Roller screen online detection system based on 3D visual identification

Abstract

The invention relates to the technical field of roller screen detection, in particular to a roller screen online detection system based on 3D visual recognition. Comprising a database, a vibration detection module, a temperature detection module, a broken pin detection module and a detection output module, according to the database, data are acquired through 3D laser scanning and infrared thermal imaging, a three-dimensional model is constructed, and temperature data are mapped. The vibration detection module analyzes vibration of a motor and a speed reducer, the temperature detection module evaluates temperature distribution and risk, the broken pin detection module checks breakage and abrasion of a connecting pin, and the detection output module processes and visualizes an early warning signal; accurate vibration detection, temperature detection and connection pin fracture detection of the roller screen are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of roller screen detection, and in particular to an online roller screen detection system based on 3D visual recognition. Background Art

[0002] In today's wave of industrial intelligence, efficient and accurate detection technology has become the key to improving production quality and efficiency. The demand for 3D visual recognition online detection systems has emerged. Traditional detection methods often rely on manual labor, which is not only inefficient but also prone to errors. However, 3D visual recognition technology, with its advantages such as high precision and non-contact measurement, has gradually become the new favorite in the field of industrial inspection.

[0003] Traditional methods for online inspection of roller screens currently face numerous challenges. Manual inspection is inefficient and difficult to meet the demands of large-scale production. Cable routing is also costly and prone to failure, impacting the stability of the inspection system. 3D visual recognition online inspection systems can effectively address these challenges.

[0004] Therefore, there is a need for an online detection system for roller screens based on 3D visual recognition. Summary of the Invention

[0005] The purpose of the present invention is to provide an online detection system for roller screens based on 3D visual recognition to solve the problems mentioned in the above background technology.

[0006] The object of the present invention can be achieved by the following technical solutions: a roller screen online detection system based on 3D visual recognition, comprising: a database, a vibration detection module, a temperature detection module, a broken pin detection module and a detection output module;

[0007] The database is used to obtain three-dimensional temperature maps and three-dimensional stereograms of key parts of the roller screen's operating status through 3D visual acquisition, including motors, reducers, bearings, and connecting pins, and save them;

[0008] The vibration detection module performs vibration detection based on the three-dimensional image of the motor and reducer in the running state of the roller screen to determine the vibration state of the motor and reducer and generate corresponding abnormal vibration signals;

[0009] The temperature detection module performs temperature detection and analysis based on the three-dimensional temperature map of each key part to determine whether there is a temperature anomaly and generate a corresponding temperature warning signal;

[0010] The broken pin detection module performs broken pin detection on the roller screen's connecting pins based on the three-dimensional image of the connecting pins to promptly detect whether the connecting pins are broken or worn, and generates corresponding connecting pin breakage warning signals or connecting pin wear signals;

[0011] The detection output module performs notification processing and visualization processing based on the received signals.

[0012] Preferably, the specific method of 3D visual acquisition is:

[0013] A 3D laser scanner is used to collect high-precision three-dimensional point cloud data of the roller screen's operating status. Based on the point cloud data, a three-dimensional digital model of the roller screen's key parts is constructed. Specific key parts include the motor, reducer, bearings, and connecting pins. Each key part is annotated in the three-dimensional digital model, and its position coordinates in the three-dimensional digital model are determined. Each key part in the three-dimensional digital model is identified and its point cloud is segmented to obtain a three-dimensional stereogram of each key part.

[0014] Scan the roller screen with an infrared thermal imaging scanner to obtain a thermal image, and align the temperature data in the thermal image with the coordinates of the key parts marked in the three-dimensional digital model to map the temperature data to the key parts in the three-dimensional digital model, thereby obtaining a three-dimensional temperature map of each key part;

[0015] Thus, a three-dimensional temperature diagram and a three-dimensional stereogram of the motor, reducer, bearing and connecting pin when the roller screen is in operation can be obtained.

[0016] Preferably, the specific method of vibration detection is:

[0017] Extract the three-dimensional images of the motor and reducer, and sort the three-dimensional images in order of acquisition time. Mark several feature points in the three-dimensional images, track the displacement trajectory of the feature points during operation, and record their real-time operation trajectory in three-dimensional space. Use image processing algorithms to calculate the displacement of the feature points per unit time, and calculate the vibration speed based on the displacement. In this way, the vibration speed of the motor and reducer at each acquisition time can be obtained, and they are recorded as V Motor and V Reducer ; where V Motor Indicates the vibration speed of the motor, V Reducer Indicates the vibration speed of the reducer;

[0018] If there is a motor vibration speed V at a certain acquisition time Motor > motor vibration limit, the motor vibration speed V Motor The collection time is the monitoring starting point, and a fixed time is set. The monitoring starting point and the fixed time constitute a monitoring window. The motor vibration speed at each collection time in the monitoring window is compared with the motor vibration limit. If V Motor > motor vibration limit, the motor vibration speed V MotorThe corresponding collection moment is marked as the abnormal vibration moment of the motor. The number of abnormal vibration moments of the motor in the monitoring window is calculated and recorded as A. If A ≥ K1, the abnormal vibration signal of the motor is generated and sent to the detection output module;

[0019] Similarly, the number of abnormal vibration moments of the reducer in the monitoring window obtained by the vibration anomaly detection of the reducer is recorded as B. If B ≥ K2, an abnormal vibration signal of the reducer is generated and sent to the detection output module, where K1 and K2 are constants.

[0020] Preferably, the specific method of temperature detection is:

[0021] 4-1: Extract the three-dimensional temperature map of each key part, sort the three-dimensional images in order of acquisition time, and divide the three-dimensional temperature map into several color areas using three reference tones;

[0022] 4-2: Set several points in each color region, identify the color of each point, and compare it with all the set colors to obtain the corresponding temperature color value; calculate the average temperature color value of each point in the color region to obtain the temperature color mean of the color region, and record it as Wi, where i = 1, 2, 3...I, I is a positive integer, I represents the total number of color regions in the three-dimensional temperature map, and i represents any color region in it; use the standard deviation mathematical model to measure the unevenness of the temperature distribution in the three-dimensional stereogram to obtain the temperature discrete value;

[0023] 4-3: Sum the temperature color mean values ​​of several blue areas, several yellow areas, and several red areas in the three-dimensional temperature map to obtain the blue temperature value, yellow temperature value, and red temperature value; calculate the area of ​​each color area using the pixel counting method, and sum the area of ​​each green area, yellow area, and red area to obtain the green area, yellow area, and red area;

[0024] The blue temperature value, yellow temperature value, red temperature value, green area, yellow area and red area in the three-dimensional temperature map are used to measure the temperature risk of key parts using a multidimensional decision model to obtain a temperature risk value.

[0025] 4-4: The temperature discrete values ​​and temperature risk values ​​of each key part are used in a weighted model to comprehensively measure the temperature status of each key part to obtain the temperature warning value. If the temperature warning value is greater than or equal to the set warning threshold, the acquisition moment is recorded as the temperature abnormality moment. The number of abnormal motor vibration moments in the monitoring window is recorded as C. If C ≥ K3, a temperature warning signal is generated and the corresponding key part name is output. K3 is a set constant.

[0026] Preferably, the standard deviation mathematical model is:

[0027]

[0028] Preferably, the multidimensional decision-making mathematical model is:

[0029]

[0030] Among them, α1, α2, and α3 are the set proportional coefficients, and α2>α3>α1>0, M green 、M yellow and M red are blue temperature value, yellow temperature value and red temperature value respectively; S green 、S yellow and S red They are green area, yellow area and red area respectively.

[0031] Preferably, the specific method of broken pin detection is:

[0032] 7-1: Extract the 3D image of the connecting pin positions, mark the connecting pins on each axis in the 3D image, and establish coordinate reference points; set each connecting pin to correspond to a standard shape;

[0033] 7-2: Match the point cloud of the connecting pin with the corresponding standard form to obtain the best match; calculate the minimum matrix value Tmin of the standard form that best matches the point cloud data;

[0034] 7-3: Calculate the three-dimensional profile change ΔD between the connecting pin and its corresponding standard form. If the three-dimensional profile change ΔD ≥ the set threshold, the connecting pin is determined to be broken, a connecting pin breakage warning signal is generated, and the corresponding connecting pin number is output. Otherwise, execute step 7-4 to detect and determine the wear status of the connecting pin.

[0035] 7-4: Compare and analyze the minimum matrix value Tmin of the connecting pin with the set wear thresholds T1 and T2 to determine the wear condition of the connecting pin, specifically:

[0036] If Tmin≥T2, a signal indicating severe wear of the connecting pin is generated and the connecting pin number is output;

[0037] If T2>Tmin≥T1, a signal indicating slight wear of the connecting pin is generated and the connecting pin number is output.

[0038] Preferably, the specific method of point cloud matching is:

[0039] FPFH is used to calculate the local geometric features of the point cloud, match the feature points, and RANSAC is used to remove incorrect matching points;

[0040] Calculate the optimal transformation matrix T(R,t) and align the current point cloud with the standard point cloud;

[0041] Record any point cloud in the current connecting pin as Calculate the current point cloud Compared with standard point cloud The optimal transformation matrix T(R, t) between is specifically:

[0042]

[0043] Where "|| ||" represents the Euclidean distance, that is, the spatial distance between two point clouds is calculated, "R" represents the rotation matrix, and "t" represents the translation vector, which can be used to align the point cloud data to obtain the optimal match. The minimum matrix value Tmin is calculated by calculating the standard form of the optimal match of the point cloud data. The specific calculation formula is:

[0044] Thus, the minimum matrix value Tmin between each connecting pin and its corresponding standard form can be obtained.

[0045] Preferably, the calculation model of the three-dimensional profile variation is:

[0046]

[0047] in is the point cloud coordinate within the current connecting pin, is the point cloud coordinate of the standard connecting pin.

[0048] Preferably, the notification processing and visualization processing of the detection output module are:

[0049] When a connection pin breakage warning signal is received, the shutdown protection is triggered;

[0050] The warning signal levels are set as follows: connecting pin breakage warning signal > temperature warning signal > motor normal vibration signal > reducer abnormal vibration signal > connecting pin severe wear signal > connecting pin slight wear signal. The received signals are sorted according to their corresponding levels and displayed visually, and warning notifications are issued in sequence according to their corresponding levels, so that the corresponding engineers can be notified in time to repair and maintain the roller screen.

[0051] Beneficial effects of the present invention:

[0052] 1. By tracking the feature points of the three-dimensional stereogram, the vibration trajectory of the motor and reducer in three-dimensional space can be accurately captured, avoiding misjudgment caused by the installation position or direction deviation of traditional single-axis vibration sensors; the monitoring window design (fixed duration and non-overlapping) effectively filters out instantaneous interference (such as the impact of equipment start-up and shutdown), and only alarms for continuous abnormal vibration (such as A ≥ K1 or B ≥ K2), reducing the false alarm rate, increasing anti-interference and robustness, and realizing accurate identification and early warning of the vibration of the roller screen motor and reducer;

[0053] 2. By aligning infrared thermal imaging data with a three-dimensional digital model, the system divides the area into blue, yellow, and red areas and calculates the average temperature value. Based on the temperature dispersion value (standard deviation mathematical model) and the temperature risk value (multidimensional decision model), the system comprehensively evaluates the degree of temperature dispersion and temperature risk, ultimately generating a temperature warning value. This system achieves three-dimensional temperature visualization, quantifies temperature risk, and combines temperature dispersion values ​​and risk values ​​to accurately assess temperature risks and generate timely temperature warnings.

[0054] 3. By using FPFH feature point matching and the RANSAC algorithm to optimize point cloud alignment, the 3D profile change (△D) and minimum matrix value are calculated to determine the fracture and wear level of the connecting pin. It can identify even tiny deformations of the connecting pin, far exceeding the accuracy of traditional visual or contact flaw detection methods. 3D vision technology does not require equipment disassembly or downtime for testing, making it particularly suitable for roller screens operating at high speeds or in high-risk environments, improving detection efficiency. It also enables precise monitoring of the connecting pins, ensuring their integrity and stability.

[0055] 4. By sorting the warning signals by level (connecting pin fracture > temperature > vibration > wear), the visual interface displays them in levels and pushes notifications. The level sorting (such as the connecting pin fracture warning has the highest level) ensures that engineers prioritize the most urgent faults and avoid secondary damage caused by improper processing sequence. The visual interface integrates a three-dimensional temperature graph, a vibration trend curve, and a connecting pin point cloud comparison graph to provide a global perspective on the root cause of the fault. It supports linkage with the PLC system and automatically triggers shutdown protection when a connecting pin fracture signal is received, minimizing equipment damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of system module connection of the present invention. DETAILED DESCRIPTION

[0057] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0058] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0059] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0060] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0061] See also Figure 1 As shown, the roller screen online detection system based on 3D visual recognition of the present invention includes: a database, a vibration detection module, a temperature detection module, a broken pin detection module and a detection output module;

[0062] The database uses 3D vision to collect and save the 3D temperature graphs and 3D stereograms of key parts of the roller screen during operation. The key parts include the motor, reducer, bearings, and connecting pins. The specific 3D vision collection process is as follows:

[0063] A 3D laser scanner is used to collect high-precision three-dimensional point cloud data of the operating status of the roller screen, and a three-dimensional digital model of the key parts of the roller screen is constructed based on the point cloud data. The specific key parts include motors, reducers, bearings and connecting pins. Each key part is marked in the three-dimensional digital model, and the position coordinates of each key part in the three-dimensional digital model are determined; wherein the connecting pin refers to the connecting pin at 8-36 shafts of the roller screen (due to different models of roller screens, the number of shafts is different, but the number of shafts of the roller screen in the specific application scenario of this application is 8-36), and the connecting pin on each shaft is used to connect the driving wheel and the driven wheel (roller) to transmit power, or to fix the shaft and the screen plate to maintain the structural integrity of the equipment; if the connecting pin is severely worn or broken, it will cause the transmission connection and the fixed connection to be unstable, and even cause the shaft to lose power transmission ability or disconnect from the screen plate, thereby causing equipment failure; identify each key part in the three-dimensional digital model and perform point cloud segmentation on it to obtain a three-dimensional stereogram of each key part;

[0064] An infrared thermal imaging scanner is used to scan the roller screen to obtain a thermal image, and the temperature data in the thermal image is aligned with the coordinates of the key parts marked in the three-dimensional digital model to map the temperature data to each key part in the three-dimensional digital model, thereby obtaining a three-dimensional temperature map of each key part. It should be noted that the three-dimensional temperature distribution map is usually displayed in color, and the temperature and color of each area in the key part are different. The closer the color is to the warm tone (red), the higher the temperature, and the closer the color is to the cool tone (blue), the lower the temperature; different colors are set to correspond to a temperature color value, and the larger the temperature color value, the closer the color is to red and the higher the temperature;

[0065] It should be noted that the synchronous data collection of the 3D laser scanner and the infrared thermal imaging scanner can ensure the time consistency of the two data sources, avoid data alignment errors caused by changes in equipment operation, and achieve accurate temperature data mapping;

[0066] Thus, the three-dimensional temperature diagram and three-dimensional stereogram of the motor, reducer, bearing and connecting pin under the running state of the roller screen can be obtained;

[0067] The vibration detection module performs vibration detection based on the three-dimensional image of the motor and reducer in the running state of the roller screen to determine the vibration status of the motor and reducer, specifically:

[0068] Extract the three-dimensional stereograms of the motor and reducer, and sort the three-dimensional stereograms in order of acquisition time to ensure that each frame of data is accurately associated with its corresponding time point; mark several feature points in the three-dimensional stereogram (specific feature points usually have significant shape, texture or position features to ensure accurate positioning in the three-dimensional stereogram, such as bolts and identification marks on the motor housing, or structural corners and surface concave and convex parts on the reducer), track the displacement trajectory of the feature points during operation, and record their real-time operation trajectory in three-dimensional space; use image processing algorithms (such as Fourier transform, time domain analysis, etc.) to calculate the displacement of the feature points per unit time, and calculate the vibration amplitude based on the displacement, which is the vibration velocity, and the unit of vibration velocity is "mm per second"; thus, the vibration velocity of the motor and reducer at each acquisition moment can be obtained, and recorded as V Motor and V Reducer ; where V Motor Indicates the vibration speed of the motor, V Reducer Indicates the vibration speed of the reducer;

[0069] Compare and analyze the motor vibration speed at each acquisition moment with the set motor vibration limit (technical personnel in this field usually set the motor vibration limit to 25mm / s). If there is a motor vibration speed V at a certain acquisition moment, Motor > motor vibration limit, the motor vibration speed V MotorThe collection time is the monitoring starting point, and a fixed time is set. The monitoring starting point and the fixed time constitute a monitoring window. The motor vibration speed at each collection time in the monitoring window is compared with the motor vibration limit. If V Motor > motor vibration limit, the motor vibration speed V Motor The corresponding collection moment is marked as the abnormal vibration moment of the motor. The number of abnormal vibration moments of the motor in the monitoring window is calculated and recorded as A. If A ≥ K1 (K1 is a set constant), the abnormal vibration signal of the motor is generated and sent to the detection output module;

[0070] It should be noted that the collection time within a monitoring window is not used as the starting point of the next monitoring window, which means that the time periods involved in each monitoring window do not overlap;

[0071] Similarly, for the detection of reducer vibration anomalies: the reducer vibration speed at each sampling moment is compared with the set reducer vibration limit (technical personnel in this field usually set the motor vibration limit to 58mm / s). If there is a reducer vibration speed V at a certain sampling moment, Reducer > reducer vibration limit, the reducer vibration speed V Reducer The collection time is the monitoring starting point, and a fixed time is set. The monitoring starting point and the fixed time constitute a monitoring window. The vibration speed of the reducer at each collection time in the monitoring window is compared with the vibration limit of the reducer. If V Reducer > reducer vibration limit, then the reducer vibration speed V Reducer The corresponding collection moment is marked as the reducer abnormal vibration moment, and the number of reducer abnormal vibration moments within the statistical monitoring window is recorded as B. If B ≥ K2 (K2 is a set constant), the reducer abnormal vibration signal is generated and sent to the detection output module;

[0072] By tracking the feature points of the three-dimensional stereogram, the vibration trajectory of the motor and reducer in three-dimensional space can be accurately captured, avoiding misjudgment caused by installation position or direction deviation of traditional single-axis vibration sensors; the monitoring window design (fixed duration and non-overlapping) effectively filters out instantaneous interference (such as equipment start-up and shutdown impact), and only alarms for continuous abnormal vibrations (such as A≥K1 or B≥K2), reducing the false alarm rate, increasing anti-interference and robustness, and realizing accurate identification and early warning of the vibration of the roller screen motor and reducer.

[0073] The temperature detection module performs temperature detection and analysis based on the three-dimensional temperature map of key parts to determine whether there is a temperature anomaly. Specifically:

[0074] Extract the three-dimensional temperature map of each key part, and sort the three-dimensional stereograms in order of acquisition time; divide the three-dimensional temperature map into several color areas with three reference tones, specifically: several blue areas, several yellow areas and several red areas, and the specific three reference tones are blue, yellow and red; set several points in each color area, identify the color of each point, and compare it with all the set colors to obtain the corresponding temperature value; it should be noted that people in this field usually use color bars to set color-temperature values, and the colors of the color bar from left to right are dark blue, light blue, Light yellow, dark yellow, light red, dark red; the closer the color is to the left of the color bar, the cooler the color tone, the smaller the temperature value, and the lower the temperature; the closer the color is to the right of the color bar, the warmer the color tone, the larger the temperature value, and the higher the temperature; the temperature value of each point in the color area is averaged to obtain the temperature mean of the color area, and it is recorded as Wi, where i = 1, 2, 3...I, I is a positive integer, I represents the total number of color areas in the three-dimensional temperature map, and i represents any color area in it; the standard deviation mathematical model is used to measure the imbalance of temperature distribution in the three-dimensional stereogram to obtain the temperature dispersion value σ W It should be noted that the larger the temperature dispersion value, the more uneven the temperature of the key parts, and the greater the safety hazard; the standard deviation mathematical model is specifically:

[0075]

[0076] The temperature color mean values ​​of several blue areas, several yellow areas and several red areas in the three-dimensional temperature map are summed up to obtain the blue temperature value, yellow temperature value and red temperature value, and are recorded as M green 、M yellow and M red ; Use pixel counting method to calculate the area of ​​each color area, and sum up the area of ​​each green area, yellow area and red area to get the green area, yellow area and red area, and record them as S respectively green 、S yellow and S red ;

[0077] The blue temperature value M in the three-dimensional temperature map green , yellow temperature value M yellow , red temperature value M red , green area S green , yellow area S yellow and the red area S red The temperature risk of key parts is measured using a multidimensional decision model to obtain the temperature risk value σ M It should be noted that when the temperature risk value is larger, it means that the temperature of the three-dimensional temperature map is higher, and there is a greater risk of temperature anomaly; the multidimensional decision model is specifically as follows:

[0078]

[0079] Where α1, α2, and α3 are respectively set proportional coefficients, and α2>α3>α1>0; for example, the value of α2 is 1.891, the value of α3 is 1.464, and the value of α3 is 1.173; in this application scenario, engineers in this field use the green color bar in the key parts to indicate that the temperature is normal. If this application is applied to other devices, it can be adjusted according to the specific conditions of the device;

[0080] The temperature discrete values ​​σ of each key part W and temperature risk value σ M The weighted model is used to comprehensively measure the temperature status of each key part to obtain the temperature warning value σ W arm, the weighted model is as follows:

[0081]

[0082] Where β1 and β2 are respectively the set proportional coefficients, and β2>β1>0; when the temperature dispersion value of a key part is larger and the temperature risk value is larger, the temperature warning value is larger, indicating that the temperature anomaly of the key part is more obvious; the temperature warning value is compared and analyzed with the set warning threshold. If the temperature warning value is greater than or equal to the set warning threshold, the acquisition moment is recorded as the temperature anomaly moment, and the number of abnormal motor vibration moments in the monitoring window is recorded as C. If C≥K3 (K3 is a set constant), a temperature warning signal is generated and the corresponding key part name is output to prevent equipment damage due to overheating or uneven temperature.

[0083] By aligning infrared thermal imaging data with the three-dimensional digital model, dividing the blue, yellow and red areas and calculating the temperature color mean, the temperature dispersion and temperature risk levels are comprehensively evaluated based on the temperature dispersion value (standard deviation mathematical model) and the temperature risk value (multidimensional decision-making model), and finally a temperature warning value is generated; it realizes three-dimensional temperature visualization, quantifies temperature risks, combines temperature dispersion values ​​and risk values, accurately evaluates temperature risks and generates temperature warnings in a timely manner.

[0084] The broken pin detection module uses 3D visual recognition technology to detect the broken pins of the roller screen and promptly detect whether the connecting pins are broken. Specifically:

[0085] Extract a 3D image of the connecting pin positions, mark the connecting pins on each axis in the 3D image, and establish coordinate reference points. Set each connecting pin to a standard form. It should be noted that the standard form refers to the original form of the current connecting pin, which is a complete form without use or wear.

[0086] Match the point cloud of the connecting pin with the corresponding standard form, and detect the connecting pin based on the matching results. Specifically:

[0087] Step 1: Use FPFH (Fast Point Feature Histogram) to calculate the local geometric features of the point cloud, match the feature points, and use RANSAC (RANdom Sampling Consensus) to remove incorrect matching points;

[0088] Step 2: Calculate the optimal transformation matrix T(R,t) and align the current point cloud with the standard point cloud;

[0089] Record any point cloud in the current connecting pin as Calculate the current point cloud Compared with standard point cloud The optimal transformation matrix T(R, t) between is specifically:

[0090]

[0091] Where "|| ||" represents the Euclidean distance, that is, the spatial distance between two point clouds is calculated, "R" represents the rotation matrix, and "t" represents the translation vector, which can align the point cloud data to obtain the best match; the above formula represents the transformation after rotation R and translation t. and The smaller the Euclidean distance is, the better the current point cloud The closer to the standard point cloud The matching effect is better; the standard form of the optimal matching of the point cloud data is calculated to obtain the minimum matrix value Tmin. The specific calculation formula is:

[0092] Thus, the minimum matrix value Tmin between each connecting pin and its corresponding standard form can be obtained;

[0093] Step 3: Calculate the three-dimensional profile change △D between the connecting pin and its corresponding standard form. The specific calculation formula is:

[0094]

[0095] in is the point cloud coordinate within the current connecting pin, is the point cloud coordinate of the standard connecting pin; if the three-dimensional contour change △D ≥ the set threshold, the connecting pin is determined to be broken, a connecting pin fracture warning signal is generated, and the corresponding connecting pin number is output; otherwise, step 4 is executed to detect and judge the wear status of the connecting pin;

[0096] Step 4: Compare and analyze the minimum matrix value Tmin of the connecting pin with the set wear thresholds T1 (slight wear) and T2 (severe wear) to determine the wear condition of the connecting pin. It should be noted that 0<T1<T2 specifically means:

[0097] If Tmin≥T2, a signal indicating severe wear of the connecting pin is generated and the connecting pin number is output;

[0098] If T2>Tmin≥T1, a signal indicating slight wear of the connecting pin is generated and the connecting pin number is output;

[0099] If Tmin < T1, it means that the wear of the connecting pin is very small and can be ignored, so no signal needs to be generated;

[0100] By using FPFH feature point matching and RANSAC algorithm to optimize point cloud alignment, calculate the three-dimensional contour change (△D) and minimum matrix value, judge the fracture and wear level of the connecting pin, and identify tiny deformations of the connecting pin, far exceeding the accuracy of traditional visual or contact flaw detection methods. 3D vision technology does not require equipment disassembly or shutdown for inspection, and is particularly suitable for roller screens in high-speed operation or high-risk environments, improving inspection efficiency; it realizes accurate monitoring of the connecting pins to ensure their integrity and stability.

[0101] The detection output module outputs and visualizes the received motor normal vibration signal, reducer abnormal vibration signal, temperature warning signal, connecting pin breakage warning signal, connecting pin severe wear signal, and connecting pin slight wear signal. Specifically:

[0102] When a warning signal of a broken connecting pin is received, shutdown protection is triggered to minimize equipment damage;

[0103] Set the warning signal levels as follows: connecting pin breakage warning signal > temperature warning signal > motor normal vibration signal > reducer abnormal vibration signal > connecting pin severe wear signal > connecting pin slight wear signal. Sort the received signals according to their corresponding levels and display them visually. Then issue warning notifications in turn according to their corresponding levels, and notify the corresponding engineers in time to repair and maintain the roller screen.

[0104] By sorting the warning signals by level (connecting pin fracture > temperature > vibration > wear), the visual interface displays them in levels and pushes notifications. The level sorting (such as the connecting pin fracture warning has the highest level) ensures that engineers give priority to the most urgent faults and avoid secondary damage caused by improper processing sequence. The visual interface integrates a three-dimensional temperature graph, a vibration trend curve and a connecting pin point cloud comparison graph to provide a global perspective on the root cause of the fault, support linkage with the PLC system, and automatically trigger shutdown protection when a connecting pin fracture signal is received, thereby minimizing equipment damage.

[0105] The above content is merely an example and explanation of the structure of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the scope of protection of the present invention.

Claims

1. The roller screen online detection system based on 3D visual recognition is characterized by: include: Database, vibration detection module, temperature detection module, broken pin detection module and detection output module; The database is used to store and save the three-dimensional temperature map and three-dimensional stereogram of each key part of the roller screen's operating status obtained through 3D vision acquisition. The specific key parts include the motor, reducer, bearing and connecting pin; The vibration detection module performs vibration detection based on the three-dimensional image of the motor and reducer in the running state of the roller screen to determine the vibration state of the motor and reducer and generate corresponding abnormal vibration signals; The temperature detection module performs temperature detection and analysis based on the three-dimensional temperature map of each key part to determine whether there is a temperature anomaly and generate a corresponding temperature warning signal; The broken pin detection module detects the broken pins of the roller screen based on the three-dimensional image of the connecting pins to detect whether the connecting pins are broken or worn in time, and generates corresponding connecting pin breakage warning signals or connecting pin wear signals; The detection output module performs notification processing and visualization processing based on the received signals.

2. The roller screen online detection system based on 3D visual recognition according to claim 1 is characterized in that: The specific methods of 3D visual acquisition are: A 3D laser scanner is used to collect high-precision three-dimensional point cloud data of the roller screen's operating status. Based on the point cloud data, a three-dimensional digital model of the roller screen's key parts is constructed. Specific key parts include the motor, reducer, bearings, and connecting pins. Each key part is annotated in the three-dimensional digital model, and its position coordinates in the three-dimensional digital model are determined. Each key part in the three-dimensional digital model is identified and its point cloud is segmented to obtain a three-dimensional stereogram of each key part. Scan the roller screen with an infrared thermal imaging scanner to obtain a thermal image, and align the temperature data in the thermal image with the coordinates of the key parts marked in the three-dimensional digital model to map the temperature data to the key parts in the three-dimensional digital model, thereby obtaining a three-dimensional temperature map of each key part; Thus, a three-dimensional temperature diagram and a three-dimensional stereogram of the motor, reducer, bearing and connecting pin when the roller screen is in operation are obtained.

3. The roller screen online detection system based on 3D visual recognition according to claim 1 is characterized in that: The specific methods of vibration detection are: Extract the three-dimensional images of the motor and reducer, and sort the three-dimensional images in order of acquisition time. Mark several feature points in the three-dimensional images, track the displacement trajectory of the feature points during operation, and record their real-time operation trajectory in three-dimensional space. Use image processing algorithms to calculate the displacement of the feature points per unit time, and calculate the vibration speed based on the displacement. In this way, the vibration speed of the motor and reducer at each acquisition time can be obtained, and they are recorded as V Motor and V Reducer ; where V Motor Indicates the vibration speed of the motor, V Reducer Indicates the vibration speed of the reducer; If there is a motor vibration speed V at a certain acquisition time Motor > motor vibration limit, the motor vibration speed V Motor The collection time is the monitoring starting point, and a fixed time is set. The monitoring starting point and the fixed time constitute a monitoring window. The motor vibration speed at each collection time in the monitoring window is compared with the motor vibration limit. If V Motor > motor vibration limit, the motor vibration speed V Motor The corresponding collection moment is marked as the abnormal vibration moment of the motor. The number of abnormal vibration moments of the motor in the monitoring window is calculated and recorded as A. If A ≥ K1, the abnormal vibration signal of the motor is generated and sent to the detection output module; Similarly, the number of abnormal vibration moments of the reducer in the monitoring window obtained by the vibration anomaly detection of the reducer is recorded as B. If B ≥ K2, an abnormal vibration signal of the reducer is generated and sent to the detection output module, where K1 and K2 are constants.

4. The roller screen online detection system based on 3D visual recognition according to claim 1 is characterized in that: The specific method of temperature detection is: 4-1: Extract the three-dimensional temperature map of each key part, sort the three-dimensional images in order of acquisition time, and divide the three-dimensional temperature map into several color areas using three reference tones; 4-2: Set several points in each color region, identify the color of each point, and compare it with all the set colors to obtain the corresponding temperature color value; calculate the average temperature color value of each point in the color region to obtain the temperature color mean of the color region, and record it as Wi, where i = 1, 2, 3...I, I is a positive integer, I represents the total number of color regions in the three-dimensional temperature map, and i represents any color region in it; use the standard deviation mathematical model to measure the unevenness of the temperature distribution in the three-dimensional stereogram to obtain the temperature discrete value; 4-3: Sum the temperature color mean values ​​of several blue areas, several yellow areas, and several red areas in the three-dimensional temperature map to obtain the blue temperature value, yellow temperature value, and red temperature value; calculate the area of ​​each color area using the pixel counting method, and sum the area of ​​each green area, yellow area, and red area to obtain the green area, yellow area, and red area; The blue temperature value, yellow temperature value, red temperature value, green area, yellow area and red area in the three-dimensional temperature map are used to measure the temperature risk of key parts using a multidimensional decision model to obtain the temperature risk value; 4-4: Discrete the temperature values ​​σ of each key part W and temperature risk value σ M The weighted model is used to comprehensively measure the temperature status of each key part to obtain the temperature warning value σ Warm , the weighted model is specifically: Where β1 and β2 are the set proportional coefficients, and β2>β1>0; If the temperature warning value is greater than or equal to the set warning threshold, the acquisition moment is recorded as the temperature abnormality moment, and the number of abnormal motor vibration moments in the monitoring window is recorded as C. If C ≥ K3, a temperature warning signal is generated and the corresponding key part name is output. K3 is a set constant.

5. The roller screen online detection system based on 3D visual recognition according to claim 4 is characterized in that: The mathematical model of standard deviation is:

6. The roller screen online detection system based on 3D visual recognition according to claim 5 is characterized in that: The multidimensional decision-making mathematical model is: Among them, α1, α2, and α3 are the set proportional coefficients, and α2>α3>α1>0, M green 、M yellow and M red are blue temperature value, yellow temperature value and red temperature value respectively; S green 、S yellow and S red They are green area, yellow area and red area respectively.

7. The roller screen online detection system based on 3D visual recognition according to claim 1 is characterized in that: The specific method of broken pin detection is: 7-1: Extract the 3D image of the connecting pin positions, mark the connecting pins on each axis in the 3D image, and establish coordinate reference points; set each connecting pin to correspond to a standard shape; 7-2: Match the point cloud of the connecting pin with the corresponding standard form to obtain the best match; calculate the minimum matrix value Tmin of the standard form that best matches the point cloud data; 7-3: Calculate the three-dimensional profile change ΔD between the connecting pin and its corresponding standard form. If the three-dimensional profile change ΔD ≥ the set threshold, the connecting pin is determined to be broken, a connecting pin breakage warning signal is generated, and the corresponding connecting pin number is output. Otherwise, execute step 7-4 to detect and determine the wear status of the connecting pin. 7-4: Compare and analyze the minimum matrix value Tmin of the connecting pin with the set wear thresholds T1 and T2 to determine the wear condition of the connecting pin, specifically: If Tmin≥T2, a signal indicating severe wear of the connecting pin is generated and the connecting pin number is output; If T2>Tmin≥T1, a signal indicating slight wear of the connecting pin is generated and the connecting pin number is output.

8. The roller screen online detection system based on 3D visual recognition according to claim 7 is characterized in that: The specific method of point cloud matching is: FPFH is used to calculate the local geometric features of the point cloud, match the feature points, and RANSAC is used to remove incorrect matching points; Calculate the optimal transformation matrix T(R,t) and align the current point cloud with the standard point cloud; Record any point cloud in the current connecting pin as Calculate the current point cloud Compared with standard point cloud The optimal transformation matrix T(R, t) between is specifically: Where "||||" represents the Euclidean distance, which is used to calculate the spatial distance between two point clouds. "R" represents the rotation matrix, and "t" represents the translation vector. This allows the point cloud data to be aligned to obtain the optimal match. The minimum matrix value Tmin is calculated by calculating the standard form of the optimal matching of the point cloud data. The specific calculation formula is: Thus, the minimum matrix value Tmin between each connecting pin and its corresponding standard form can be obtained.

9. The roller screen online detection system based on 3D visual recognition according to claim 8 is characterized in that: The calculation model of the 3D profile change is: in is the point cloud coordinate within the current connecting pin, is the point cloud coordinate of the standard connecting pin.

10. The roller screen online detection system based on 3D visual recognition according to claim 1 is characterized in that: The notification processing and visualization processing of the detection output module are as follows: When a connection pin breakage warning signal is received, the shutdown protection is triggered; The warning signal levels are set as follows: connecting pin breakage warning signal > temperature warning signal > motor normal vibration signal > reducer abnormal vibration signal > connecting pin severe wear signal > connecting pin slight wear signal. The received signals are sorted according to their corresponding levels and displayed visually, and warning notifications are issued in sequence according to their corresponding levels, so that the corresponding engineers can be notified in time to repair and maintain the roller screen.

Images