Method for closing discharge port of leaked waste rock filling hydraulic support

Abstract

The invention discloses a method for closing a discharge port of a waste rock leakage filling hydraulic support. The method comprises the following steps that 1, images of waste rock falling from the discharge port are collected; 2, segmenting the falling gangue image, and calculating the projection area of the gangue in the image; 3, according to the projection area, the actual area of gangue falling from the discharging opening is converted; 4, according to the actual area of the falling gangue, the accumulated mass of the falling gangue at the discharging opening is measured and calculated; 5, calculating a filling gangue mass threshold value; 6, according to the filling gangue mass threshold value and the falling gangue accumulated mass, whether the falling gangue mass of the discharging opening meets the requirement or not is judged; and 7, when the quality of the gangue falling from the discharging opening meets the requirement, the discharging opening of the hydraulic support is closed. According to the method for closing the discharging opening of the waste rock leakage filling hydraulic support, the problems that in the filling control process of an existing waste rock leakage filling hydraulic support, the manual judgment error is large, and the filling amount cannot be monitored in real time are solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of coal mine filling mining, and particularly relates to a method for closing a discharge port of a gangue leakage filling hydraulic support. BACKGROUND

[0002] With the deep mining of coal resources and the promotion of green mine construction, solid filling mining technology is widely used because it can effectively control the roof subsidence, improve the resource recovery rate and process solid waste. In this technology, coal gangue is used as the main filling material, and the goaf is filled through the gangue leakage filling hydraulic support. However, there are still the following technical problems in the existing filling process: At present, the opening and closing of the discharge port of most gangue leakage filling hydraulic supports still rely on manual observation or time control mode for adjustment, and the filling state cannot be judged in real time according to the actual falling gangue quantity, which easily leads to excessive or insufficient filling, thereby affecting the goaf filling rate and the roof stability. At the same time, due to the large dust and poor illumination in the underground environment, the traditional sensor cannot accurately obtain the falling gangue mass data of each discharge port, which leads to the fact that the gangue leakage filling process cannot be accurately controlled and closed-loop adjusted, and the filling operation process is still mainly controlled by experience. In the existing technology, most filling systems only set total quantity monitoring devices on the ground or in the return airway, and cannot independently judge the filling state of each discharge port. At present, the gangue discharge mode is mainly grouped, and the single discharge port is closed in advance or continuously discharged, which easily causes uneven filling. SUMMARY

[0003] The purpose of the present application is to provide a method for closing the discharge port of a gangue leakage filling hydraulic support, which solves the problems of large manual judgment error and inability to monitor the filling quantity in real time in the filling control of the existing gangue leakage filling hydraulic support.

[0004] The technical solution adopted by the present application is a method for closing the discharge port of a gangue leakage filling hydraulic support, comprising the following steps: Step 1: collecting the image of the falling gangue of the discharge port; Step 2: segmenting the image of the falling gangue and calculating the projection area of the gangue in the image; Step 3: converting the actual area of the falling gangue of the discharge port according to the projection area; Step 4: measuring the cumulative mass of the falling gangue of the discharge port according to the actual area of the falling gangue; Step 5: calculating the filling gangue mass threshold; Step 6: judging whether the mass of the falling gangue of the discharge port meets the requirements according to the filling gangue mass threshold and the cumulative mass of the falling gangue; Step 7: closing the discharge port of the hydraulic support when the mass of the falling gangue of the discharge port meets the requirements.

[0005] The present application has the following characteristics: Step 1 is specifically: Step 1.1, install a high-speed camera under the top beam of the hydraulic support after the gangue filling; Step 1.2, real-time acquisition of gangue transient images falling from the discharge port of the multi-hole bottom discharge scraper conveyor by the high-speed camera, and the high-speed camera transmits the gangue transient images to the processor.

[0006] Step 1.3, after the processor receives the images transmitted by the high-speed camera, the gangue images are preprocessed for denoising, brightness equalization and contrast enhancement.

[0007] In step 1.1, one high-speed camera is installed on each gangue filling hydraulic support, and two discharge ports are monitored simultaneously; in step 1.2, the falling gangue is the filling gangue after being crushed by the crusher, the particle size range of the falling gangue is 40-60mm, the high-speed camera real-time acquires the transient image sequence of the gangue falling process with a fixed sampling period T, and transmits the falling image sequence to the processor; the sampling frequency of the high-speed camera is self-adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size.

[0008] Step 2 is specifically: Step 2.1, the processor performs image segmentation processing on the preprocessed gangue transient images based on the constructed image segmentation model; Step 2.2, after the segmentation processing, the processor extracts the effective pixel area of the gangue in the sampling area and calculates the projection area of the gangue in the image.

[0009] In step 2.1, the image segmentation model is a segmentation network model based on deep learning, the image segmentation model selects a SegFormer semantic segmentation model based on the Transformer structure, the SegFormer semantic segmentation model extracts accurate contour features of the gangue after being trained on a gangue sample dataset containing different particle sizes, illuminations and dust conditions, and outputs a binary segmentation mask image containing gangue pixels.

[0010] Step 3 is specifically: the processor uses a proportional conversion coefficient k , according to formula (1), converts the gangue pixel area S p in the binary segmentation mask image output by the image segmentation model into the projection area S r of the gangue in the actual space. (1); Proportional conversion coefficient kAccording to the installation height of the camera, the lens focal length and the field angle parameters, the camera calibration is determined in advance; the camera calibration process includes placing a calibration board with a known size under the camera installation position, and calculating the actual length value corresponding to a unit pixel by shooting.

[0011] Step 4 is specifically: assuming that the falling gangue is an approximate regular body, the volume of the gangue is calculated according to the set representative particle size parameter, and the cumulative mass of the gangue falling from the discharge port is calculated according to formulas (2) and (3); (2); (3); wherein, d is the particle size, p is the gangue density, m is the mass of the gangue in the sampling area, is the cumulative mass of the gangue falling from the discharge port; the gangue density parameter is set according to on-site sampling test, and is automatically corrected according to different seam conditions.

[0012] Step 5 is specifically: Step 5.1, according to the volume of the goaf V , the design filling rate φ and the gangue packing density after ramming p p , the total mass of the filling gangue under the design filling rate is calculated according to formula (4) M ; (4); Step 5.2, assuming that the current working face is completed by n sets of gangue leakage filling hydraulic supports, and two discharge ports are set for each support, the filling mass threshold value in the monitoring range of each high-speed camera is calculated according to formula (5); (5); Design filling rate φ According to the filling process, the coal seam thickness and the roof subsidence control requirements, the dynamic correction is carried out to ensure the uniformity of the goaf filling; the gangue packing density after ramming is obtained through the loose gangue compaction test.

[0013] Step 6 is specifically: the processor judges whether the gangue falling amount of the current discharge port meets the requirements according to the preset mass threshold value m t When the cumulative value of the falling gangue mass calculated in real time m c reaches the threshold value m t , it is determined that the gangue filling amount of the two discharge ports monitored by the high-speed camera has met the design requirements.

[0014] The step 7 is specifically: when the gangue quality cumulative value reaches the set threshold value, the processor sends a closing instruction to the hydraulic support control module to control the hydraulic actuator to close the corresponding discharge port, so that the automatic and quantitative closing control of the discharge port of the gangue filling hydraulic support is realized; the processor is connected with the hydraulic support control system through the Ethernet communication interface to realize the real-time transmission of the control instruction and the action closed loop.

[0015] The beneficial effects of the present application are: The gangue filling hydraulic support discharge port closing method provided by the present application arranges a high-speed camera below the top beam of the gangue filling hydraulic support to collect image information of the falling gangue of the discharge port in real time, extracts the actual area of the gangue by combining the image segmentation model, calculates the cumulative mass of the falling gangue according to the gangue density and particle size, and then compares it with the mass threshold value calculated according to the goaf volume, the designed filling rate and the stacking density. When the monitored cumulative mass reaches the threshold value, the system automatically determines that the filling amount has reached the design requirement, sends a closing instruction to the hydraulic control system, so that the accurate determination of the gangue filling process is realized. Through the present application, independent monitoring and accurate judgment of multiple discharge ports can be realized in the complex environment of underground high dust and low illumination, and the discharge ports can be closed, which effectively prevents the local under-filling or over-filling phenomenon of the goaf, significantly improves the filling uniformity and the automation level of the operation, and provides effective technical support for solid accurate filling mining. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flow chart of the gangue filling hydraulic support discharge port closing method of the present application; Figure 2 is a high-speed camera arrangement schematic diagram in embodiment 8 of the present application. DETAILED DESCRIPTION

[0017] The present application will be described in detail below in combination with specific embodiments.

[0018] Embodiment 1 The gangue filling hydraulic support discharge port closing method proposed in this embodiment, as shown in Figure 1 , includes the following steps: Step 1, collecting the image of the falling gangue of the discharge port; Step 2, segmenting the falling gangue image to calculate the projection area of the gangue in the image; Step 3, converting the actual area of the falling gangue of the discharge port according to the projection area; Step 4, calculating the cumulative mass of the falling gangue of the discharge port according to the actual area of the falling gangue; Step 5, calculating the filling gangue mass threshold value; Step 6, judging whether the falling gangue mass of the discharge port reaches the requirement according to the filling gangue mass threshold value and the cumulative mass of the falling gangue; Step 7, when the quality of the falling gangue at the discharge port meets the requirements, close the discharge port of the hydraulic support.

[0019] Embodiment 2 The gangue leakage filling hydraulic support discharge port closing method proposed in this embodiment, as shown in Figure 1 , includes the following steps: Step 1, collect the falling gangue image at the discharge port; Specifically: Step 1.1, install a high-speed camera under the rear top beam of the gangue leakage filling hydraulic support; In step 1.1, one high-speed camera is installed for each gangue leakage filling hydraulic support, and two discharge ports are monitored at the same time; in step 1.2, the falling gangue is the filling gangue after being crushed by a crusher, the particle size range of the falling gangue is 40-60 mm, the high-speed camera collects the transient image sequence of the gangue falling process in real time with a fixed sampling period T, and transmits the falling image sequence to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size; Step 1.2, collect the transient image of the falling gangue at the discharge port of the multi-hole bottom discharge type scraper conveyor through the high-speed camera, and transmit the transient image of the gangue to the processor by the high-speed camera; Step 1.3, after the processor receives the image transmitted by the high-speed camera, the gangue image is preprocessed for denoising, brightness equalization and contrast enhancement; Step 2, segment the falling gangue image, and calculate the projection area of the gangue in the image; Step 3, convert the actual area of the falling gangue at the discharge port according to the projection area; Step 4, calculate the cumulative mass of the falling gangue at the discharge port according to the actual area of the falling gangue; Step 5, calculate the filling gangue mass threshold; Step 6, determine whether the quality of the falling gangue at the discharge port meets the requirements according to the filling gangue mass threshold and the cumulative mass of the falling gangue; Step 7, when the quality of the falling gangue at the discharge port meets the requirements, close the discharge port of the hydraulic support.

[0020] Embodiment 3 The gangue leakage filling hydraulic support discharge port closing method proposed in this embodiment, as shown in Figure 1 , includes the following steps: Step 1, collect the falling gangue image at the discharge port; Specifically: Step 1.1, install a high-speed camera under the rear top beam of the gangue leakage filling hydraulic support; In step 1.1, one high-speed camera is installed on each gangue filling hydraulic support to monitor two discharge ports; in step 1.2, the falling gangue is the filling gangue crushed by a crusher, the particle size range of the falling gangue is 40-60 mm, the high-speed camera collects the transient image sequence of the gangue falling process in real time at a fixed sampling period T, and the falling image sequence is transmitted to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size; In step 1.2, the high-speed camera collects the transient image of the falling gangue at the discharge port of the multi-hole bottom discharge type scraper conveyor in real time, and the high-speed camera transmits the transient image of the gangue to the processor. In step 1.3, after the processor receives the image transmitted by the high-speed camera, the gangue image is preprocessed by denoising, brightness equalization and contrast enhancement; In step 2, the falling gangue image is segmented, and the projection area of the gangue in the image is calculated. Specifically, In step 2.1, the processor performs image segmentation processing on the preprocessed transient image of the gangue based on the constructed image segmentation model. In step 2.1, the image segmentation model is a segmentation network model based on deep learning, the image segmentation model selects a SegFormer semantic segmentation model based on a Transformer structure, the SegFormer semantic segmentation model extracts accurate contour features of the gangue after being trained on a gangue sample data set containing different particle sizes, illuminations and dust conditions, and outputs a binary segmentation mask image containing gangue pixels. In step 2.2, after the segmentation processing, the processor extracts the effective pixel area of the gangue in the sampling area, and calculates the projection area of the gangue in the image. In step 3, the actual area of the falling gangue at the discharge port is converted according to the projection area. In step 4, the cumulative mass of the falling gangue at the discharge port is calculated according to the actual area of the falling gangue. In step 5, the filling gangue mass threshold is calculated. In step 6, whether the mass of the falling gangue at the discharge port meets the requirements is judged according to the filling gangue mass threshold and the cumulative mass of the falling gangue. In step 7, when the mass of the falling gangue at the discharge port meets the requirements, the discharge port of the hydraulic support is closed.

[0021] Embodiment 4 The gangue falling hydraulic support discharge port closing method proposed in this embodiment, as shown in Figure 1 includes the following steps: Step 1, collect the falling gangue image of the discharge port; Specifically, Step 1.1, install a high-speed camera under the top beam of the hydraulic support after the gangue filling; In step 1.1, one high-speed camera is installed on each gangue filling hydraulic support, and two discharge ports are monitored at the same time; in step 1.2, the falling gangue is the filling gangue after being crushed by a crusher, the particle size range of the falling gangue is 40-60 mm, the high-speed camera collects the transient image sequence of the gangue falling process in real time with a fixed sampling period T, and transmits the falling image sequence to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size; Step 1.2, real-time collection of falling gangue transient images at the discharge port of the multi-hole bottom discharge type scraper conveyor by the high-speed camera, and transmission of the gangue transient images to the processor by the high-speed camera; Step 1.3, after the processor receives the images transmitted by the high-speed camera, the gangue images are preprocessed for denoising, brightness equalization and contrast enhancement; Step 2, segmenting the falling gangue image and calculating the projection area of the gangue in the image; Specifically: Step 2.1, the processor performs image segmentation processing on the preprocessed gangue transient images based on the constructed image segmentation model; In step 2.1, the image segmentation model is a segmentation network model based on deep learning, the image segmentation model selects a SegFormer semantic segmentation model based on a Transformer structure, the SegFormer semantic segmentation model extracts accurate contour features of the gangue after being trained on a gangue sample data set containing different particle sizes, illuminations and dust conditions, and outputs a binary segmentation mask containing gangue pixels; Step 2.2, after segmentation processing, the processor extracts the effective pixel area of the gangue in the sampling area and calculates the projection area of the gangue in the image; Step 3, converting the actual area of the falling gangue at the discharge port according to the projection area; Specifically, the processor uses a proportional conversion coefficient k according to formula (1) to convert the gangue pixel area in the binary segmentation mask output by the image segmentation model S p to the projection area of the gangue in the actual space S r ; (1); the proportional conversion coefficient kThe camera is pre-calibrated based on its installation height, lens focal length, and field of view. The camera calibration process includes placing a calibration plate of known dimensions below the camera installation position and calculating the actual length value corresponding to each pixel by taking pictures. Step 4: Calculate the cumulative mass of the falling gangue at the unloading port based on the actual area of ​​the falling gangue; Step 5: Calculate the backfill gangue quality threshold; Step 6: Based on the backfill gangue quality threshold and the cumulative mass of falling gangue, determine whether the mass of falling gangue at the discharge port meets the requirements; Step 7: When the quality of the gangue falling from the discharge port meets the requirements, close the discharge port of the hydraulic support.

[0022] Example 5 The method for closing the unloading port of the hydraulic support for backfilling leaking gangue proposed in this embodiment is as follows: Figure 1 As shown, it includes the following steps: Step 1: Acquire images of the falling gangue at the unloading port; Specifically: Step 1.1: Install a high-speed camera below the top beam after the hydraulic support for backfilling the leaking gangue; In step 1.1, a high-speed camera is installed on each hydraulic support for backfilling leaking gangue, simultaneously monitoring two unloading ports. In step 1.2, the falling gangue is the backfilling gangue after being crushed by a crusher, with a particle size range of 40~60mm. The high-speed camera acquires transient image sequences of the gangue falling process in real time at a fixed sampling period T, and transmits the falling image sequences to the processor. The sampling frequency of the high-speed camera is adaptively adjusted according to the falling speed of the gangue to ensure that the displacement of gangue between consecutive frames does not exceed 20% of the particle size. Step 1.2: Real-time transient images of the gangue falling from the discharge port of the multi-hole bottom discharge scraper conveyor are acquired using a high-speed camera, and the high-speed camera transmits the transient images of the gangue to the processor. Step 1.3: After receiving the image transmitted from the high-speed camera, the processor performs noise reduction, brightness equalization, and contrast enhancement preprocessing on the gangue image; Step 2: Segment the image of falling gangue and calculate the projected area of ​​the gangue in the image; Specifically: Step 2.1: Based on the constructed image segmentation model, the processor performs image segmentation processing on the preprocessed transient image of gangue; In step 2.1, the image segmentation model is a segmentation network model based on deep learning. The image segmentation model selected is the SegFormer semantic segmentation model based on the Transformer structure. After training on a gangue sample dataset containing different particle sizes, illumination and dust conditions, the SegFormer semantic segmentation model extracts the accurate contour features of the gangue and outputs a binary segmentation mask containing gangue pixels. Step 2.2: After segmentation, the processor extracts the effective pixel area of ​​gangue within the sampling area and calculates the projected area of ​​gangue in the image. Step 3: Calculate the actual area of ​​the waste rock falling from the unloading port based on the projected area; Specifically, the processor uses a proportional conversion factor based on the fixed distance between the camera and the falling area and the actual area of ​​the camera's sampling area. k According to equation (1), the area of ​​the gangue pixels in the binary segmentation mask output by the image segmentation model is... S p This is converted into the projected area of ​​the gangue in actual space. S r ; (1); Proportional conversion factor k The camera is pre-calibrated based on its installation height, lens focal length, and field of view. The camera calibration process includes placing a calibration plate of known dimensions below the camera installation position and calculating the actual length value corresponding to each pixel by taking pictures. Step 4: Calculate the cumulative mass of the falling gangue at the unloading port based on the actual area of ​​the falling gangue; Specifically: Assuming the falling gangue is an approximately regular body, the gangue volume is calculated based on the set representative particle size parameters, and the cumulative mass of the falling gangue at the discharge port is calculated according to formulas (2) and (3); (2); (3); in, d For particle size, p Density of gangue m The mass of gangue in the sampling area. The cumulative mass of gangue falling from the unloading port; the gangue density parameter is set based on on-site sampling and testing, and is automatically corrected according to different ore layer conditions; Step 5: Calculate the backfill gangue quality threshold; Step 6: Based on the backfill gangue quality threshold and the cumulative mass of falling gangue, determine whether the mass of falling gangue at the discharge port meets the requirements; Step 7: When the quality of the gangue falling from the discharge port meets the requirements, close the discharge port of the hydraulic support.

[0023] Example 6 The method for closing the unloading port of the hydraulic support for backfilling leaking gangue proposed in this embodiment is as follows: Figure 1 As shown, it includes the following steps: Step 1: Acquire images of the falling gangue at the unloading port; Specifically: Step 1.1, install a high-speed camera under the top beam of the hydraulic support after the gangue filling; In step 1.1, one high-speed camera is installed on each gangue filling hydraulic support, and two discharge ports are monitored at the same time; in step 1.2, the falling gangue is the filling gangue after being crushed by a crusher, the particle size range of the falling gangue is 40-60 mm, the high-speed camera collects the transient image sequence of the gangue falling process in real time with a fixed sampling period T, and transmits the falling image sequence to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size; Step 1.2, real-time collection of falling gangue transient images at the discharge port of the multi-hole bottom discharge type scraper conveyor by the high-speed camera, and transmission of the gangue transient images to the processor by the high-speed camera; Step 1.3, after the processor receives the images transmitted by the high-speed camera, the gangue images are preprocessed for denoising, brightness equalization and contrast enhancement; Step 2, segmenting the falling gangue image and calculating the projection area of the gangue in the image; Specifically: Step 2.1, the processor performs image segmentation processing on the preprocessed gangue transient images based on the constructed image segmentation model; In step 2.1, the image segmentation model is a segmentation network model based on deep learning, the image segmentation model selects a SegFormer semantic segmentation model based on a Transformer structure, the SegFormer semantic segmentation model extracts accurate contour features of the gangue after being trained on a gangue sample data set containing different particle sizes, illuminations and dust conditions, and outputs a binary segmentation mask containing gangue pixels; Step 2.2, after segmentation processing, the processor extracts the effective pixel area of the gangue in the sampling area and calculates the projection area of the gangue in the image; Step 3, converting the actual area of the falling gangue at the discharge port according to the projection area; Specifically, the processor uses a proportional conversion coefficient k according to formula (1) to convert the gangue pixel area in the binary segmentation mask output by the image segmentation model S p to the projection area of the gangue in the actual space S r ; (1); the proportional conversion coefficient kAccording to the camera installation height, lens focal length and field angle parameters, the camera calibration is determined in advance; the camera calibration process includes placing a calibration board with a known size under the camera installation position, and calculating the actual length value corresponding to a unit pixel by shooting; Step 4, according to the actual area of the falling gangue, the cumulative mass of the falling gangue at the discharge port is calculated; Specifically, it is assumed that the falling gangue is an approximate regular body, the volume of the gangue is calculated according to the set representative particle size parameter, and the cumulative mass of the falling gangue at the discharge port is calculated according to formulas (2) and (3); (2); (3); wherein, d is the particle size, p is the gangue density, m is the mass of the gangue in the sampling area, is the cumulative mass of the falling gangue at the discharge port; the gangue density parameter is set according to on-site sampling test and automatically corrected according to different seam conditions; Step 5, calculate the filling gangue mass threshold; Specifically, Step 5.1, according to the volume of the goaf V , the design filling rate φ and the gangue packing density after ramming p p , the total filling gangue mass under the design filling rate is calculated according to formula (4) M ; (4); Step 5.2, assuming that the current working face is completed by n sets of gangue leakage filling hydraulic supports, and each set of support is provided with two discharge ports, the filling mass threshold in the monitoring range of each high-speed camera is calculated according to formula (5); (5); The design filling rate φ is dynamically corrected according to the filling process, coal seam thickness and roof subsidence control requirements to ensure the uniformity of the goaf filling; the gangue packing density after ramming is obtained through the loose gangue compaction test; Step 6, according to the filling gangue mass threshold and the cumulative mass of the falling gangue, it is judged whether the mass of the falling gangue at the discharge port meets the requirements; Step 7, when the mass of the falling gangue at the discharge port meets the requirements, the discharge port of the hydraulic support is closed.

[0024] Example 7 The gangue leakage filling hydraulic support discharge port closing method proposed in this embodiment, as shown in Figure 1 , includes the following steps: Step 1, collect the falling gangue image of the discharge port; Specifically, Step 1.1, install a high-speed camera under the top beam of the gangue filling hydraulic support; In step 1.1, one high-speed camera is installed on each gangue filling hydraulic support, and two discharge ports are monitored at the same time; in step 1.2, the falling gangue is the filling gangue after being crushed by a crusher, the particle size range of the falling gangue is 40-60 mm, the high-speed camera collects the transient image sequence of the gangue falling process in real time with a fixed sampling period T, and transmits the falling image sequence to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the gangue falling speed to ensure that the gangue displacement between consecutive frames does not exceed 20% of the particle size; Step 1.2, real-time collection of falling gangue transient images of multi-hole bottom discharge type scraper conveyor discharge port by high-speed camera, and transmission of gangue transient images to processor by high-speed camera; Step 1.3, after the processor receives the images transmitted by the high-speed camera, the gangue images are preprocessed for denoising, brightness equalization and contrast enhancement; Step 2, segment the falling gangue image and calculate the projection area of the gangue in the image; Specifically, Step 2.1, the processor performs image segmentation processing on the preprocessed gangue transient image based on the constructed image segmentation model; In step 2.1, the image segmentation model is a segmentation network model based on deep learning, the image segmentation model selects a SegFormer semantic segmentation model based on a Transformer structure, the SegFormer semantic segmentation model extracts accurate contour features of the gangue after being trained on a gangue sample data set containing different particle sizes, illuminations and dust conditions, and outputs a binary segmentation mask containing gangue pixels; Step 2.2, after segmentation processing, the processor extracts the effective pixel area of the gangue in the sampling area and calculates the projection area of the gangue in the image; Step 3, according to the projection area, the actual area of the falling gangue of the discharge port is converted; Specifically, the processor converts the pixel area of the gangue in the binary segmentation mask output by the image segmentation model according to formula (1) by using the proportional conversion coefficient k , according to formula (1), the pixel area of the gangue in the binary segmentation mask output by the image segmentation model S p , is converted into the projection area of the gangue in the actual space S r ; (1); Proportional conversion coefficient kAccording to the camera installation height, lens focal length and field angle parameters, the camera calibration is determined in advance; the camera calibration process includes placing a calibration board with a known size under the camera installation position, and calculating the actual length value corresponding to a unit pixel by shooting; Step 4, according to the actual area of the falling gangue, the cumulative mass of the falling gangue at the discharge port is calculated; Specifically, it is assumed that the falling gangue is an approximate regular body, the volume of the gangue is calculated according to the set representative particle size parameter, and the cumulative mass of the falling gangue at the discharge port is calculated according to formula (2) and (3); (2); (3); Wherein, d is the particle size, p is the gangue density, m is the mass of the gangue in the sampling area, is the cumulative mass of the falling gangue at the discharge port; the gangue density parameter is set according to on-site sampling test, and is automatically corrected according to different seam conditions; Step 5, calculate the filling gangue mass threshold; Specifically, Step 5.1, according to the volume of the goaf V , the design filling rate φ and the gangue packing density after ramming p p , the total filling gangue mass under the design filling rate is calculated according to formula (4) M ; (4); Step 5.2, assuming that the current working face is completed by n gangue leakage filling hydraulic supports, and each support is provided with two discharge ports, the filling mass threshold in the monitoring range of each high-speed camera is calculated according to formula (5); (5); The design filling rate φ is dynamically corrected according to the filling process, coal seam thickness and roof subsidence control requirements to ensure the uniformity of the goaf filling; the gangue packing density after ramming is obtained through loose gangue compaction test; Step 6, according to the filling gangue mass threshold and the cumulative mass of the falling gangue, whether the mass of the falling gangue at the discharge port meets the requirements is judged; Specifically, the processor judges whether the falling amount of the gangue at the current discharge port meets the requirements according to the preset mass threshold m t , when the cumulative value of the falling gangue mass calculated in real time m c reaches the threshold m tIf the high-speed camera detects that the amount of gangue filling at the two unloading ports has reached the design requirements, then it is determined that the amount of gangue filling at the two unloading ports has reached the design requirements. Step 7: When the quality of the gangue falling from the discharge port meets the requirements, close the discharge port of the hydraulic support; Specifically, when the accumulated mass of gangue reaches a set threshold, the processor sends a closing command to the hydraulic support control module, controlling the hydraulic actuator to close the corresponding unloading port, thereby realizing automated and quantitative closing control of the unloading port of the hydraulic support for backfilling gangue; the processor connects to the hydraulic support control system through an Ethernet communication interface to realize real-time transmission of control commands and closed-loop operation.

[0025] Example 8 The method for closing the unloading port of the hydraulic support for backfilling leaking gangue proposed in this embodiment is as follows: Figure 1 As shown, it includes the following steps: like Figure 2 As shown, the system of this invention includes: a hydraulic support for backfilling leaked gangue, a high-speed camera located below the rear top beam of the hydraulic processor support for backfilling leaked gangue, a multi-hole bottom-discharge scraper conveyor, discharge ports, and a control module. Each hydraulic support for backfilling leaked gangue has two discharge ports located below it. The high-speed camera is installed at the center below the rear top beam, enabling simultaneous monitoring of both discharge ports. The high-speed camera is connected to an image processor, which in turn communicates with the control module. The control module then links with the execution control system of the hydraulic support to achieve automatic opening and closing of the discharge ports. The specific implementation steps are as follows: Step 1: Installation and deployment of image acquisition devices; High-speed cameras are fixedly installed below the top beam of the hydraulic support for backfilling the leaking gangue. Each high-speed camera monitors the falling area of ​​the discharge ports of two multi-hole bottom-discharge scraper conveyors. The camera lens faces the falling gangue area, with a fixed installation distance of 1.5m, and the field of view covers the falling trajectory range of the two discharge ports. The resolution of the high-speed camera is set to 1280×1024, the frame rate is 500FPS, and it stably acquires clear images under high dust and low light conditions. Step 2: Transient image acquisition and data transmission; Once the discharge port is opened, the gangue is conveyed by a scraper conveyor to the area below the gangue support. A high-speed camera acquires images at a fixed sampling period T. The camera is installed at a distance of 1.5m, covering the drop area of ​​both discharge ports. A single frame image can completely capture the distribution of gangue drop within this range. The system is set to acquire one frame image every T=1s and transmit it to an industrial processor for analysis in real time. Within each sampling period, the system accumulates the mass of gangue drop within the sampling range based on the image recognition results. Step 3: Image processing and gangue identification; The industrial processor denoises and enhances the collected image, and then uses a pre-trained image segmentation model (such as the SegFormer network) to segment the gangue area in the image to obtain the pixel area of the gangue S p . By knowing the geometric relationship between the camera and the sampling area, an area conversion coefficient k is established to convert the pixel area to the actual area S r : (1); wherein k is obtained according to the camera resolution and the distance calibration.

[0026] Step 4, calculation of the volume and mass of the falling gangue; The gangue block is approximated as a cube or a regular block, the particle size is taken as the average value d = 50 mm, and the gangue bulk density p = 2.1 × 10 3 kg / m 3 is combined to obtain the mass of the falling gangue in the sampling area: (2); (3); Step 5, calculation of the gangue mass threshold for filling; According to the volume of the goaf V , the designed filling rate φ , and the gangue bulk density after ramming p p , the total mass of the filling gangue under the designed filling rate is calculated: (4); Suppose that the current working face is completed by n gangue leakage filling hydraulic supports, and each support is provided with two discharge ports, then the filling mass threshold in the monitoring range of each high-speed camera is (5); that is, when the cumulative filling gangue mass in the double discharge port area monitored by a high-speed camera reaches m t , it is considered that the filling task of the monitoring area is completed.

[0027] Step 6, judgment of the gangue mass threshold falling from the discharge port; The processor judges whether the gangue falling amount of the current discharge port meets the requirements according to the preset mass threshold m t , when the cumulative value of the real-time calculated falling gangue mass m c reaches the threshold m tIf the high-speed camera detects that the amount of gangue filling at the two unloading ports has reached the design requirements, then it is determined that the amount of gangue filling at the two unloading ports has reached the design requirements. Step 7: Accurately close and control the discharge port; When the accumulated mass of gangue reaches the set threshold, the processor sends a closing command to the hydraulic support control module, which controls the hydraulic actuator to close the corresponding unloading port, thereby realizing the automated, quantitative and accurate closing control of the unloading port of the hydraulic support for backfilling gangue. Taking a backfilling mining face in a certain mine as an example, its main parameters are: goaf volume V = 1200 m³ 3 Design filling rate φ The compacted bulk density of the gangue is 0.8. p p =2.0×10 3 kg / m 3 The number of hydraulic supports for backfilling the leaking gangue is n=100. Each support is equipped with one high-speed camera, and the sampling field of view covers the gangue drop range of both unloading ports. The backfilling gangue particle size is 0.05m, and the camera sampling conversion factor k=4.5×10⁻⁶. -6 m 2 / pixel. From the formula M=V× φ × p p =1200×0.8×2.0×10 3 kg = 1.92 × 10 6 kg, the total filling mass is 1920 tons; The filling quality threshold for each high-speed camera is: ; During the filling process, assume that the area segmented within the rockfall region sampled by the high-speed camera is... S p =5.2×10 5 pixels, then: ; Based on a gangue particle size of d=0.05m, the volume of gangue in the current sampling area; ; The processor calculates the cumulative mass within the current camera's sampling range in real time. When the detection result reaches 9600 kg, the system immediately issues a shutdown command, and the hydraulic control unit executes the action to close the corresponding two discharge ports. If the cumulative mass does not reach the threshold, sampling and calculation continue until the filling volume meets the design requirements.

[0028] Therefore, the application can automatically determine the closing time according to image recognition and quality accumulation, realizes accurate and self-adaptive control of the discharge port of the hydraulic support for gangue filling, effectively avoids overfilling or underfilling, and significantly improves the filling efficiency and uniformity.

Claims

1. A method for closing the unloading port of a hydraulic support for backfilling leaking gangue, characterized in that, Includes the following steps: Step 1: Acquire images of the falling gangue at the unloading port; Step 2: Segment the image of falling gangue and calculate the projected area of ​​the gangue in the image; Step 3: Calculate the actual area of ​​the waste rock falling from the unloading port based on the projected area; Step 4: Calculate the cumulative mass of the falling gangue at the unloading port based on the actual area of ​​the falling gangue; Step 5: Calculate the backfill gangue quality threshold; Step 6: Based on the backfill gangue quality threshold and the cumulative mass of falling gangue, determine whether the mass of falling gangue at the discharge port meets the requirements; Step 7: When the quality of the gangue falling from the discharge port meets the requirements, close the discharge port of the hydraulic support.

2. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 1, characterized in that, Step 1 specifically involves: Step 1.1: Install a high-speed camera below the top beam after the hydraulic support for backfilling the leaking gangue; Step 1.2: Real-time transient images of the gangue falling from the discharge port of the multi-hole bottom discharge scraper conveyor are acquired using a high-speed camera, and the high-speed camera transmits the transient images of the gangue to the processor. Step 1.3: After receiving the image transmitted from the high-speed camera, the processor performs noise reduction, brightness equalization and contrast enhancement preprocessing on the gangue image.

3. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 2, characterized in that, In step 1.1, a high-speed camera is installed on each leaking gangue filling hydraulic support to simultaneously monitor two unloading ports; in step 1.2, the falling gangue is the filling gangue after being crushed by a crusher, and the particle size range of the falling gangue is 40~60mm. The high-speed camera collects transient image sequences of the gangue falling process in real time with a fixed sampling period T, and transmits the falling image sequences to the processor; the sampling frequency of the high-speed camera is adaptively adjusted according to the falling speed of the gangue to ensure that the displacement of the gangue between consecutive frames does not exceed 20% of the particle size.

4. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 3, characterized in that, Step 2 specifically involves: Step 2.1: Based on the constructed image segmentation model, the processor performs image segmentation processing on the preprocessed transient image of gangue; Step 2.2: After segmentation, the processor extracts the effective pixel area of ​​gangue within the sampling area and calculates the projected area of ​​gangue in the image.

5. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 4, characterized in that, In step 2.1, the image segmentation model is a segmentation network model based on deep learning. The image segmentation model selected is the SegFormer semantic segmentation model based on the Transformer structure. After training on a gangue sample dataset containing different particle sizes, illumination and dust conditions, the SegFormer semantic segmentation model extracts the accurate contour features of the gangue and outputs a binary segmentation mask containing the gangue pixels.

6. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 5, characterized in that, Step 3 specifically involves the processor using a proportional conversion factor based on the fixed distance between the camera and the falling area and the actual area of ​​the camera's sampling area. k According to equation (1), the area of ​​the gangue pixels in the binary segmentation mask output by the image segmentation model is... S p This is converted into the projected area of ​​the gangue in actual space. S r ; （1）； The proportional conversion factor k The camera is pre-calibrated based on its installation height, lens focal length, and field of view parameters. The camera calibration process includes placing a calibration plate of known size below the camera installation position and calculating the actual length value corresponding to each unit pixel by taking pictures.

7. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 6, characterized in that, The specific steps of step 4 are as follows: assuming that the falling gangue is an approximately regular body, the gangue volume is calculated according to the set representative particle size parameters, and the cumulative mass of the falling gangue at the discharge port is calculated according to formulas (2) and (3). （2）； （3）； in, d For particle size, ρ Density of gangue m The mass of gangue in the sampling area. The cumulative mass of gangue falling from the unloading port; the gangue density parameter is set based on on-site sampling and testing, and is automatically corrected according to different ore layer conditions.

8. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 7, characterized in that, Step 5 specifically involves: Step 5.1: Based on the volume of the goaf V Design filling rate φ and compacted gangue stock density ρ p The total mass of the backfilled gangue under the design filling ratio is calculated according to formula (4). M ; （4）； Step 5.2: Assume that the current working face is filled by n hydraulic supports for filling loose gangue. Each support has two unloading ports. Calculate the filling quality threshold within the monitoring range of each high-speed camera according to formula (5). （5）； The design fill rate φ Dynamic adjustments are made based on the filling process, coal seam thickness, and roof settlement control requirements to ensure uniform filling of the goaf; the compacted gangue bulk density is obtained through loose gangue compaction tests.

9. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 8, characterized in that, Step 6 specifically involves the processor determining a preset quality threshold. m t To determine whether the current amount of gangue falling from the unloading port meets the requirements, the cumulative mass of falling gangue is calculated in real time. m c Reaching the threshold m t If the high-speed camera detects that the amount of gangue filling at the two unloading ports has reached the design requirements, it is determined that the amount of gangue filling at the two unloading ports has reached the design requirements.

10. The method for closing the unloading port of the hydraulic support for backfilling leaking gangue according to claim 9, characterized in that, Step 7 specifically involves the following steps: when the accumulated mass of gangue reaches a set threshold, the processor sends a closing command to the hydraulic support control module to control the hydraulic actuator to close the corresponding unloading port, thereby achieving automated and quantitative closing control of the unloading port of the hydraulic support for backfilling gangue; the processor is connected to the hydraulic support control system via an Ethernet communication interface to achieve real-time transmission of control commands and closed-loop operation.

Images