machine vision acquisition module, analyzes the hole diameter deviation, hole position offset, and hole wall roughness, and obtains the actual hole diameter data by combining the laser diameter measuring instrument, and compares it with the design parameters to determine the hole quality level. The data storage and interaction module encrypts and stores the framework design drawings, visually acquired images, hole parameters, and quality inspection results to the local server and cloud database. It supports viewing the data via touch screen and mobile APP, and can also receive remote control commands to adjust the hole parameters.

[0007] Furthermore, it also includes a hole diameter thermal deformation compensation module. This module, before the hole parameter planning module generates CNC code, combines the thermal expansion characteristics of the thermally broken aluminum profile with the compensation differences for different wall thicknesses, and uses a formula... Calculate the compensated opening diameter ,in The compensated opening diameter, To design the opening diameter, Basic compensation amount, The coefficient of linear expansion of thermally broken aluminum alloy is denoted as . This represents the temperature change during the hole-opening process. After calculating the compensation diameter, this module also calls the historical compensation database to compare the past compensation effects of similar profiles and hole diameters, and makes fine adjustments. The numerical value is also specifically verified during the quality inspection phase to match the aperture after cooling. The deviation.

[0008] Furthermore, it also includes a hole path optimization module. After the hole parameter planning module generates the initial path, the module identifies the coordinates of obstacles such as pre-installed holes and scratches on the frame surface, and then uses a formula to optimize the path. Calculate the optimal aperture path length ,in To determine the optimal path length, The path start time. This is the end time of the path. , , They are respectively The module measures the movement speed of the CNC platform in the x, y, and z axes, with the integral term reflecting the cumulative magnitude of the velocity vector along the entire path. This module groups multiple holes by coordinates, grouping holes on the same horizontal line together. Continuous hole opening within a group reduces the number of Z-axis rises and falls, and the path between different groups avoids obstacles. At the same time, the path turning angle is limited to ≤45° to reduce positioning errors. Actual production verification shows that the hole opening time for a single frame can be shortened by 15%-20%, while the hole position offset is reduced by 8%-12%.

[0009] Furthermore, it also includes a multi-specification frame adaptation module. This module is equipped with an AI model for model recognition. Through machine learning training, it learns the cross-sectional features of mainstream thermally broken aluminum frames. After the machine vision acquisition module acquires the cross-sectional image of the frame, the model recognition accuracy reaches over 99%, and it automatically calls the corresponding specification profile database. It adjusts the opening logic based on customer requirements parameters: the 55 series frame has a thinner wall thickness, the 70 series frame has more cavities, so it prioritizes opening in the cavity interval area, and the 80 series frame has dense stiffeners, so it automatically calculates the safe opening distance between the stiffeners. The module supports manual input of new specification frame parameters, and automatically generates an adapted opening parameter template after input. Production can be switched without changing hardware. It also has a parameter self-learning function. After 50 consecutive openings of the same specification frame, it automatically adjusts the optimal tool speed and feed rate.

[0010] Furthermore, it also includes a tool wear monitoring module, which integrates three data sources: torque monitoring, cutting edge visual inspection, and hole diameter deviation analysis. The module collects cutting torque through a torque sensor and calculates the average torque of five consecutive holes. It captures images of the tool cutting edge using an industrial camera and analyzes the degree of edge dulling. Combined with the hole diameter deviation data from the quality inspection module, a tool wear assessment model is established. The module records the number of holes opened, usage time, and total amount of profile processed for each tool, generating a tool life prediction curve to support advance stocking. It also automatically performs trial cutting calibration after replacing a tool.

[0011] Furthermore, it includes a safety protection module that constructs a multi-level protection system: Level 1 protection is an infrared light grid, which immediately sends a stop command when personnel or foreign objects enter the danger zone; Level 2 protection is area alert, which displays a yellow warning on the touchscreen and issues an audible and visual alert when an object appears in the alert zone; Level 3 protection is emergency stop control, with three emergency stop buttons around the equipment that cut off the main power supply to the equipment upon triggering, while saving the current opening data; the module monitors the equipment's operating parameters in real time; the equipment automatically performs a safety check before starting, and can only enter the operating mode after passing the check, while also storing the protection trigger records for the past 3 months.

[0012] Furthermore, it includes a remote operation and maintenance module, which establishes a real-time connection between the device and the cloud platform via 4G / 5G networks. The cloud platform is equipped with an AI fault diagnosis model that can identify more than 20 common faults. When a device malfunctions, the platform pushes the cause of the fault and repair steps to the operation and maintenance personnel's mobile APP within 10 seconds, supporting remote parameter debugging. The module regularly generates device health reports, and pushes repair reminders when the score is below 60. It pushes the device maintenance plan monthly, and uploads the records to the cloud after maintenance is completed. It supports centralized management of multiple devices, and operation and maintenance personnel can view the online status and production progress of all devices through the platform, realizing remote monitoring and batch scheduling.

[0013] Furthermore, a method for intelligent opening of thermally broken aluminum window and door frames based on machine vision is proposed, including the following steps: Step 1: Frame loading and vision acquisition. Place the thermally broken aluminum window frame on the fixture of the opening platform, start the fixture to fix the frame, start the machine vision acquisition module, and six industrial CCD cameras acquire images of the six sides of the frame in batches under the assistance of 360° ring light source. After Gaussian filtering and noise reduction and edge enhancement preprocessing, the images are transmitted to the frame positioning and calibration module. Step 2: Frame positioning and calibration. The frame positioning and calibration module extracts the corner points of the frame border, the endpoints of the reinforcing ribs, and the center of the preset reference holes in the image through the feature point recognition algorithm, establishes the image coordinate system, and uses the laser range sensor to measure the actual length, width and height of the frame. The positioning deviation is calculated by comparing it with the theoretical dimensions in the design drawings, and a positioning calibration matrix is ​​generated. Step 3: Hole opening parameter planning. The staff inputs the customer's required parameters through the touch screen. The hole opening parameter planning module calls the frame profile parameters collected by machine vision to plan the hole opening path, calls the hole diameter thermal deformation compensation module to calculate the compensated hole diameter, and sets the tool speed and initial feed speed in combination with the profile material density to generate CNC code and preview the hole opening path. Step 4: Intelligent hole drilling execution. The intelligent execution module receives the CNC code, and the three-axis CNC platform drives the tool to move to the first hole position. The automatic tool changer selects the appropriate tool, starts the tool to rotate to the set speed, and the torque sensor monitors the cutting torque in real time. When the torque exceeds the preset threshold, the feed speed is reduced. After completing one hole position, the tool returns to a safe height and moves to the next hole position until all holes are completed. Step 5: Quality Inspection and Feedback. The quality inspection module activates the machine vision acquisition module to acquire images of the opening area for the second time, analyzes the opening diameter deviation, hole position offset, and hole wall roughness, and compares the actual hole diameter measured by the laser diameter measuring instrument with the design parameters to determine the quality level. If it is qualified, the data is recorded; if it is unqualified, the correction suggestions are pushed to the intelligent execution module for secondary correction. Step Six: Data Storage and Update. The data storage and interaction module encrypts and stores the framework design drawings, visual acquisition images, opening parameters, and quality inspection results to the local server and cloud database. It generates production reports in Excel format according to the production date and uses the data as a sample to update the algorithm model of the opening parameter planning module.

[0014] Furthermore, it also includes a dynamic compensation step. During the intelligent drilling process in step four, a temperature sensor collects the temperature of the cutting area of ​​the tool and the non-cutting area of ​​the frame every 0.5 seconds to calculate ΔT, and a torque sensor collects the real-time torque value every 0.3 seconds. Combining the cutting edge wear data from the tool wear monitoring module, and using the formula... Adjust feed rate ,in The compensated feed rate, The initial feed rate, The initial torque threshold for the cutting tool. Here, β is the real-time torque value, ΔT is the temperature influence coefficient, and ΔT is the real-time temperature change. This step fine-tunes the compensation coefficient according to the hole-opening stage: β is set to 0.006 in the initial stage of hole opening to enhance temperature compensation; β is restored to 0.005 in the middle stage of hole opening; and torque compensation weight is increased in the later stage of hole opening. The weighting is increased to 0.6; simultaneously, the diameter of each hole is measured in real time every 5 holes using a laser diameter gauge, and compared with the compensated theoretical diameter. If the deviation exceeds 0.01mm, it is corrected immediately. .

[0015] Furthermore, this includes a multi-batch production adaptation step. Before the frame is loaded in step one, the system uses the data storage and interaction module to generate production data, statistically analyze common opening parameters, quality pass rate, and average production time according to frame specifications, and generate multi-batch production templates. When more than 10 frames of the same specification are loaded consecutively, the system automatically identifies the frame specifications and loads the corresponding template. After the operator confirms that the frame has no obvious defects through the machine vision acquisition module, they can directly proceed to step two, positioning and calibration. The template supports hierarchical editing permissions: administrators can modify all parameters, while operators can only adjust non-core parameters. After each batch of production is completed, the system automatically calculates the pass rate and reasons for non-conformity of the batch, generates a batch quality analysis report, and marks the direction for improvement. At the same time, each frame is assigned a unique QR code, which can be scanned to query the batch to which the frame belongs, opening parameters, quality inspection results, and operator.

[0016] Compared with existing technologies, the beneficial effects of this invention are: Positioning accuracy and opening rationality are significantly improved, reducing scrap rate. The machine vision acquisition module uses a six-directional camera and a ring light source to fully capture the three-dimensional contour of the frame, the distribution of internal reinforcing ribs, and surface defects. Combined with a laser rangefinder, it achieves millimeter-level positioning calibration, avoiding frame cracking caused by manual positioning errors. The opening parameter planning module automatically avoids reinforcing ribs and weak areas, while reserving expansion space through a thermal deformation compensation module. After cooling, the hole diameter accuracy meets design requirements, eliminating the need for secondary hole enlargement, reducing material waste and process time. The multi-specification frame adaptation module uses AI to identify the frame model and automatically call the corresponding parameter template. When changing to different series of frames, manual debugging is not required, allowing for rapid production switching to meet the needs of multiple batches and small production runs, improving equipment utilization.

[0017] Production efficiency and quality stability are significantly optimized, reducing labor costs. The hole-opening path optimization module reduces idle travel time through optimal path calculation, while grouping hole opening reduces the number of CNC platform lifts, resulting in a significant improvement in single-frame hole opening efficiency. The tool wear monitoring module integrates torque, vision, and quality data to provide early warnings of tool dulling conditions, preventing excessive hole wall roughness and reducing rework. The quality inspection module automatically completes hole diameter, hole position, and roughness inspection using machine vision and laser diameter measuring instruments, eliminating the need for manual intervention, greatly improving inspection efficiency, reducing the missed inspection rate, and ensuring consistent hole opening quality for each batch of frames. Multi-batch production adaptation steps generate templates, eliminating the need to repeatedly input parameters when continuously producing frames of the same specifications, further shortening production preparation time.

[0018] The system comprehensively enhances equipment safety and ease of operation and maintenance, ensuring production continuity. The safety protection module constructs a multi-level protection system encompassing infrared gratings, area alerts, and emergency stop controls, proactively preventing personnel from accidentally touching dangerous areas and reducing the risk of accidents. The remote operation and maintenance module uploads equipment data in real time through a cloud platform, and AI fault diagnosis quickly identifies problem types. Most software faults can be remotely debugged and resolved without waiting for on-site engineers, reducing downtime. Simultaneously, the data storage and interaction module enables systematic management of production data, making frame opening parameters and quality results traceable. This allows for rapid troubleshooting of subsequent problems, facilitating production process optimization. Overall, the system achieves intelligent, automated, and precise opening of thermally broken aluminum window and door frames, improving industry processing levels and providing strong support for cost reduction and efficiency improvement for enterprises. Attached Figure Description

[0019] Figure 1 This is a schematic block diagram of the intelligent opening system for thermally broken aluminum window and door frames based on machine vision proposed in this invention. Figure 2 This is a schematic block diagram of the intelligent hole-opening method for thermally broken aluminum window and door frames based on machine vision proposed in this invention. Figure 3 A comparison chart of frame positioning errors using different positioning methods; Figure 4 This is a graph showing the relationship between tool wear warning and the remaining number of holes to be drilled. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0023] Reference Figures 1 to 4 A machine vision-based intelligent opening system for thermally broken aluminum window and door frames includes the following modules: The machine vision acquisition module consists of 6 industrial CCD cameras and a 360° ring light source. The cameras are arranged in six directions: up, down, left, right, front, and back of the frame. Each camera has an image resolution of no less than 2048×1536 and a frame rate of no less than 30fps. The color temperature of the ring light source can be adjusted from 3000K to 6500K. The light source intensity can be automatically switched according to the reflectivity of the frame surface to acquire information on the three-dimensional contour of the frame, the thickness of the profile wall, the position of the reinforcing ribs, and surface defects. The frame positioning and calibration module is equipped with a feature point recognition algorithm. It automatically extracts feature points such as frame edge corners, stiffener endpoints, and preset reference holes, establishes a mapping relationship between the image coordinate system and the world coordinate system, obtains the actual size of the frame through a laser rangefinder (measurement accuracy ±0.01mm), compares the size with the design drawings to perform deviation calibration, and generates a positioning calibration matrix to ensure that the positioning error of the frame on the opening platform is ≤0.05mm. The hole parameter planning module receives customer requirement parameters (hole type such as screw hole, drainage hole, lock hole, hole diameter, hole spacing, hole depth) and frame profile parameters (wall thickness, cavity structure, material density). Combined with the actual frame structure collected by machine vision, it automatically avoids reinforcing ribs and weak areas of the cavity, plans the hole path, and generates CNC code containing hole coordinates, tool speed, and feed rate. The tool speed and feed rate are dynamically adjusted according to the profile thickness. The intelligent execution module consists of a three-axis CNC platform (positioning accuracy ±0.02mm, repeatability ±0.01mm), an automatic tool changer (capable of loading 8 types of tools including drills and milling cutters), and a torque sensor. The CNC platform drives the tool movement according to the CNC code generated by the hole-opening parameter planning module. The torque sensor monitors the cutting torque during the hole-opening process in real time (monitoring range 0-50N・m, accuracy ±0.1N・m). When the torque exceeds the preset threshold, the feed rate is automatically reduced to avoid tool damage or frame deformation. The quality inspection module, after the hole is drilled, acquires images of the drilled area a second time through the machine vision acquisition module, analyzes the hole diameter deviation, hole position offset, and hole wall roughness, and obtains the actual hole diameter data by combining a laser diameter gauge (measuring range 0.5-50mm, accuracy ±0.005mm), and compares it with the design parameters to determine the hole quality level (qualified, needs correction, unqualified). The data storage and interaction module encrypts and stores the framework design drawings, visually acquired images, opening parameters, and quality inspection results to a local server (storage capacity ≥ 2TB) and a cloud database. It supports viewing data via a touch screen (10.1-inch, resolution 1920×1200) and a mobile APP, and can export production reports in Excel format. It can also receive remote control commands to adjust the opening parameters.

[0024] This invention also includes a hole diameter thermal deformation compensation module. Before the hole parameter planning module generates the CNC code, this module, considering the thermal expansion characteristics of the thermally broken aluminum profile and the compensation differences for different wall thicknesses, uses a formula... Calculate the compensated opening diameter ,in The compensated opening diameter (in mm). For designing the opening diameter (unit: mm). Basic compensation amount (unit: mm, wall thickness: 3 mm) =0.02, wall thickness 5mm =0.03, wall thickness 8mm =0.04 (set according to profile wall thickness gradient). The coefficient of linear expansion of thermally broken aluminum (unit: °C⁻) 1 The value is 2.3 × 10⁻ 5 ), The temperature change during the hole-opening process (in °C) is measured by two temperature sensors, which collect the temperatures of the cutting area of ​​the tool and the non-cutting area of ​​the frame, respectively. The difference between the two is taken as the effective value. (Sampling frequency 1Hz); After calculating the compensation diameter, this module also calls the historical compensation database to compare the past compensation effects of similar materials and apertures, and makes fine adjustments. The numerical value is also specifically verified during the quality inspection phase to match the aperture after cooling. The deviation is automatically updated when it exceeds 0.01mm to ensure that the final opening size accuracy meets the design requirements. After testing and cooling, the hole diameter deviation can be controlled within ±0.008mm.

[0025] This invention also includes an opening path optimization module. After the opening parameter planning module generates the initial path, this module identifies the coordinates of obstacles such as pre-installed holes and scratches on the frame surface, and then uses a formula to optimize the path. Calculate the optimal aperture path length ,in The optimal path length (in mm). The path start time (in seconds). The path end time (in seconds). , , They are respectively The module measures the movement speed (in mm / s) of the CNC platform along the x, y, and z axes, with the integral term reflecting the cumulative magnitude of the velocity vector throughout the path. It groups multiple holes by coordinates, grouping holes on the same horizontal line together. Continuous hole opening within a group reduces the number of Z-axis rises and falls, and the path between different groups avoids obstacles. Simultaneously, it limits the path turning angle to ≤45° to reduce positioning errors. After optimization, it also calculates the path positioning accuracy (by simulating path operation and statistically analyzing the deviation between the theoretical and actual coordinates of each hole), ensuring a deviation ≤0.02mm. Actual production verification shows that the hole opening time for a single frame can be reduced by 15%-20%, while the hole offset is reduced by 8%-12%.

[0026] This invention also includes a multi-specification frame adaptation module. This module is equipped with an AI model recognition model, which is trained through machine learning to recognize the cross-sectional features (such as the number of cavities, stiffener spacing, and frame width) of mainstream thermally broken aluminum frames such as the 55 series, 60 series, 70 series, and 80 series. After the machine vision acquisition module acquires the frame cross-sectional image, the model recognition accuracy reaches over 99%. It automatically calls the corresponding profile database (stores the wall thickness distribution, cavity load-bearing capacity, and weak area location of different series). Based on customer requirements, the opening logic is adjusted: for the 55 series frame with a thinner wall thickness (≤1.4mm), the feed speed is reduced. 10% of the time, and tools with a diameter ≤5mm are selected. For the 70 series frame with more cavities (≥3), the cavity interval area (spacing ≥10mm) is selected for drilling. For the 80 series frame with dense reinforcing ribs (spacing ≤8mm), the safe drilling distance between ribs is automatically calculated (≥3mm). The module supports manual input of new frame specifications (such as custom wall thickness and cavity structure). After input, the module automatically generates a matching drilling parameter template, and production can be switched without changing hardware. It also has a parameter self-learning function. After drilling 50 times in a row for the same frame specification, the optimal tool speed and feed rate are automatically adjusted to improve drilling efficiency by another 5%.

[0027] This invention also includes a tool wear monitoring module, which integrates three data sources: torque monitoring, cutting edge visual inspection, and hole diameter deviation analysis. The module collects cutting torque via a torque sensor (recording once every 0.5 seconds) and calculates the average torque for five consecutive holes. It also captures images of the tool cutting edge using an industrial camera (taking one image every 20 holes) and analyzes the degree of edge dulling (wear ≥ 0.02 mm is considered mild dulling). Combined with hole diameter deviation data from the quality inspection module, a tool wear assessment model is established. When the average torque increases by 15% from the initial value or the cutting edge is mildly dulled, and the hole diameter deviation is ≤ 0.03 mm, the tool is judged to have mild wear, and a yellow indicator is displayed on the touchscreen. The system provides color-coded reminders and indicates the recommended replacement time (remaining drilling cycles ≤ 50). When the average torque increases by 30% or the cutting edge wear is ≥ 0.05mm, and the hole diameter deviation exceeds 0.05mm, the system determines that the tool is severely worn. The system automatically pauses the drilling process, locks the tool drive unit, and displays a red replacement prompt. The module records the number of drilling cycles, usage time, and total amount of profile processed for each tool, and generates a tool life prediction curve (e.g., high-speed steel drill bits have a lifespan of approximately 800 drilling cycles, and carbide drill bits approximately 1500 cycles). It supports advance stocking and automatically performs trial cutting calibration after replacing the tool (test cutting one standard hole to calibrate the tool coordinate system) to ensure the drilling accuracy of the new tool.

[0028] This invention also includes a safety protection module, which constructs a multi-level protection system: Level 1 protection is an infrared grating (protection height 1.2m, response time ≤10ms, covering a 50cm danger zone around the tool). When personnel or foreign objects enter the danger zone, a pause command is immediately sent, the tool stops rotating within 0.1s, and the CNC platform is locked; Level 2 protection is a zone warning (a warning zone is set 50cm outside the danger zone). When an object appears in the warning zone, the touchscreen displays a yellow warning and emits an 80dB audible and visual alert (the light flashes yellow); Level 3 protection is an emergency stop control, with 3 emergency stop buttons (spaced ≤2m) around the equipment. Triggering these buttons cuts off the main power supply to the equipment. In addition to the control system, it also saves the current drilling data; the module monitors the equipment operating parameters in real time: when the current (monitoring range 0-50A, accuracy ±0.1A) exceeds the rated value by 1.2 times, it automatically reduces the drive power to 80%; when the tool speed exceeds the set value by 10%, it immediately reduces the speed to a safe range; when the frame fixation is loose (the pressure of the fixing fixture is detected by the pressure sensor, and the pressure ≤50N is judged as loose), the drilling is paused and a prompt is made to re-fix it; before the equipment starts, it automatically performs a safety check (checking whether the grating, emergency stop, and current are normal), and can only enter the operation mode after the check is passed. At the same time, it stores the protection trigger records of the past 3 months to facilitate the analysis of safety hazards.

[0029] This invention also includes a remote operation and maintenance module. This module establishes a real-time connection between the device and the cloud platform via a 4G / 5G network, uploading device operating data (tool speed, feed rate, torque value, current, temperature) every 1 second, and visually acquired images and quality inspection results every 5 minutes. The cloud platform is equipped with an AI fault diagnosis model that can identify more than 20 common faults (such as tool jamming, camera defocusing, CNC platform deviation, and sensor failure), with a fault identification accuracy rate of over 95%. When a fault occurs, the platform pushes the cause of the fault (such as "sudden increase in torque: possible tool breakage") and repair steps (with disassembly instructions) within 10 seconds. The system allows for remote parameter debugging (such as correcting the positioning calibration matrix and resetting the torque threshold via the app) to maintenance personnel's mobile app. Over 80% of software-related faults can be resolved remotely. The module regularly generates equipment health reports (scored from three aspects: operational stability, component wear, and energy consumption, with a maximum score of 100). When the score is below 60, a maintenance reminder is pushed. Monthly equipment maintenance plans are pushed (such as lubricating CNC guideways, cleaning camera lenses, and calibrating laser diameter gauges). After maintenance is completed, the records are uploaded to the cloud. The system supports centralized management of multiple devices. Maintenance personnel can view the online status and production progress of all devices through the platform, enabling remote monitoring and batch scheduling.

[0030] This invention also proposes a method for intelligent opening of thermally broken aluminum window and door frames based on machine vision, including the following steps: Step 1: Frame loading and vision acquisition. Place the thermally broken aluminum window and door frame on the fixture of the opening platform, start the fixture to fix the frame (the fixture pressure is set to 80-120N, adjusted according to the frame wall thickness), start the machine vision acquisition module, and six industrial CCD cameras acquire images of the six sides of the frame in batches under the assistance of a 360° ring light source (3 images are acquired for each side, and clear images are selected). After the images are preprocessed by Gaussian filtering for noise reduction and edge enhancement, they are transmitted to the frame positioning and calibration module. Step 2: Frame positioning and calibration. The frame positioning and calibration module extracts the corner points of the frame border, the endpoints of the reinforcing ribs, and the center of the preset reference holes from the image using a feature point recognition algorithm. It establishes an image coordinate system and measures the actual length, width, and height of the frame using a laser rangefinder (measures 3 points in each dimension and takes the average value). It then calculates the positioning deviation (such as X-axis deviation, Y-axis deviation, and rotation deviation) by comparing it with the theoretical dimensions in the design drawings. This generates a positioning calibration matrix, drives the platform to fine-tune the frame position, and ensures that the positioning error is ≤0.05mm. After calibration, the platform is locked. Step 3: Hole Opening Parameter Planning. Staff input customer requirements (hole type, hole diameter, hole spacing, hole depth) via a touchscreen. The hole opening parameter planning module calls upon frame profile parameters acquired by machine vision (wall thickness, cavity structure, reinforcing rib location), automatically marks areas where holes cannot be opened (reinforcing rib center, weak wall of the cavity), plans the hole opening path (sorted by hole position coordinates), and calls the hole diameter thermal deformation compensation module to calculate the compensated hole diameter. The tool speed is then set based on the profile material density (aluminum alloy density 2.7g / cm³). 3 At that time, the rotational speed is set to 2000-3000 r / min and the initial feed rate is set to 10-20 mm / min. The CNC code is generated and the hole opening path is previewed. Step 4: Intelligent Drilling Execution. The intelligent execution module receives the CNC code, and the three-axis CNC platform drives the tool to move to the first drilling position. The automatic tool changer selects the appropriate tool according to the drilling type (twist drill for screw holes, end mill for lock holes). The tool is started to rotate to the set speed, and the torque sensor monitors the cutting torque in real time. When the torque exceeds the preset threshold (e.g., the torque threshold for high-speed steel drill bits is set to 8-12 N·m), the feed rate is reduced by 10%-20%. After drilling one hole, the tool returns to a safe height (10 mm above the frame surface) and moves to the next hole until all holes are drilled. Step 5: Quality Inspection and Feedback. The quality inspection module activates the machine vision acquisition module to acquire images of the opening area a second time (two images are acquired for each hole, one for the hole diameter and one for the hole wall). The module analyzes the hole diameter deviation (actual hole diameter - designed hole diameter), hole position offset (actual hole center - theoretical hole center distance), and hole wall roughness (analyzed by image grayscale values). The actual hole diameter is measured using a laser diameter gauge (measured in four directions for each hole, and the average value is taken). The quality level is determined by comparing the measurement with the design parameters (deviation ≤ 0.02mm is acceptable, 0.02-0.05mm requires correction, and > 0.05mm is unacceptable). If the quality is acceptable, the quality data is recorded. If the quality is unacceptable, correction suggestions (such as re-enlarging the hole if the hole diameter is too small) are pushed to the intelligent execution module for secondary correction. Step Six: Data Storage and Update. The data storage and interaction module encrypts and stores the framework design drawings, original and processed images acquired by vision, hole opening parameters (NC code, compensation parameters), and quality inspection results (grade, deviation data) to the local server and cloud database. It generates production reports in Excel format (including production quantity, pass rate, and number of failures) according to the production date. At the same time, it uses the hole opening data as a sample to update the algorithm model of the hole opening parameter planning module (adjusting path optimization weights and compensation coefficients) to improve the accuracy and efficiency of subsequent hole openings.

[0031] This invention also includes a dynamic compensation step. During the intelligent drilling process in step four, a temperature sensor collects the temperatures of the cutting area and the non-cutting area of ​​the frame every 0.5 seconds to calculate ΔT. A torque sensor collects the real-time torque value τreal every 0.3 seconds. Combined with the cutting edge wear data from the tool wear monitoring module, the results are calculated using the formula... Adjust feed rate ,in The feed rate after compensation (unit: mm / s). Initial feed rate (unit: mm / s). The initial torque threshold for the tool (in N·m, set according to the tool type, for carbide tools) =15N・m), β is the real-time torque value (in N·m), and β is the temperature influence coefficient (in °C⁻). 1 ΔT is the real-time temperature change (unit: °C), and is set to 0.005. This step fine-tunes the compensation coefficient according to the hole-opening stage: In the initial stage (before tool preheating, for the first 3 holes), the β value is set to 0.006 to enhance temperature compensation; in the middle stage (4-20 holes, temperature stabilizes), the β value returns to 0.005; in the later stage (after 20 holes, tool may wear), the torque compensation weight is increased. The weighting is increased to 0.6; simultaneously, the diameter of each hole is measured in real time every 5 holes using a laser diameter gauge, and compared with the compensated theoretical diameter. If the deviation exceeds 0.01mm, it is corrected immediately. This ensures that the diameter deviation of the opening is ≤0.02mm throughout the entire process, thereby improving the stability of the opening quality.

[0032] This invention also includes a multi-batch production adaptation step. Before the frame loading in step one, this step retrieves historical production data from the past three months through the data storage and interaction module. Common opening parameters (hole diameter, hole position, tool type), quality pass rate, and average production time are statistically analyzed according to frame specifications (e.g., 55 series, 70 series). This generates multi-batch production templates (one template per specification, containing complete opening parameters and calibration data). When more than 10 frames of the same specification are continuously loaded, the system automatically identifies the frame specification and loads the corresponding template. The operator only needs to confirm through the machine vision acquisition module that the frame has no obvious defects (e.g., deformation, scratches exceeding 0.5mm) before proceeding directly to step two. Positioning and calibration eliminate the need to repeatedly input customer-required parameters; templates support tiered editing permissions: administrators can modify all parameters, while operators can only adjust non-core parameters such as feed rate and tool speed, preventing misoperation; after each batch of production is completed (e.g., 50 frames per batch), the system automatically calculates the batch's pass rate and reasons for non-conformity (e.g., hole diameter deviation, hole position offset), generates a batch quality analysis report, and indicates improvement directions (e.g., if the pass rate is below 95%, it is recommended to recalibrate the tool); simultaneously, each frame is assigned a unique QR code, which can be scanned to query the batch to which the frame belongs, opening parameters, quality inspection results, and operator, enabling full-process data traceability and facilitating quality problem investigation and responsibility determination.

[0033] The following two examples further illustrate the specific implementation of this system: Example 1: Mass production of openings for 55 series thermally broken aluminum window and door frames (ordinary residential project) I. Scenarios and Basic Information This embodiment is applied to a mass production line for 55-series thermally broken aluminum frames in a door and window processing plant. It undertakes orders for ordinary residential projects, requiring the processing of 1000 sets of 55-series frames (wall thickness 1.4mm, 2 cavities, 6mm stiffener spacing, single frame size 1200mm×800mm). The customer requires each frame to have 20 4mm screw holes (100mm spacing) and 4 8mm drainage holes (located at the bottom corners and center of the frame). The hole drilling accuracy requirements are: hole diameter deviation ≤ ±0.02mm, hole position offset ≤ ±0.05mm. Equipment configuration: 6 Hikvision industrial CCD cameras (resolution 2048×1536, frame rate 30fps), Keyence laser rangefinder (accuracy ±0.01mm), three-axis CNC platform (positioning accuracy ±0.02mm), 8-station automatic tool changer (equipped with 4mm / 8mm high-speed steel twist drills, end mills, etc.).

[0034] II. Detailed Implementation Process 1. Frame loading and visual acquisition (Step 1 detailed) The 55 series frame was placed on the pneumatic fixture of the opening platform, with the fixture pressure set to 80N (to avoid deformation of the thin-walled frame due to excessive pressure). The 360° ring light source was activated (the color temperature was adjusted to 4500K and the intensity to 60% based on the reflectivity of the anodized layer on the frame surface). Six CCD cameras were used to acquire images of the frame from six sides: the top camera (monitoring the top screw hole area), the bottom camera (focusing on the drainage hole location), the left / right side cameras (capturing the distribution of the side reinforcing ribs), and the front / back cameras (recording the overall outline and surface defects of the frame). Three images were acquired from each side. After denoising with Gaussian filtering (5×5 convolution kernel) and edge enhancement with Sobel operator, the images were transmitted to the frame positioning and calibration module. The image clarity reached over 98%, with no feature loss due to reflection.

[0035] 2. Frame positioning and calibration (step two detailed) The frame positioning and calibration module uses a feature point recognition algorithm to automatically extract the four corner points of the frame, the eight reinforcing rib endpoints, and the two preset reference holes. A laser rangefinder measures the actual dimensions of the frame: length 1200.03mm, width 800.02mm, and calculates the X-axis deviation +0.03mm, Y-axis deviation +0.02mm, and rotation deviation 0.01°. A positioning calibration matrix is ​​then generated. The CNC platform is driven to fine-tune the frame position. After calibration, the positioning error is measured again and found to be 0.04mm, which meets the requirement of ≤0.05mm. The platform is then locked to prevent displacement.

[0036] 3. Hole opening parameter planning and thermal deformation compensation (step three refinement) Staff input customer requirements via a 10.1-inch touchscreen: 20 screw holes (4mm each), 4 drainage holes (8mm each), and a hole depth of 1.2mm (not penetrating the cavity). The hole parameter planning module calls upon profile data acquired by machine vision (wall thickness 1.4mm, reinforcing rib diameter 2mm), automatically marks areas where holes cannot be drilled (within ±1.5mm of the reinforcing rib center), and plans the initial path (in the order of "top screw holes → bottom drainage holes"). The hole diameter thermal deformation compensation module is then activated, and the formula is substituted... : Screw hole: D=4mm, ΔD=0.02mm (for 55 series with a wall thickness of 1.4mm, corresponding to foundation compensation), α=2.3×10⁻ 5 ℃⁻ 1 ΔT = 5℃ (temperature sensor collects data: cutting area 35℃, non-cutting area 30℃, difference 5℃), calculated as follows: ; Drain hole: D=8mm, ΔD=0.03mm, ΔT=5℃, calculated as follows Combined with the 55 series aluminum alloy density of 2.7 g / cm³ 3 Set the tool rotation speed: 2500r / min for 4mm twist drill and 2000r / min for 8mm twist drill, initial feed rate 12mm / min, generate G-code format CNC code, and preview the path to ensure there are no intersections or obstacles.

[0037] 4. Intelligent hole opening execution and path optimization (step four refinement) The intelligent execution module receives CNC code, the automatic tool changer selects a 4mm high-speed steel twist drill, and the three-axis CNC platform drives the tool to move to the first screw hole position (coordinates X=100mm, Y=100mm, Z=0); the tool rotation is started, and the torque sensor monitors the cutting torque in real time (initial threshold 8N・m). During the hole-opening process, the torque stabilizes at 7.5-7.8N・m, not exceeding the threshold; simultaneously, the hole-opening path optimization module optimizes the hole according to the formula... Optimized path: Divide the 20 screw holes into 4 groups according to the Y-axis coordinate (5 holes in each group, on the same horizontal line), and make continuous holes within the group (only raise and lower once on the Z-axis). The moving path between groups avoids the reinforcing rib, and the turning angle of the path is 30° (≤45°). The path length before optimization was 1800mm, and after optimization it was 1500mm. The idle travel time was reduced from 20s to 15s. After completing all screw holes, the tool was automatically changed to an 8mm twist drill, and the drainage holes were opened according to the optimized path. The total time for opening holes in a single frame was reduced from the traditional 3min to 2.4min, improving efficiency by 20%.

[0038] 5. Quality Inspection and Tool Wear Monitoring (Step 5 in detail) The quality inspection module initiated secondary machine vision acquisition: two images were captured for each screw hole (front hole diameter and side hole wall). A laser diameter gauge measured the hole diameter (measured in four directions for each hole, and the average value was taken). The results showed that the actual screw hole diameter was 4.0200mm (deviation -0.000023mm), the drainage hole diameter was 8.0300mm (deviation -0.0000345mm), the hole position offset was 0.03mm, and the hole wall roughness Ra≤1.6μm, thus meeting the acceptable standard. The tool wear monitoring module recorded: after continuously drilling 50 frames, 4m... The average torque of the twist drill increased to 9.2 N·m (a 15% increase from the initial 8 N·m). The visual inspection of the cutting edge showed a wear of 0.02 mm, indicating slight wear. The touchscreen displayed a yellow warning: "50 holes remain to be drilled; stocking up is recommended." After drilling 100 frames, the torque increased to 10.4 N·m (a 30% increase), triggering a severe wear alarm. The system paused drilling, replaced the tool, and automatically test-cut a 4 mm standard hole (diameter 4.0201 mm, deviation 0.000077 mm). Production resumed after the tool coordinate system was calibrated.

[0039] 6. Data storage and production reports (Step 6 in detail) The data storage and interaction module encrypts and stores 1,000 sets of frame design drawings (CAD format), visual images (JPG format), hole opening parameters (CNC code, compensation value), and quality results (Excel spreadsheet) on a local 2TB server and Alibaba Cloud, generating production reports: the overall pass rate is 99.5% (5 sets were unqualified due to initial tool adjustment deviations), and the average hole opening time per frame is 2.4 minutes, which is 30% shorter than traditional manual auxiliary equipment; the reports can be exported to the workshop management system for project progress tracking and process optimization.

[0040] III. Effect Verification and Table Analysis Table 1: Performance Comparison of Traditional Hole-Opening Equipment and the System of the Invention for 55 Series Frames

[0041] Table 1 shows that traditional equipment relies on manual positioning and inspection, resulting in large positioning errors that easily lead to frame failure due to holes touching reinforcing ribs, with a scrap rate exceeding 8%. This system, through six-directional visual acquisition and laser calibration, controls the positioning error within 0.05mm. Combined with thermal deformation compensation and path optimization, hole diameter deviation and hole opening time are significantly reduced, and the scrap rate is reduced to below 1%. For example, when traditional equipment processes 55 series thin-walled frames, due to the lack of thermal deformation compensation, the 4mm screw hole shrinks to 3.98mm after cooling, requiring secondary hole enlargement. This system calculates the compensation diameter in advance, and the hole diameter remains at 4.02mm after cooling, eliminating the need for rework. At the same time, it provides early warning of tool wear, avoiding rough hole walls due to dulling, significantly improving production efficiency and product quality.

[0042] Example 2: Mixed production of multi-specification thermally broken aluminum frames (high-end residential projects) I. Scenarios and Basic Information This embodiment is applied to a multi-specification production line of a custom door and window factory, undertaking high-end residential projects. It requires mixed processing of 55 series (500 sets, 1500mm×900mm) and 70 series (300 sets, 1.8mm wall thickness, 3 cavities, 1800mm×1000mm) frames. Customer requirements: 55 series requires 24 5mm screw holes and 4 8mm drainage holes; 70 series requires 30 6mm screw holes and 2 10mm lock holes. The requirement is a multi-specification switching time of ≤10 minutes, remote monitoring of equipment operation status, and full-process data traceability. Additional equipment configurations include: an AI model recognition module, a 4G / 5G remote operation and maintenance module, and a multi-batch template storage unit.

[0043] II. Detailed Implementation Process 1. Frame loading and multi-specification identification (Step 1 detailed) Workers alternately feed 55 series and 70 series frames. The pneumatic clamps automatically adjust the pressure according to the frame specifications: 80N for 55 series and 100N for 70 series (adjusting pressure as wall thickness increases). The machine vision acquisition module activates the AI ​​model for model recognition. By comparing the cross-sectional features of the frames (2 cavities for 55 series, 6mm spacing between reinforcing ribs; 3 cavities for 70 series, 8mm spacing), the recognition accuracy reaches 99.2%, completing model determination within 2 seconds and automatically calling the corresponding profile database (1.4mm wall thickness for 55 series, weak area is the side wall of the cavity; 1.8mm wall thickness for 70 series, the lock hole area must avoid the internal hardware mounting groove).

[0044] 2. Multi-batch template loading and parameter planning (step three refinement) The data storage and interaction module retrieves historical production data and generates multiple batch templates for the 55 series and 70 series. 55 series template: screw holes 5mm ( , ), Drain hole 8mm ( The tool rotation speed is 2800 r / min and the feed rate is 13 mm / min. 70 series template: screw holes 6mm ( , ), 10mm lock hole (end mill, speed 2200r / min, feed speed 10mm / min); when continuously feeding 10 55 series frames, the system automatically loads the 55 series template. The staff only needs to visually confirm that the frame is not deformed (surface scratches ≤0.3mm) and directly enter the positioning and calibration step. There is no need to repeatedly input parameters. The specification switching time is shortened from the traditional 1.5h to 8min.

[0045] 3. Intelligent opening and dynamic compensation (Step 4 + refinement of claim 9) When the intelligent execution module processes the 70 series frame lock holes, a dynamic compensation step is initiated: the temperature sensor collects data every 0.5 seconds at 38°C in the cutting area and 32°C in the non-cutting area (ΔT=6°C), and the torque sensor collects real-time torque data every 0.3 seconds. =12N·m (initial threshold) =15N・m), substituting into the formula : =10mm / min (initial feed rate of 70 series keyhole), β=0.005℃⁻ 1 Calculated ; During the middle of the hole opening process (15th screw hole), ΔT dropped to 4°C. Increased to 13.5 N·m, recalculated. After dynamic adjustment, the diameter deviation of the 70 series lock holes is stable at ±0.015mm, and the hole walls are burr-free, meeting the requirements for high-end residential hardware installation.

[0046] 4. Remote operation and maintenance and fault handling (refined according to claim 7) When the equipment was processing the 200th 70-series frame, the torque suddenly increased to 20 N·m (exceeding the threshold of 15 N·m). The remote operation and maintenance module uploaded the data to the cloud platform in real time. AI fault diagnosis model analysis: Sudden increase in torque + 5% decrease in speed, determined to be "tool breakage", and push the fault cause to the maintenance personnel's mobile APP within 10 seconds, with repair steps ("turn off the power of the equipment → replace the 10mm end mill → test cut calibration"). Maintenance personnel remotely sent a "parameter reset command" to temporarily reduce the keyhole feed speed to 8mm / min. After instructing the on-site operator to replace the tool, they test-cut one keyhole (diameter 10.03mm, deviation 0.00mm), and the equipment resumed production with a downtime of only 15 minutes (traditionally it would take 2 hours). The cloud platform generated an equipment health report: operational stability 92 points, component wear 78 points (prompting "CNC guide rail needs lubrication"), and pushed the monthly maintenance plan to the workshop management APP.

[0047] 5. Quality Inspection and Data Traceability (Step 5 + Detailed Explanation of Claim 10) The quality inspection module tested the 55 / 70 series frames separately: the screw hole diameter of the 55 series was 5.025mm (deviation 0.00mm), and the lock hole diameter of the 70 series was 10.03mm (deviation 0.00mm), with a pass rate of 99.8%. Each frame was assigned a unique QR code, which could be scanned for further information. Production information: processing time, operator, equipment number; Technical parameters: Hole coordinates, compensation value, tool type; Quality data: hole diameter deviation, hole position offset, and inspection personnel; after multiple batches of production are completed, a quality analysis report is generated: 2 sets of 70 series lock holes are unqualified (due to the initial dynamic compensation parameters not being adapted, which has been optimized), and the subsequent production pass rate is 100%. The data can be exported to the project client to meet the traceability requirements of customized projects.

[0048] III. Effect Verification and Table Analysis Table 2: Performance Comparison of Traditional Multi-Specification Frame Production and the System of the Invention

[0049] Table 2 shows that traditional multi-specification production requires manual adjustment of equipment parameters, with a switchover time exceeding 1.5 hours, and the lack of template support leads to errors due to repeated parameter input. This system, through AI model recognition and multiple batch templates, reduces the switchover time to within 10 minutes. Remote maintenance solves the problem of delayed on-site repairs, significantly reducing downtime. For example, when processing 70-series keyholes, traditional equipment, lacking dynamic compensation for temperature and torque changes, results in a keyhole diameter deviation of up to 0.04mm, requiring manual grinding. This system adjusts the feed speed in real time, controlling the deviation within 0.015mm. Simultaneously, the full-process QR code traceability meets the stringent quality traceability requirements of high-end residential projects, improving customer satisfaction and helping companies secure high-end customized orders.

[0050] Reference Figure 3 This diagram clearly demonstrates the core advantage of the system in terms of positioning accuracy. Traditional manual auxiliary equipment relies on measuring tape markings and experience-based judgment, resulting in positioning errors typically ranging from 0.5 to 1.0 mm. This can easily lead to the opening position touching reinforcing ribs or weak walls of the cavity, causing frame cracking and scrapping (especially for the 55 series thin-walled frames, where the scrap rate exceeds 8%). The system of this invention uses a six-directional CCD camera to completely capture the frame outline, combined with the millimeter-level measurement accuracy of a laser rangefinder, to stably control the positioning error within 0.04-0.05 mm, far below the industry-required threshold of 0.1 mm. For example, in sample 3, the traditional equipment has a positioning error of 0.8 mm, causing the screw holes to deviate from the preset position, requiring re-drilling. The system's error is 0.05 mm, ensuring the opening position perfectly matches the design requirements, eliminating the need for rework, and significantly reducing material waste and production cycle. Reference Figure 4This diagram illustrates the forward-looking nature of the tool wear monitoring module. Traditional manual tool wear monitoring relies on visual inspection of the cutting edge, often only discovering wear after the tool has become severely dull. At this point, only 3-10 drilling cycles remain, and the machined holes may have excessive surface roughness (Ra > 3.2μm) due to tool dulling, requiring rework and grinding. The system of this invention, through torque monitoring, visual inspection of the cutting edge, and hole diameter deviation analysis, issues an early warning when the tool shows slight wear (around 400 drilling cycles), allowing 40-50 drilling cycles for replacement preparation time, ensuring the quality of the holes machined before replacement is up to standard. For example, in the sample, when 410 drilling cycles had been completed, the system warned of 45 remaining cycles, allowing staff to prepare a new tool in advance and replace it promptly at 450 cycles, resulting in no defective products. Traditional manual methods only detect wear at 450 cycles, by which time quality issues have already occurred in the previous 10 drilling cycles, leading to a rework rate exceeding 20%.

[0051] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A machine vision-based intelligent opening system for thermally broken aluminum window and door frames, characterized in that, Includes the following modules: The machine vision acquisition module consists of 6 industrial CCD cameras and a 360° ring light source. The cameras are arranged in six directions: up, down, left, right, front, and back of the frame. The light source intensity can be automatically switched according to the reflectivity of the frame surface to acquire information on the three-dimensional contour of the frame, the thickness of the profile wall, the position of the reinforcing ribs, and surface defects. The frame positioning and calibration module is equipped with a feature point recognition algorithm. It automatically extracts the feature points of the frame edge corners, stiffener endpoints, and preset reference holes, establishes the mapping relationship between the image coordinate system and the world coordinate system, obtains the actual size of the frame through a laser range sensor, and performs deviation calibration by comparing it with the size of the design drawings. The hole opening parameter planning module receives customer requirement parameters and frame profile parameters, combines them with the actual frame structure acquired by machine vision, automatically avoids reinforcing ribs and weak areas of the cavity, plans the hole opening path, and generates CNC code containing hole opening coordinates, tool speed, and feed speed. The intelligent execution module consists of a three-axis CNC platform, an automatic tool changer, and a torque sensor. The CNC platform drives the tool to move according to the CNC code generated by the hole opening parameter planning module, and the torque sensor monitors the cutting torque in real time during the hole opening process. The quality inspection module, after the hole is opened, acquires images of the hole area a second time through the machine vision acquisition module, analyzes the hole diameter deviation, hole position offset, and hole wall roughness, and obtains the actual hole diameter data by combining the laser diameter measuring instrument, and compares it with the design parameters to determine the hole quality level. The data storage and interaction module encrypts and stores the framework design drawings, visually acquired images, hole parameters, and quality inspection results to the local server and cloud database. It supports viewing the data via touch screen and mobile APP, and can also receive remote control commands to adjust the hole parameters.

2. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes a hole diameter thermal deformation compensation module. Before the hole parameter planning module generates the CNC code, this module combines the thermal expansion characteristics of the thermally broken aluminum profile with the compensation differences for different wall thicknesses, using a formula... Calculate the compensated opening diameter ,in The compensated opening diameter, To design the opening diameter, Basic compensation amount, The coefficient of linear expansion of thermally broken aluminum alloy is denoted as . This represents the temperature change during the hole-opening process. After calculating the compensation diameter, this module also calls the historical compensation database to compare the past compensation effects of similar profiles and hole diameters, and makes fine adjustments. The numerical value is also specifically verified during the quality inspection phase to match the aperture after cooling. The deviation.

3. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes an opening path optimization module. After the opening parameter planning module generates the initial path, the module identifies the coordinates of obstacles on the frame surface and then uses a formula to optimize the path. Calculate the optimal aperture path length ,in To determine the optimal path length, The path start time. This is the end time of the path. , , They are respectively The moving speed of the CNC platform in the x, y, and z axes at any time; the integral term reflects the cumulative magnitude of the velocity vector along the entire path. This module groups multiple holes by coordinates, with holes on the same horizontal line grouped together. Continuous hole opening within a group reduces the number of Z-axis lifting and lowering operations, and the path between different groups avoids obstacles. At the same time, the path turning angle is limited to ≤45° to reduce positioning error. Actual production verification shows that the hole opening time for a single frame can be shortened by 15%-20%, while the hole position offset is reduced by 8%-12%.

4. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes a multi-specification frame adaptation module, which is equipped with an AI model for model recognition. Through machine learning training, it learns the cross-sectional features of mainstream thermally broken aluminum frames. After the machine vision acquisition module acquires the cross-sectional image of the frame, the model recognition accuracy reaches over 99%, and it automatically calls the corresponding specification profile database. It adjusts the opening logic based on customer requirements parameters: the 55 series frame has a thinner wall thickness, the 70 series frame has more cavities, so it prioritizes opening in the cavity interval area, and the 80 series frame has dense stiffeners, so it automatically calculates the safe opening distance between the stiffeners. The module supports manual input of new specification frame parameters, and automatically generates an adapted opening parameter template after input. Production can be switched without changing hardware. It also has a parameter self-learning function. After 50 consecutive openings of the same specification frame, it automatically adjusts the optimal tool speed and feed rate.

5. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes a tool wear monitoring module, which integrates three data sources: torque monitoring, cutting edge visual inspection, and hole diameter deviation analysis. The module collects cutting torque through a torque sensor and calculates the average torque of five consecutive holes. It captures images of the tool cutting edge using an industrial camera and analyzes the degree of cutting edge dulling. Combined with the hole diameter deviation data from the quality inspection module, a tool wear assessment model is established. The module records the number of holes opened, usage time, and total amount of profile processed for each tool, generates a tool life prediction curve, supports advance stocking, and automatically performs trial cutting calibration after replacing a new tool.

6. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes a safety protection module, which constructs a multi-level protection system: Level 1 protection is an infrared light grid, which immediately sends a stop command when personnel or foreign objects enter the danger zone; Level 2 protection is area warning, which displays a yellow warning on the touchscreen and issues an audible and visual alert when an object appears in the warning zone; Level 3 protection is emergency stop control, with 3 emergency stop buttons around the equipment, which cut off the main power supply to the equipment upon triggering and save the current opening data; the module monitors the equipment's operating parameters in real time; the equipment automatically performs a safety check before starting, and can only enter the operating mode after passing the check, while storing the protection trigger records for the past 3 months.

7. The intelligent opening system for thermally broken aluminum window and door frames based on machine vision according to claim 1, characterized in that, It also includes a remote operation and maintenance module, which establishes a real-time connection between the device and the cloud platform via a 4G / 5G network; the cloud platform is equipped with an AI fault diagnosis model that can identify more than 20 common faults; when the device malfunctions, the platform pushes the cause of the fault and repair steps to the operation and maintenance personnel's mobile APP within 10 seconds, and supports remote parameter debugging; the module regularly generates device health reports, and pushes repair reminders when the score is below 60. The system pushes out equipment maintenance plans monthly, and uploads records to the cloud after maintenance is completed. It supports centralized management of multiple devices, and maintenance personnel can view the online status and production progress of all devices through the platform, enabling remote monitoring and batch scheduling.

8. A method for intelligent opening of thermally broken aluminum window and door frames based on machine vision as described in any one of claims 1-7, characterized in that, Includes the following steps: Step 1: Frame loading and vision acquisition. Place the thermally broken aluminum window frame on the fixture of the opening platform, start the fixture to fix the frame, start the machine vision acquisition module, and six industrial CCD cameras acquire images of the six sides of the frame in batches under the assistance of 360° ring light source. After Gaussian filtering and noise reduction and edge enhancement preprocessing, the images are transmitted to the frame positioning and calibration module. Step 2: Frame positioning and calibration. The frame positioning and calibration module extracts the corner points of the frame border, the endpoints of the reinforcing ribs, and the center of the preset reference holes in the image through the feature point recognition algorithm, establishes the image coordinate system, and uses the laser range sensor to measure the actual length, width and height of the frame. The positioning deviation is calculated by comparing it with the theoretical dimensions in the design drawings, and a positioning calibration matrix is ​​generated. Step 3: Hole opening parameter planning. The staff inputs the customer's required parameters through the touch screen. The hole opening parameter planning module calls the frame profile parameters collected by machine vision to plan the hole opening path, calls the hole diameter thermal deformation compensation module to calculate the compensated hole diameter, and sets the tool speed and initial feed speed in combination with the profile material density to generate CNC code and preview the hole opening path. Step 4: Intelligent hole drilling execution. The intelligent execution module receives the CNC code, and the three-axis CNC platform drives the tool to move to the first hole position. The automatic tool changer selects the appropriate tool, starts the tool to rotate to the set speed, and the torque sensor monitors the cutting torque in real time. When the torque exceeds the preset threshold, the feed speed is reduced. After completing one hole position, the tool returns to a safe height and moves to the next hole position until all holes are completed. Step 5: Quality Inspection and Feedback. The quality inspection module activates the machine vision acquisition module to acquire images of the opening area for the second time, analyzes the opening diameter deviation, hole position offset, and hole wall roughness, and compares the actual hole diameter measured by the laser diameter measuring instrument with the design parameters to determine the quality level. If it is qualified, the data is recorded; if it is unqualified, the correction suggestions are pushed to the intelligent execution module for secondary correction. Step Six: Data Storage and Update. The data storage and interaction module encrypts and stores the framework design drawings, visual acquisition images, opening parameters, and quality inspection results to the local server and cloud database. It generates production reports in Excel format according to the production date and uses the data as a sample to update the algorithm model of the opening parameter planning module.

9. The method for intelligent opening of thermally broken aluminum window and door frames based on machine vision according to claim 8, characterized in that, It also includes a dynamic compensation step. In this step, during the intelligent drilling process in step four, the temperature of the cutting area of ​​the tool and the non-cutting area of ​​the frame are collected every 0.5 seconds by a temperature sensor to calculate ΔT, and the real-time torque value is collected every 0.3 seconds by a torque sensor. Combining the cutting edge wear data from the tool wear monitoring module, and using the formula... Adjust feed rate ,in The compensated feed rate, The initial feed rate, The initial torque threshold for the cutting tool. β is the real-time torque value, ΔT is the temperature influence coefficient, and ΔT is the real-time temperature change. This step fine-tunes the compensation coefficient according to the hole-opening stage: the β value is set to 0.006 at the initial stage of hole opening to enhance temperature compensation. The β value recovers to 0.005 in the mid-stage of orifice opening; the torque compensation weight is increased in the late stage of orifice opening. The weighting is increased to 0.6; simultaneously, the diameter of each hole is measured in real time every 5 holes using a laser diameter gauge, and compared with the compensated theoretical diameter. If the deviation exceeds 0.01mm, it is corrected immediately. .

10. The method for intelligent opening of thermally broken aluminum window and door frames based on machine vision according to claim 8, characterized in that, The system also includes a multi-batch production adaptation step. Before the frame is loaded in step one, the system uses the data storage and interaction module to generate production data, statistically analyze common opening parameters, quality pass rate, and average production time according to the frame specifications, and generate multi-batch production templates. When more than 10 frames of the same specifications are loaded consecutively, the system automatically identifies the frame specifications and loads the corresponding template. The staff only needs to confirm that the frame has no obvious defects through the machine vision acquisition module before proceeding directly to step two, positioning and calibration. The templates support hierarchical editing permissions: administrators can modify all parameters, while operators can only adjust non-core parameters. After each batch of production is completed, the system automatically calculates the pass rate and reasons for non-conformity of the batch, generates a batch quality analysis report, and marks the direction for improvement. At the same time, each frame is assigned a unique QR code, which can be scanned to query the batch to which the frame belongs, opening parameters, quality inspection results, and operator.

Images