An intelligent spraying dust-settling method and system based on segmented image recognition and sensor fusion

Abstract

The present application relates to a kind of intelligent spray dust-settling method and system based on segmented image recognition and sensor fusion, belong to mine safety environmental protection technical field.The present application aims at solving the technical problems such as low intelligent level of existing spray dust-settling mode, personnel mis-spray and poor dust space perception accuracy.Technical scheme includes: using visual module to divide the monitoring field of view into left, middle and right three sections, and running personnel detection algorithm and dust concentration grading logic in parallel;Through the fusion of image definition, gray value and sensor reading, comprehensive score is calculated, and concentration grading determination in sub-region is realized;According to the subarea result, linkage control of dust rising and stopping is realized.The present application realizes accurate dust-settling, improves dust-settling efficiency greatly while ensuring the comfort of operating personnel, reduces system cost and enhances reliability in harsh environment.

Description

Technical Field

[0001] This invention belongs to the field of mine safety and environmental protection technology, and relates to an intelligent spray dust suppression method and system based on segmented image recognition and sensor fusion. Background Technology

[0002] Mining processes, including drilling, blasting, crushing, transportation, and loading / unloading, generate large amounts of dust. This dust not only pollutes the working environment, accelerates equipment wear, and reduces visibility, leading to safety accidents, but also poses a threat to the health of on-site workers, being a leading cause of occupational diseases such as pneumoconiosis. Therefore, effectively controlling mine dust is a crucial task for ensuring safe production and promoting the construction of green mines.

[0003] Currently, dust control in mines mainly relies on spray dust suppression technology. However, existing spray control methods suffer from problems such as low level of intelligence, poor precision, and significant resource waste.

[0004] First, the manual control mode relies too heavily on the experience of staff to observe the field and start or stop the spraying, resulting in a delayed response. Furthermore, long-term exposure of personnel to high dust environments poses extremely high health risks.

[0005] Secondly, the timed automatic control mode causes the equipment to start and stop at preset time intervals, making it impossible to detect the actual dust concentration. The equipment continues to spray during dust-free or low-dust periods, resulting in a significant waste of water and electricity.

[0006] Furthermore, existing automatic dust control modes typically employ a combination of dust concentration sensors and passive infrared (PIR) sensors for monitoring. When the dust concentration exceeds the limit, spraying begins; if personnel are detected passing by, spraying stops and a countdown timer starts. While this mode combines dust suppression and spray prevention functions, it has significant drawbacks: 1. Regarding personnel detection, passive infrared PIR sensors have a narrow detection range and cannot accurately determine whether a person is within the spray area. While increasing sensor deployment and setting a countdown timer to stop spraying can compensate for some shortcomings, it is impossible to quickly restart the spray after personnel leave, significantly reducing dust suppression time. Furthermore, when personnel remain stationary within the area, passive infrared PIR sensors cannot detect their presence, potentially causing them to be sprayed and delaying work.

[0007] 2. Regarding dust concentration detection, dust concentration sensors are point-based measurements, only reflecting the concentration near the installation point and unable to detect the spatial concentration distribution within the area. When used with a single spray actuator, the accuracy of spray coverage is typically low.

[0008] Therefore, there is an urgent need in this field for an intelligent spray dust suppression solution that can accurately balance dust suppression efficiency, resource conservation, and user-friendly operation. This solution needs to achieve real-time, segmented sensing of dust spatial distribution and the presence of people in the area, thereby achieving the intelligent control goal of spraying as soon as dust starts to rise, stopping as soon as dust clears, avoiding areas with people, and restarting when people leave. Summary of the Invention

[0009] In view of this, the purpose of the present invention is to provide an intelligent spray dust suppression method and system based on segmented image recognition and sensor fusion.

[0010] To achieve the above objectives, the present invention provides the following technical solution: A smart spray dust suppression method based on segmented image recognition and sensor fusion includes the following steps: Step 1, System Initialization: Perform hardware initialization on the vision module and dust concentration sensor, and load the preset personnel detection model; Step 2, Image Acquisition and Region Division: Real-time image information is captured by the vision module, and the monitoring field of view of the vision module is divided into three independent monitoring regions in the horizontal direction: left, middle and right. Each monitoring region corresponds to an independent spray execution unit. Step 3, Personnel Detection by Zone: Run a personnel detection algorithm based on the lightweight object detection (FasterObjects, More Objects, FOMO) model in each monitoring area to determine in real time whether there are personnel in each monitoring area and obtain the personnel presence status; Step 4, Dust Concentration Classification: Extract image features from each monitoring area and obtain sensor readings from the dust concentration sensor. Calculate the comprehensive concentration score for each monitoring area using a fusion decision algorithm. According to the comprehensive concentration score The threshold range in which the dust concentration is located divides the dust concentration into two levels: high and low. Step 5, Intelligent Linkage Control: Based on the presence status of personnel and the dust concentration level of each monitoring area, a logical judgment is made. If and only if the dust concentration level in a certain monitoring area is high and there are no personnel in that monitoring area, the spray execution unit corresponding to that monitoring area is triggered to start spraying through the general purpose input / output (GPIO) pin.

[0011] Furthermore, in step four, the extracted image features include the reciprocal of image sharpness. and average gray value The image sharpness is achieved by calculating the variance of the Laplacian operator.

[0012] Furthermore, in step four, the comprehensive concentration score... The calculation formula is:

[0013] In the formula, , , These are the weighting coefficients for the reciprocal of sharpness, the average gray value, and the sensor reading, respectively. The reading is the sensor reading of the dust concentration sensor.

[0014] Furthermore, the weighting coefficients , , Determined through the following steps: Acquire multiple sets of sample data under different dust concentration conditions. Each set of sample data includes the reciprocal of the sharpness acquired simultaneously. Average gray value Auxiliary dust monitor readings and the calibrated standard concentration values ; A system of linear equations is constructed from the multiple sets of sample data, and the weighting coefficients are obtained by solving the least squares method. , , .

[0015] Furthermore, the comprehensive concentration score The threshold range is determined through the following steps: Based on the legal dust concentration limit, the concentration score corresponding to each sample data is calculated using the aforementioned weighting coefficients. ; exist Find all standard concentration values ​​on the numerical axis. Concentration score corresponding to samples that are less than or equal to the statutory dust concentration limit The maximum value is set as the lower limit of the advanced threshold. ; when At that time, the dust concentration level was low; when At that time, the dust concentration level was high.

[0016] Furthermore, during system trial operation, the lower limit of the advanced threshold is determined based on the rate of decrease in dust concentration. Make dynamic adjustments.

[0017] Furthermore, in step three, the personnel detection algorithm is trained on an edge optimization model using transfer learning technology. The training dataset of the edge optimization model includes real work images, personnel images in different poses, and images that have undergone enhancement processing.

[0018] A smart spray dust suppression system for implementing the method includes: The perception layer includes a vision module as the main sensor and a dust concentration sensor, which are used to collect image information and dust concentration data. The decision-making layer includes a microprocessor, which embeds image processing algorithms, data fusion algorithms, and hierarchical judgment logic to achieve intelligent decision-making at the edge. The execution layer includes a main control unit and three spray execution units corresponding to the monitoring areas of the left, middle and right sections, respectively. The main control unit controls the actions of the spray execution units of each section according to the instructions of the decision layer.

[0019] Furthermore, the vision module is an OpenMV vision module, the main control unit is an STM32 series microcontroller; the spray execution unit includes an electric ball valve and a spray frame that are connected to each other.

[0020] The beneficial effects of this invention are as follows: (1) By segmenting the images of the monitored area, the system achieves refined management by spraying only where there is dust. This method reduces ineffective or inefficient spraying behavior and can significantly shorten the time required for dust suppression compared to the traditional single-point spraying mode, while effectively saving water resources.

[0021] (2) The system innovatively introduces a real-time personnel sensing function, which can intelligently avoid situations where the spray will be turned off immediately when someone is present and turned on immediately when someone leaves. This not only greatly improves the work comfort and acceptance of on-site staff, but also fundamentally solves the problem of system failure caused by human intervention, ensuring the long-term stable operation of the dust suppression facilities.

[0022] (3) This invention uses a low-cost vision module as its core, replacing multiple expensive traditional dust sensors, making large-area, grid-based monitoring economically feasible. The edge computing architecture eliminates the need for complex data transmission links and central processing units, resulting in a more compact system structure, fewer points of failure, and better adaptability to harsh environments such as underground mines and construction tunnels.

[0023] (4) By fusing the spatial distribution information of images with the precise point data from sensors, this invention overcomes the limitations of single monitoring technologies. This weighted fusion of multi-source information makes the assessment of dust concentration more comprehensive and reliable, significantly improving the accuracy of system control decisions.

[0024] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a block diagram of an intelligent spray dust suppression system based on segmented image recognition and sensor fusion. Figure 2 This is a flowchart of an intelligent spray dust suppression method based on segmented image recognition and sensor fusion. Detailed Implementation

[0026] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0027] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0028] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0029] 1. System Composition and Hardware Connection like Figure 1 As shown, the present invention provides an intelligent spray dust suppression system based on segmented image recognition and sensor fusion, whose hardware architecture is divided into a perception layer, a decision layer and an execution layer.

[0030] Perception Layer: The vision module uses OpenMV N6, which serves as the core of edge computing and is responsible for real-time image acquisition and feature analysis. Dust concentration sensors, model GCD1000 / GCG1000, are deployed at the left, center, and right positions of the monitoring area, respectively. They connect to the main control unit via a Universal Asynchronous Receiver / Transmitter (UART) interface, providing precise point-based concentration data.

[0031] Decision layer: Composed of the main control unit (STM32 F-series microcontroller) and the embedded algorithms built into the vision module. The vision module is responsible for running the lightweight object detection (Faster Objects, More Objects, FOMO) person recognition model and image feature extraction algorithms.

[0032] Execution layer: The main control unit connects to the relay module via general purpose input / output (GPIO) pins to drive the electric ball valves of the three zones. Each zone's spray frame is independently controlled, achieving segmented spraying in physical space.

[0033] 2. Example 1: Segmented Image Feature Extraction and Dust Classification In this embodiment, a detailed description is provided of how to quantify dust concentration using image information.

[0034] After capturing an image, the vision module first horizontally divides it into three regions of interest (ROIs): left, middle, and right. For each ROI, the system executes the following algorithm in parallel: 1. Sharpness Extraction: Calculate the variance of the Laplacian operator in the image. As dust concentration increases, image details are obscured, edge information is reduced, leading to a decrease in sharpness value. Decrease, its reciprocal It increases accordingly.

[0035] 2. Average Gray Scale Calculation: Calculate the average brightness value of all pixels within the ROI. Under auxiliary lighting conditions in mines, dust particles scatter light, causing the background to appear grayish-white. The denser the dust, the higher the average gray value. The larger.

[0036] 3. Fusion computing: The system acquires dust sensor readings corresponding to each zone. Substitute into the formula:

[0037] Calculate the comprehensive concentration score .

[0038] 4. Level Determination: [The following is a list of steps / mechanisms] With preset advanced threshold (e.g., 4.0) for comparison. If The dust level of the area is marked as "high" otherwise as "low".

[0039] 3. Example 2: Person perception and avoidance based on the FOMO model This embodiment describes how the system ensures that workers are not sprayed while dust is being reduced.

[0040] like Figure 2 As shown, while calculating dust concentration, the system runs a FOMO (Fear of Discharge) personnel detection model specifically optimized for the mining environment in each ROI area.

[0041] 1. Model characteristics: This model uses MobileNet V2 as the pre-training base for transfer learning and can recognize various postures such as walking, standing, and bending over.

[0042] 2. Control Logic: The system makes logical decisions for each partition.

[0043] Activation condition: The GPIO output is high to activate the corresponding solenoid valve only when the dust level of a certain ROI is "high" and no personnel are detected in the FOMO model within that ROI.

[0044] Instantaneous stop spraying: If personnel enter the ROI area during spraying, the vision module will detect the "person" status and respond within milliseconds to shut down the spraying in that zone.

[0045] Automatic reset: Once personnel leave the area and dust levels remain above the limit, the system immediately resumes spraying without manual intervention.

[0046] 4. Example 3: On-site calibration method for weighting coefficients This embodiment relates to the parameter calibration process during system deployment.

[0047] To ensure the coefficients in the scoring formula , , To accurately reflect the on-site working conditions, the following steps are adopted: 1. Baseline Acquisition: A calibrated dust sampler is deployed on-site to obtain standard concentration values. .

[0048] 2. Sample Construction: Collect no fewer than 50 sets of samples containing the reciprocal of sharpness. Average gray level Sensor readings and standard values Synchronized data.

[0049] 3. Solve using the least squares method: Substitute the data into the linear equation system for regression analysis.

[0050] For example, the coefficients obtained after calibration at a coal mine transfer point are: .

[0051] 4. Threshold binding: Based on the total dust limit of 4 mg / m³ in the "Coal Mine Safety Regulations". 3 The standard will correspond to this concentration. The value is set as the trigger threshold.

[0052] 5. Example 4: Dynamic Optimization Strategy for Thresholds To adapt to changes in different seasons or operating conditions, the system has statistical optimization functions.

[0053] 1. Response Lag Adjustment: If the sensor reading spikes rapidly after the system records multiple dust level determinations as "low," it indicates that the intervention was too late, and the system will automatically adjust the level appropriately. .

[0054] 2. Over-response adjustment: If the spray starts frequently but the concentration remains extremely low, the threshold should be appropriately increased to conserve resources.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A smart spray dust suppression method based on segmented image recognition and sensor fusion, characterized in that: Includes the following steps: Step 1, System Initialization: Perform hardware initialization on the vision module and dust concentration sensor, and load the preset personnel detection model; Step 2, Image Acquisition and Region Division: Real-time image information is captured by the vision module, and the monitoring field of view of the vision module is divided into three independent monitoring regions in the horizontal direction: left, middle and right. Each monitoring region corresponds to an independent spray execution unit. Step 3, Personnel Detection by Zone: Run a personnel detection algorithm based on a lightweight target detection FOMO model in each monitoring zone to determine in real time whether there are personnel in each monitoring zone and obtain the personnel presence status; Step 4, Dust Concentration Classification: Extract image features from each monitoring area and obtain sensor readings from the dust concentration sensor. Calculate the comprehensive concentration score for each monitoring area using a fusion decision algorithm. According to the comprehensive concentration score The threshold range in which the dust concentration is located divides the dust concentration into two levels: high and low. Step 5, Intelligent Linkage Control: Based on the presence status of personnel and the dust concentration level of each monitoring area, a logical judgment is made. If and only if the dust concentration level in a certain monitoring area is high and there are no personnel in that monitoring area, the spray execution unit corresponding to that monitoring area is triggered to start spraying through the general purpose input / output (GPIO) pin.

2. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 1, characterized in that: In step four, the extracted image features include the reciprocal of image sharpness. and average gray value The image sharpness is achieved by calculating the variance of the Laplacian operator.

3. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 2, characterized in that: In step four, the comprehensive concentration score The calculation formula is: In the formula, , , These are the weighting coefficients for the reciprocal of sharpness, the average gray value, and the sensor reading, respectively. The reading is the sensor reading of the dust concentration sensor.

4. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 3, characterized in that: The weighting coefficient , , Determined through the following steps: Acquire multiple sets of sample data under different dust concentration conditions. Each set of sample data includes the reciprocal of the sharpness acquired simultaneously. Average gray value Auxiliary dust monitor readings and the calibrated standard concentration values ; A system of linear equations is constructed from the multiple sets of sample data, and the weighting coefficients are obtained by solving the least squares method. , , .

5. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 3, characterized in that: The comprehensive concentration score The threshold range is determined through the following steps: Based on the legal dust concentration limit, the concentration score corresponding to each sample data is calculated using the aforementioned weighting coefficients. ; exist Find all standard concentration values ​​on the numerical axis. Concentration score corresponding to samples that are less than or equal to the statutory dust concentration limit The maximum value is set as the lower limit of the advanced threshold. ; when At that time, the dust concentration level was low; when At that time, the dust concentration level was high.

6. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 5, characterized in that: During system trial operation, the lower limit of the advanced threshold is determined based on the rate of decrease in dust concentration. Make dynamic adjustments.

7. The intelligent spray dust suppression method based on segmented image recognition and sensor fusion according to claim 1, characterized in that: In step three, the personnel detection algorithm is trained on an edge optimization model using transfer learning techniques. The training dataset of the edge optimization model includes real work images, personnel images in different poses, and images that have undergone enhancement processing.

8. An intelligent spray dust suppression system implementing the method of any one of claims 1 to 7, characterized in that: include: The perception layer includes a vision module as the main sensor and a dust concentration sensor, which are used to collect image information and dust concentration data. The decision-making layer includes a microprocessor, which embeds image processing algorithms, data fusion algorithms, and hierarchical judgment logic to achieve intelligent decision-making at the edge. The execution layer includes a main control unit and three spray execution units corresponding to the monitoring areas of the left, middle and right sections, respectively. The main control unit controls the actions of the spray execution units of each section according to the instructions of the decision layer.

9. The intelligent spray dust suppression system according to claim 8, characterized in that: The vision module is an OpenMV vision module, and the main control unit is an STM32 series microcontroller; the spray execution unit includes an electric ball valve and a spray frame that are connected to each other.

Images