Plateau hydroelectric sandstone aggregate particle size on-line detection system based on machine vision

Abstract

The invention discloses a plateau hydroelectric sandstone aggregate particle size on-line detection system based on machine vision, and the system comprises an image collection module which comprises an industrial camera and a light source and is installed above a belt conveyor; a protection device and a supporting structure; the image processing module is used for image preprocessing, Mask R-CNN-based instance segmentation and equivalent ellipse Feret short diameter particle size calculation; the data analysis module is used for calculating the particle size distribution and the supersonic / eson diameter rate in real time; the control module is used for being linked with the crushing and screening equipment to realize closed-loop control; according to the invention, non-contact image acquisition is carried out by using the industrial camera, and rapid, accurate and automatic detection of the particle size of the gravel aggregate is realized by combining an instance segmentation algorithm based on Mask R-CNN and an equivalent ellipse Feret short diameter calculation method, so that the efficiency and precision of particle size analysis are significantly improved, and the method is suitable for large-scale popularization and application. The problems of poor timeliness and subjective errors existing in manual sampling and laboratory screening are solved.

Description

Technical Field

[0001] This invention relates to the field of sand and gravel aggregate production testing and quality control technology, and in particular to an online particle size detection system for sand and gravel aggregates in high-altitude hydropower projects based on machine vision. Background Technology

[0002] When constructing large hydropower stations in plateau regions, it is often necessary to build sand and gravel processing systems near the dam site to meet the continuous supply of aggregates required for large-volume concrete. The particle size distribution of sand and gravel aggregates directly affects the mixing ratio, density, temperature control effect, and structural crack resistance of concrete.

[0003] In the unique environment of high-altitude areas, the aforementioned traditional methods have four shortcomings. First, the thin air, strong ultraviolet radiation, and large temperature differences between day and night in high-altitude regions, coupled with frequent dust storms and extreme weather, make manual on-site sampling difficult and unsafe, and the representativeness of the samples is easily affected by environmental interference. Second, the laboratory screening process is cumbersome and time-consuming, making it difficult to reflect the real-time fluctuations in aggregate particle size in the production line, which can easily lead to deviations in concrete gradation. Third, the construction of hydropower station dams requires long-term, large-scale, and continuous material supply, and laboratory testing cannot meet this high-frequency testing requirement. Fourth, existing testing methods lack effective data communication and intelligent linkage mechanisms with crushing, screening, and other production equipment, making it difficult to achieve real-time closed-loop control of production parameters and affecting the long-term stability of aggregate gradation.

[0004] Therefore, in response to the problems mentioned above, this invention proposes an online particle size detection system for high-altitude hydropower aggregates based on machine vision. Summary of the Invention

[0005] To overcome the problems of high safety risks, poor real-time performance, insufficient scale requirements, and inadequate linkage with the production system caused by manual inspection, this invention proposes an online particle size detection system for sand and gravel aggregates in high-altitude hydropower projects. Through image acquisition and intelligent analysis, the system achieves real-time and accurate detection of sand and gravel aggregate particle size and forms a closed-loop control with the production line, thereby ensuring the continuous stability of aggregate quality.

[0006] The technical solution of this invention is: an online particle size detection system for high-altitude hydropower aggregates based on machine vision, comprising: The image acquisition module, including an industrial camera and an LED light source, is fixedly installed above the belt conveyor to acquire images of sand and gravel aggregates in real time while the belt is running. The industrial camera has a frame rate of no less than 30fps and is equipped with a polarizing filter to suppress surface reflection of the aggregates. The light source is symmetrically installed on both sides of the belt at an inclined angle to form low-angle dark field illumination, highlighting the particle outline. The module is fixedly installed 1.5 meters to 2.5 meters above the belt conveyor to ensure that the field of view covers the entire effective width of the belt. The protective device is designed to be resistant to low temperatures, dust, water, and ultraviolet rays, and encapsulates the image acquisition module. The supporting structure uses a vibration-resistant aluminum alloy frame to fix the image acquisition module and protective devices; The image processing module, connected to the image acquisition module, is used for preprocessing, target segmentation, and particle size calculation of the acquired images; The data analysis module, connected to the image processing module, is used to calculate the particle size distribution, oversize ratio, undersize ratio, and medium size ratio of sand and gravel aggregates in real time, and to automatically identify mixed and mismatched materials. The feedback and control module, connected to the data analysis module, is used to work in conjunction with the crushing and screening equipment to adjust production parameters based on the detection results, forming a closed-loop control. The data storage and display module is used to present data in the form of reports, screening curves, and real-time curves, and archive it to the database; The edge computing server, located in the control room, communicates with the image processing module and the data analysis module to execute image processing algorithms and data calculations, avoiding exposure to the high-altitude outdoor environment.

[0007] Preferably, the industrial camera in the image acquisition module is a high-speed industrial camera, which can capture clear images while the belt is running at high speed.

[0008] Preferably, the LED light source in the image acquisition module is a uniform illumination source, used to provide compensating illumination under strong light conditions at high altitudes, reducing image shadows and reflections. The color temperature of the LED light source is 5000K to 6000K, and the illuminance on the belt surface can reach more than 8000 Lux, thereby ensuring stable image brightness under strong ambient light changes at high altitudes.

[0009] Preferably, the protective device includes a sealed outer shell, a temperature compensation unit, and an ultraviolet filter cover to ensure the normal operation of internal equipment in high-altitude, low-temperature, strong ultraviolet, and dusty environments.

[0010] Preferably, the image processing module includes: The preprocessing unit is used to perform noise reduction, motion blur removal, brightness equalization, and gamma correction on the image to adapt to the high-altitude environment with strong light and dust. The target segmentation unit, based on the Mask R-CNN instance segmentation model, identifies and segments sand and gravel particles one by one, avoiding particle overlap or adhesion. The particle size calculation unit is used to perform morphological fitting on the segmented particles, calculate the equivalent ellipse Feret minor axis, and convert it into the actual physical particle size through a calibration plate.

[0011] Preferably, the system uses an industrial camera and LED light source mounted above the belt conveyor to acquire real-time images of sand and gravel aggregates in a high-altitude environment. The acquired images are then denoised, motion blurred, brightness equalized, and gamma-corrected to eliminate the effects of strong light and dust at high altitudes. Next, an instance segmentation model based on Mask R-CNN is used to identify and segment the sand and gravel particles one by one. Morphological fitting is performed on the segmented particles to calculate the equivalent ellipse Feret minor diameter, which is then converted to the actual physical particle size using a calibration plate. The system then calculates the particle size distribution, oversize ratio, undersize ratio, and median ratio in real time, and determines whether there is mixing or incorrect material. Finally, a particle size histogram, sieve residue curve, and real-time data report are generated and stored in a database. When the oversize ratio or undersize ratio exceeds a set threshold, an alarm is triggered, and the parameters of the crusher and screening machine are automatically adjusted to achieve closed-loop control.

[0012] Preferably, the acquired image also includes a contrast enhancement operation, employing an adaptive histogram equalization method to improve image clarity under low-light conditions at high altitudes.

[0013] Preferably, the Mask R-CNN model is trained on a dataset of sand and gravel images collected in a high-altitude environment, and is able to identify sand and gravel particles under conditions of dust, shadow, and particle adhesion.

[0014] Preferably, the particle size calculation specifically includes: fitting an equivalent ellipse to each segmented sand and gravel particle, then calculating the Feret minor axis of the ellipse as the representative particle size, and finally converting the pixel size into the actual physical size through a pre-calibrated pixel-physical size ratio.

[0015] As a preferred option: when the oversize ratio exceeds the threshold, the crusher discharge port is automatically reduced; when the undersize ratio exceeds the threshold, the vibration frequency or tilt angle of the screening machine is automatically adjusted to optimize the aggregate gradation in real time.

[0016] The beneficial effects of this invention are: 1. This invention utilizes an industrial camera for non-contact image acquisition and combines an instance segmentation algorithm based on Mask R-CNN and a method for calculating the Feret minor axis of the equivalent ellipse to achieve rapid, accurate, and automated detection of sand and gravel aggregate particle size. This significantly improves the efficiency and accuracy of particle size analysis and overcomes the problems of poor timeliness and subjective error in manual sampling and laboratory sieving.

[0017] 2. This invention can calculate and output key quality indicators such as particle size distribution, oversize rate, and undersize rate in real time, and can automatically identify abnormal situations such as mixing and mismatched materials. This meets the urgent need for high-frequency and rapid monitoring of aggregate quality in large-scale continuous production of hydropower projects, and provides a strong guarantee for the stability of the production process.

[0018] 3. By employing specialized protective devices that are resistant to low temperatures, dust, water, and ultraviolet radiation, as well as a vibration-resistant support structure, this invention enables the system to operate stably for extended periods under harsh environmental conditions in high-altitude regions. This effectively overcomes the adverse effects of low oxygen, low temperature, strong ultraviolet radiation, and frequent vibrations on the reliability and lifespan of testing equipment in high-altitude areas. At the same time, the implementation of this invention significantly reduces the amount of manual on-site sampling and testing required in harsh high-altitude environments, effectively reducing the labor intensity and safety risks for construction personnel. While improving quality control capabilities, it also significantly enhances production safety and overall management efficiency.

[0019] 4. By linking real-time detection results with crushing and screening equipment, this invention can automatically adjust production parameters based on the threshold of oversize or undersize ratio, forming a closed-loop system from quality detection to production control, fundamentally improving the stability of aggregate gradation and the level of intelligence in the overall production process. Attached Figure Description

[0020] Figure 1 The diagram shown is a schematic representation of the system framework of the present invention. Figure 2 The diagram shown is a schematic representation of the detection process of this invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, but 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.

[0022] Please see Figure 1 and Figure 2 This invention provides an embodiment: an online particle size detection system for high-altitude hydropower aggregates based on machine vision. In this embodiment, a high-speed industrial area scan camera is selected for the image acquisition module. Its global shutter effectively avoids motion blur caused by the high-speed operation of the conveyor belt. The camera is connected to the subsequent processing unit via a gigabit Ethernet interface. To overcome the influence of strong and uneven ambient light (such as direct sunlight and shadows) at high altitudes, the system is equipped with a high-intensity LED linear light source. This light source is installed on both sides of the camera and projects onto the surface of the conveyor belt material at a low angle, forming a uniform bright field illumination. This effectively highlights the contours and textures of the sand and gravel particles and suppresses ambient light interference. The camera's frame rate and exposure time are set according to the conveyor belt speed to ensure that the acquired image sequence is continuous and free of motion blur.

[0023] Designed for the high-altitude environment, the protective device features a fully sealed stainless steel shell with an IP67 protection rating, effectively preventing dust and moisture intrusion. The shell integrates a temperature control module, including a heater and cooling fan, maintaining internal electronic components within their normal operating temperature range of -30 to 50°C. An anti-UV coated lens and an automatically opening / closing physical wiper are installed in front of the lens to cope with strong UV radiation and rain / snow. The support structure uses a high-strength aluminum alloy frame, securely mounted to the conveyor truss with anchor bolts. High-performance shock-absorbing pads are installed at the contact points between the frame and the conveyor belt, effectively isolating mechanical vibrations caused by conveyor operation and strong winds at high altitudes, ensuring stable image acquisition.

[0024] The image processing and analysis module, including an image processing module and a data analysis module, is connected to the image acquisition module. The image processing module is used to preprocess the acquired images, segment the target, and calculate the particle size. The data analysis module is used to calculate the particle size distribution, oversize ratio, undersize ratio, and medium size ratio of sand and gravel aggregates in real time, and automatically identify mixed and mismatched materials. The feedback and control module, connected to the image processing and analysis module, is used to link with the crushing and screening equipment, adjust production parameters based on the detection results, and form a closed-loop control. The data storage and display module is used to present data in the form of reports, sieve residue curves, and real-time curves, and archive them to the database.

[0025] The detection process of this invention will be described in detail: After the system is powered on, the industrial camera continuously captures images of sand and gravel aggregates on the conveyor belt at a acquisition frequency of 5 frames per second. The LED light source is triggered synchronously to provide stable illumination. The acquired raw RGB images are transmitted to the edge computing server in real time via the network.

[0026] In the image preprocessing stage, the system preprocesses the acquired raw sand and gravel aggregate images to address common factors in plateau environments such as uneven strong light, dust interference, and belt movement. First, nonlocal mean denoising or wavelet threshold denoising algorithms are used to effectively suppress image sensor noise and slight motion blur while preserving particle edge information. Then, adaptive histogram equalization technology is used to perform local histogram equalization on image blocks to improve overall contrast and make particle details in both dark and bright areas clearly distinguishable. Finally, gamma correction is applied to perform nonlinear transformation on image grayscale to compensate for camera response or illumination nonlinearity effects, improve image quality, and lay a reliable foundation for subsequent target segmentation and particle size calculation.

[0027] A Mask R-CNN-based instance segmentation model is employed to accurately identify and segment preprocessed sand and gravel images down to the individual particle level. This model framework adds a branch to Faster R-CNN to predict the binary mask for each region of interest, thereby achieving instance segmentation. ResNet-101 combined with a feature pyramid network is selected as the backbone feature extractor to fully utilize deep semantic features and shallow contour features, which is beneficial for detecting sand and gravel particles of different scales. Candidate target regions are generated through a region proposal network. The model is trained using a dataset of sand and gravel images collected under various conditions in plateau terrain and annotated at the pixel level. Data augmentation techniques such as rotation, brightness variation, and simulated dust noise are introduced to improve the robustness of the model. During training, classification loss, bounding box regression loss, and mask segmentation loss are optimized simultaneously. In forward inference, the model outputs the class probability, bounding box coordinates, and corresponding binary mask for each detected particle to the input image. This mask accurately marks the pixel region of the particle in the image, thus effectively solving the segmentation problem caused by particle adhesion and overlap.

[0028] Morphological analysis was performed on the binary mask of each sand and gravel particle successfully identified and output by the Mask R-CNN instance segmentation model. First, morphological closing operations were used to fill any small holes in the mask region to optimize the integrity of the particle outline. Then, an equivalent ellipse with the same second-order central moment as the mask region was calculated, and the minor axis length of the equivalent ellipse was extracted as the Feret minor diameter of the particle at the image pixel scale. Finally, using the pixel-to-physical size conversion relationship accurately established by the system in the initial calibration stage, the obtained Feret minor diameter (in pixels) was converted into the actual physical particle size (in millimeters), thereby obtaining the representative physical size of all detected sand and gravel particles one by one.

[0029] Based on the list of all physical particle sizes output by the particle size calculation unit, the system performs real-time particle size distribution statistics, generates particle size histograms and cumulative distribution curves, and accurately calculates key quality indicators including oversize rate, undersize rate, and medium size rate. At the same time, by comparing with the preset particle size distribution model, the system can automatically identify possible mixing or incorrect material abnormalities in the production line. All detection results, including real-time data, statistical charts, and alarm information, are visualized through the monitoring interface in the form of reports and real-time trend curves, and the complete data (including original images, processing results, and timestamps) is stored.

[0030] When the calculated oversize or undersize ratio exceeds the set threshold, the system will immediately trigger an audible and visual alarm to alert the operator. The system can also send adjustment commands to the PLC control systems of the crusher (e.g., adjusting the discharge port size) and vibrating screen (e.g., adjusting the vibration frequency or inclination angle) via industrial communication protocols such as OPC UA or Modbus TCP, thereby achieving automatic fine-tuning of production parameters and forming an efficient closed-loop quality control loop to ensure the long-term stability of aggregate gradation.

[0031] This invention provides Embodiment 1: Basalt aggregate produced by a sand and gravel processing system of a high-altitude hydropower station was selected, with a target gradation of 5-20mm continuous gradation. Multiple sets of samples with known particle size distribution were prepared as standard samples (the true particle size distribution was determined by precision sieving).

[0032] This embodiment sets up two sets of comparative examples. Comparative example 1 is the traditional manual sieving method, which involves manually sampling and reducing the size of the same batch of materials in strict accordance with national standards, and then performing sieving analysis in the laboratory using standard sieves. Comparative example 2 is the traditional image processing method, which uses traditional image processing technology based on threshold segmentation and edge detection, combined with the Watershed watershed algorithm to process adhered particles for particle size analysis.

[0033] In this embodiment 1, a complete online detection system is built according to the above embodiment. The experimental environment is achieved by constructing a test chamber in the laboratory that can simulate the plateau environment. This environmental chamber can adjust the temperature (range -10℃ to 30℃), simulate the intensity of strong ultraviolet light on the plateau, and inject a controllable concentration of dust to reproduce the dust conditions on site.

[0034] Under standard experimental conditions, 10 groups of standard samples were tested using Comparative Example 1, Comparative Example 2 and Example 1 of this embodiment, respectively, with each group of samples tested three times.

[0035] Using the results of Comparative Example 1 (manual sieving) as the baseline true value, the mean absolute error and root mean square error of Example 1 and Comparative Example 2 on the key particle size sieve residues (e.g., 4.75 mm, 9.5 mm, 16 mm, 19 mm sieves) were calculated. The time required to test a single batch of samples for the three examples was also compared. The specific results are shown in the table below:

[0036] The results show that Example 1, due to the use of deep learning for instance segmentation, has a much stronger ability to segment adhered and overlapping particles than the traditional image method in Comparative Example 2. Its detection accuracy is very close to the laboratory screening results, and the time consumption is extremely short, achieving high-precision online detection.

[0037] This invention provides Embodiment 2: The systems of Example 1 and Comparative Example 2 were placed in a simulated environment chamber, and the test samples were fixed. Stability tests were then conducted for 8 consecutive hours under the following three conditions: Low temperature conditions, ambient temperature -10℃, normal light, dust-free; Under strong light conditions, with an ambient temperature of 20°C, high-intensity simulated ultraviolet light was applied. High dust conditions, ambient temperature 20℃, normal lighting, injection of high-concentration dust to simulate on-site dust concentration.

[0038] Throughout the entire test process, it was recorded whether Example 1 could continuously and stably output images and perform detection, and the fluctuation range of particle size detection results (measured by standard deviation) was statistically analyzed and compared with the baseline results under standard conditions. The specific results are shown in the table below.

[0039]

[0040] The results show that, thanks to the optimization of the protective devices and preprocessing algorithms, the model can operate stably under all three harsh conditions. The image preprocessing steps effectively overcome the interference of light and dust. The Mask R-CNN model demonstrates strong robustness, with the standard deviation of the detection results showing only a slight increase compared to the standard environment, far less than Comparative Example 2, proving its excellent adaptability to the high-altitude environment.

[0041] This invention provides embodiment 3: In this embodiment, the system of the present invention was deployed on an actual sand and gravel production line of a high-altitude hydropower station for a 30-day continuous operation test. The test results were compared with the results of manual sampling and screening at fixed daily time points. Simultaneously, the system's closed-loop control function was activated, automatically adjusting the crusher parameters when the oversize ratio >5% or the undersize ratio >8%.

[0042] The correlation coefficient between the daily average particle size index of this invention and the results of manual screening was calculated, and the pass rate fluctuation of the key particle size distribution of aggregates (such as 5-10mm, 10-15mm, 15-20mm) was statistically analyzed over 30 days. At the same time, the number of times the system triggered automatic control and the average time for the index to return to normal after adjustment were recorded. The specific results are shown in the table below.

[0043]

[0044] The results show that the detection results of this invention are highly correlated with the results of manual screening, with a correlation coefficient R. 2The aggregate grade pass rate was >0.98, and under the closed-loop control of the system, the aggregate grade pass rate remained stable at over 98.5% within 30 days, with a fluctuation range far smaller than before the system deployment. The system triggered automatic control 15 times, and the average time from alarm to parameter adjustment and return to normal was about 10 minutes, which greatly shortened the quality fluctuation cycle, avoided the generation of a large number of unqualified products, and significantly improved the level of intelligence in production and the stability of product quality.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A machine vision-based online particle size detection system for sand and gravel aggregates used in high-altitude hydropower projects, characterized in that, Including: The image acquisition module, including an industrial camera and an LED light source, is fixedly installed above the belt conveyor to acquire images of sand and gravel aggregates in real time while the belt is running. The protection and installation module includes a protective device and a support structure. The protective device is designed to be resistant to low temperatures, dust, water, and ultraviolet rays, and encapsulates the image acquisition module. The support structure uses a vibration-resistant aluminum alloy frame to fix the image acquisition module and the protective device. The image processing and analysis module, including the image processing module and the data analysis module, is connected to the image acquisition module. The image processing module is used to preprocess the acquired images, segment the target, and calculate the particle size. The data analysis module is used to calculate the particle size distribution, oversize ratio, undersize ratio, and medium size ratio of sand and gravel aggregates in real time, and automatically identify mixed and mismatched materials. The feedback and control module, connected to the image processing and analysis module, is used to link with the crushing and screening equipment, adjust production parameters based on the detection results, and form a closed-loop control. The data storage and display module is used to present data in the form of reports, screening curves, and real-time curves, and archive it to the database; The edge computing server, located in the control room, communicates with the image processing module and the data analysis module to execute image processing algorithms and data calculations, avoiding exposure to the high-altitude outdoor environment.

2. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 1, characterized in that: The industrial camera in the image acquisition module is a high-speed industrial camera, capable of capturing clear images while the belt is running at high speed.

3. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 1, characterized in that: The LED light source in the image acquisition module is a uniform illumination source, used to provide compensating illumination under strong light conditions at high altitudes, reducing image shadows and reflections.

4. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 1, characterized in that: The protective device includes a sealed outer shell, a temperature compensation unit, and an ultraviolet filter cover, which are used to ensure the normal operation of internal equipment in high-altitude, low-temperature, strong ultraviolet and dusty environments.

5. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 1, characterized in that: The image processing module includes: The preprocessing unit is used to perform noise reduction, motion blur removal, brightness equalization, and gamma correction on the image to adapt to the high-altitude environment with strong light and dust. The target segmentation unit, based on the Mask R-CNN instance segmentation model, identifies and segments sand and gravel particles one by one, avoiding particle overlap or adhesion. The particle size calculation unit is used to perform morphological fitting on the segmented particles, calculate the equivalent ellipse Feret minor axis, and convert it into the actual physical particle size through a calibration plate.

6. A machine vision-based online particle size detection system for high-altitude hydropower aggregates according to any one of claims 1-5, characterized in that: The system uses an industrial camera and LED light source mounted above the belt conveyor to acquire real-time images of sand and gravel aggregates in a high-altitude environment. The acquired images are then denoised, motion blurred, brightness equalized, and gamma-corrected to eliminate the effects of strong light and dust at high altitudes. Next, an instance segmentation model based on Mask R-CNN is used to identify and segment the sand and gravel particles one by one. Morphological fitting is performed on the segmented particles to calculate the equivalent ellipse Feret minor axis, which is then converted to the actual physical particle size using a calibration plate. The system then calculates the particle size distribution, oversize rate, undersize rate, and median rate in real time, and determines whether there is mixing or incorrect material. Finally, a particle size histogram, sieve residue curve, and real-time data report are generated and stored in a database. When the oversize rate or undersize rate exceeds a set threshold, an alarm is triggered, and the parameters of the crusher and screening machine are automatically adjusted to achieve closed-loop control.

7. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 6, characterized in that: The acquired images also include a contrast enhancement operation, employing an adaptive histogram equalization method to improve image clarity under low-light conditions at high altitudes.

8. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 6, characterized in that: The Mask R-CNN model, trained on a dataset of sand and gravel images collected in a high-altitude environment, is able to identify sand and gravel particles under conditions of dust, shadow, and particle adhesion.

9. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 6, characterized in that, The particle size calculation specifically includes: fitting an equivalent ellipse to each segmented sand and gravel particle, then calculating the Feret minor axis of the ellipse as the representative particle size, and finally converting the pixel size into the actual physical size through a pre-calibrated pixel-physical size ratio.

10. The online particle size detection system for high-altitude hydropower aggregates based on machine vision according to claim 6, characterized in that: When the oversize ratio exceeds the threshold, the crusher discharge port is automatically reduced; when the undersize ratio exceeds the threshold, the vibration frequency or tilt angle of the screening machine is automatically adjusted to optimize the aggregate gradation in real time.

Images