An unmanned aerial vehicle-based hydropower station operation ventilation equipment state monitoring method
By using UAV image processing technology, feature recognition, and Hough transform to determine the status of ventilation ducts, the problem of the inability to monitor the status of ventilation ducts in existing technologies has been solved, enabling safe and reliable monitoring of ventilation equipment and improving the safety of hydropower station operations.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DADU RIVER HYDROPOWER DEV
- Filing Date
- 2025-11-28
- Publication Date
- 2026-05-29
Smart Images
Figure CN122116191A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of operational ventilation equipment status in hydropower stations, and more specifically, to a method for monitoring the status of operational ventilation equipment in hydropower stations based on unmanned aerial vehicles (UAVs). Background Technology
[0002] Currently, the design and maintenance of ventilation systems are key factors in ensuring workplace safety. Ventilation ducts, as an important component of the ventilation system, directly affect the efficiency and effectiveness of airflow. However, in practical applications, identifying whether ventilation ducts are in a ventilated state often presents challenges. Ventilation duct identification can only identify the duct itself, but cannot determine whether it is in a ventilated state. When workers are actually operating the system, if the ventilation status cannot be guaranteed, it can lead to safety issues for the workers. Summary of the Invention
[0003] The main objective of this application is to provide a method for monitoring the status of ventilation equipment in hydropower stations based on unmanned aerial vehicles (UAVs), in order to solve the problem that it is not easy to monitor whether the ventilation ducts are in a ventilated state.
[0004] To achieve the above objectives, the first aspect of this application proposes a method for monitoring the condition of ventilation equipment in hydropower stations based on unmanned aerial vehicles (UAVs), comprising: Acquire images taken by drones while they are operating in the target space; The image is subjected to feature recognition to determine whether a ventilation duct exists in the image; If the ventilation duct exists in the image, then based on the minimum circumscribed rectangle area of the ventilation duct, the image is divided into multiple regions along the spiral direction of the outer spiral of the ventilation duct, and the crease edge points in each region are obtained. Map the crease edge points in each region to Hough space to obtain the effective set of lines for each region; If the set of valid straight lines in each region is less than a preset number threshold, then determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to a preset angle threshold, and obtain the determination result. If the judgment result is greater than or equal to the preset angle threshold, the ventilation duct is determined to be in a ventilation state.
[0005] In some feasible methods, the step of acquiring images taken by the UAV while operating in the target space includes: Construct a flight path for the unmanned aerial vehicle (UAV), wherein the flight path includes at least one target space operation point; The images were obtained by taking pictures using a drone while operating in the target space at the target space work site.
[0006] In some feasible methods, the step of performing feature recognition on the image to determine whether a ventilation duct exists in the image includes: The image is subjected to grayscale processing to obtain the grayscale processed image; The image after grayscale processing is subjected to edge information feature recognition to determine whether there is an object with a spiral pattern. If an object with a spiral pattern exists, determine whether the overall shape of the object conforms to a cylindrical shape, whether the length of the cylindrical shape is greater than or equal to a preset length threshold, whether one end is connected to a cylindrical object, and whether the other end extends to a circular well opening, to obtain the judgment result of the object. If the judgment result of the object is yes, then it is determined that the ventilation duct exists in the image, wherein the ventilation duct includes a spiral-patterned ventilation duct and a cylindrical fan.
[0007] In some feasible methods, the step of dividing the image into multiple regions along the spiral direction of the outer spiral of the ventilation duct based on the minimum circumscribed rectangle area of the ventilation duct, and obtaining the crease edge points in each region, includes: If the ventilation duct exists in the image, calculate the area of the ventilation duct in the image to obtain the area of the smallest bounding rectangle of the ventilation duct in the image; If the area of the minimum bounding rectangle is greater than or equal to a preset area threshold, the area is divided into multiple regions based on the spiral direction of the spiral line on the outer surface of the ventilation duct to obtain the analysis region. Using an edge detection algorithm, the image edge points of each analysis region are calculated to form spiral crease features, thus obtaining the crease edge points in each region.
[0008] In some feasible methods, the step of mapping the crease edge points in each of the regions to Hough space to determine the valid straight lines includes: For the pixel coordinates of the crease edge points in each region, the corresponding extreme radius value is calculated using the Hough transform formula under a preset set of θ angles, and each crease edge point is mapped to the Hough parameter space composed of the set of θ angles and the extreme radius value to obtain the Hough parameter space point set of each region. The Hough parameter space point set of each region is statistically analyzed, and the number of mapping points in each parameter unit is counted to construct the accumulator matrix corresponding to each region. The parameter unit includes a set of θ angles and polar radius values. In the accumulator matrix, local extreme points with a number exceeding a preset threshold are identified to obtain a candidate peak set for each region; Based on the candidate peaks in the candidate peak set of each region, the corresponding straight line equation is calculated, and the straight line formed by the edge points that conform to the straight line equation is determined as the valid straight line, thus obtaining the valid straight line set of each region.
[0009] In some implementable methods, if the number of valid straight lines in each region is less than a preset threshold, then the step of determining whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to a preset angle threshold, and obtaining the determination result, includes: If the total number of valid lines in the set of valid lines in all the regions is less than the preset number threshold, then the tangent angle of the curved profile of the ventilation duct at the wellhead is calculated to obtain the folding angle of the ventilation duct. The folding angle is compared with a preset angle threshold to generate the judgment result.
[0010] In some feasible methods, the step of determining that the ventilation duct is in a ventilation state if the judgment result is greater than or equal to a preset angle threshold includes: If the folding angle is greater than or equal to the preset angle threshold in the judgment result, the state of the ventilation duct is marked as ventilation state, and a normal ventilation state signal is generated. If the folding angle is less than the preset angle threshold in the judgment result, the ventilation duct is marked as non-ventilated and an abnormal ventilation signal is generated.
[0011] Secondly, this application provides a drone-based hydropower station ventilation equipment status monitoring system, applied to the aforementioned drone-based hydropower station ventilation equipment status monitoring method, comprising: The acquisition unit is used to acquire images taken by the UAV while it is operating in the target space. A feature recognition unit is used to perform feature recognition on the image to determine whether a ventilation duct exists in the image; The region division unit is used to divide the ventilation duct into multiple regions along the spiral direction of the spiral line outside the ventilation duct based on the minimum bounding rectangle area of the ventilation duct if the ventilation duct exists in the image, and obtain the crease edge points in each region. A mapping unit is used to map the crease edge points in each region to Hough space to obtain the effective set of lines for each region; The judgment unit is used to determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to the preset angle threshold if the effective straight line set of each region is less than a preset number threshold, and to obtain the judgment result. The result unit is used to determine that the ventilation duct is in a ventilation state if the judgment result is greater than or equal to a preset angle threshold.
[0012] Thirdly, this application provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the aforementioned method.
[0013] Fourthly, this application provides a computer program that, when executed by a processor, implements the steps of the aforementioned method.
[0014] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This application presents a method for monitoring the status of ventilation equipment in hydropower stations based on unmanned aerial vehicles (UAVs). By capturing images with a UAV and employing an image processing chain, automated and non-contact intelligent monitoring of the ventilation equipment's status is achieved. Its advantages are as follows: First, it accurately distinguishes ventilation ducts from other objects through feature recognition. Second, it utilizes the significant difference in the straight-line characteristics of the fold edges of the ventilation duct's spiral pattern in its unfolded (ventilated) and folded (non-ventilated) states (quantified by the number of effective straight lines after Hough transform), combined with double verification using the folding angle at the duct inlet, thus reliably distinguishing the "ventilated" and "non-ventilated" states of the ventilation duct. This method effectively overcomes the limitation of existing technologies that can only identify the existence of ducts but cannot determine their functional status, providing real-time safety status warnings for confined space operations such as hydropower stations. It fundamentally avoids safety accidents caused by ventilation failure and significantly improves the level of operational safety. Attached Figure Description
[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 A flowchart of a method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) provided in this application; Figure 2 A logic diagram of a method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) provided in this application. Detailed Implementation
[0016] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0017] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0018] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.
[0019] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.
[0020] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.
[0021] The ventilation systems of hydroelectric power stations typically have their ducts installed in relatively spacious main powerhouse areas such as generator floors, turbine floors, and installation rooms, or in semi-open corridors with large cross-sectional dimensions, such as inlets / outlets and water diversion tunnels. These areas usually have sufficient space and height, providing basic conditions for the safe flight and filming of drones. Ventilation ducts are generally made of flexible materials such as reinforced PVC, with regular, continuous spiral ring patterns on their walls, giving them an overall regular cylindrical shape. One end of these ducts connects to cylindrical fan equipment via standard interfaces, while the other end extends and is fixed to a circular wellhead or opening leading to the work surface, forming a stable and predictable "fan-duct-wellhead" spatial layout.
[0022] When the ventilation system is working normally, i.e., in ventilation mode, the fan starts, causing airflow to fill the duct. The internal pressure causes the flexible duct to fully expand and tighten. At this time, the duct maintains an approximately straight cylindrical shape, and the spiral patterns on its surface become clearer and more regular. The creases appear as straight or uniformly curved continuous lines in the image. This shape is a direct external representation of unobstructed ductwork and smooth airflow.
[0023] Therefore, this application utilizes drones to capture images in the aforementioned operational areas with flyable conditions, effectively capturing these features. The significant visual identifiers and spatial relationships of the pipelines ensure the accuracy of target identification, while the morphological changes directly resulting from their ventilation status are transformed into differences in linear features in the image. This makes it possible to determine the status using algorithms, ultimately achieving automated, non-contact intelligent monitoring of the ventilation status.
[0024] like Figure 1 As shown, in a first aspect, this application provides a method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs), comprising: S100 acquires images taken by the drone while it is operating in the target space.
[0025] Specifically, acquiring images taken by a drone while operating in the target space may include the following steps: S101, constructing drone flight routes.
[0026] The flight path includes at least one target space operation point.
[0027] Specifically, based on a 3D map or prior knowledge of the target space of the hydropower station (such as the main powerhouse or large tunnels), the ventilation equipment that needs to be monitored is identified. Subsequently, based on the principle of covering all target equipment and avoiding obstacles, and in combination with the 3D map and the location of the ventilation equipment, a path is planned to form the UAV flight route. The flight path contains at least one hovering operation point corresponding to each target ventilation equipment.
[0028] S102, at the target space operation point, the image is obtained by taking pictures using a drone while operating in the target space.
[0029] Specifically, after flying along its flight path and reaching the target work site, the drone adjusts its yaw, pitch, and gimbal angles to align the camera's viewfinder with the target ventilation equipment. This ensures that it can capture images from a near-side view, including the overall outline of the ventilation duct, its surface texture, and its connection to the fan and wellhead. After capturing the images, the drone transmits them in real-time or near real-time to the ground station or backend data processing system via a wireless data link.
[0030] It should be noted that after the drone captures images, it first performs coarse alignment to adjust the camera angle. Specifically, after the drone reaches the target space, it captures a first image of the operation. The ventilation equipment in the first image is then identified. If the ventilation equipment is not in the first image, or is only partially in it, the camera angle is adjusted based on the content of the first image, and a second image is captured. This second image is then used to identify the ventilation equipment again, until the ventilation equipment is fully visible in the image. In other words, the images capture images to identify the edges of the ventilation equipment. The image where all the edges of the ventilation equipment are identified is used as the target image for subsequent steps.
[0031] In addition, the images captured by the drone can be manually adjusted by the personnel operating the drone, who can then manually adjust the images based on the location of the ventilation equipment.
[0032] It should also be noted that, based on the pose of the images captured by the UAV that meet the requirements and the location of the UAV, the pose data such as the UAV's GPS coordinates, flight altitude, yaw angle, pitch angle, roll angle, and gimbal pitch angle at the current moment are recorded to form the pose data of the current target space operation point. This data is then added to the flight path so that when the UAV inspects and monitors the current target space operation point again, it can directly use the pose data to take images, thereby improving work efficiency.
[0033] S200, perform feature recognition on the image to determine whether a ventilation duct exists in the image.
[0034] The purpose of step S200 is to distinguish the ventilation ducts of the hydropower station from other objects in the image by utilizing their visual characteristics (spiral pattern) and fixed spatial connections (connecting the fan and the wellhead).
[0035] Specifically, determining whether a ventilation duct exists in the image may include the following steps: S201, perform grayscale processing on the image to obtain the grayscale processed image.
[0036] Specifically, the color image is converted to a grayscale image to reduce interference from color information and enhance the brightness differences between different objects, facilitating subsequent edge feature extraction. For example, the red, green, and blue components of each pixel in the image are converted into a single grayscale value through a weighted average. Subsequently, methods such as contrast stretching are used to expand the dynamic range of the image's grayscale levels, making the differences between the outline and surface texture of the ventilation duct and the background more pronounced.
[0037] S202, perform edge information feature recognition on the grayscale processed image to determine whether there is an object with a spiral pattern.
[0038] Specifically, step S202 identifies the boundaries of all objects in the image and performs an initial screening based on texture features. For example, an edge detection algorithm (which identifies pixels with drastic changes in grayscale values in the image) is used to obtain the edge contours of all objects in the entire image. Among the extracted edge contours, combinations of approximately parallel curve pairs that appear periodically are identified. This edge combination pattern is a typical representation of spiral textures. Edge regions matching this pattern are preliminarily identified as objects that may contain spiral textures.
[0039] It should be noted that when identifying spiral patterns, a minimum number of consecutive cycles (e.g., at least three pairs of parallel curves) can be set as a valid criterion. For areas with discontinuous edges due to viewing angle, lighting, or dirt, morphological operations (such as closing, a morphological image processing operation that is essentially a combination of dilation and erosion) can be used to connect the edges to improve the robustness of pattern recognition.
[0040] S203, if there is an object with a spiral pattern, then determine whether the overall shape of the object conforms to a cylindrical shape, whether the length of the cylindrical shape is greater than or equal to a preset length threshold, whether one end is connected to a cylindrical object, and whether the other end extends to a circular well opening, and obtain the judgment result of the object.
[0041] Specifically, step S203 ensures that the object selected in the previous step is a genuine ventilation duct, and not other objects with similar textures (such as certain hoses, railings, etc.).
[0042] For example, for each object with a spiral pattern that has been initially selected, the following four verifications are performed in sequence, and all conditions must be met simultaneously: 1. Shape verification: Determine whether the overall external contour of an object can be fitted by a cylindrical geometric model.
[0043] 2. Size verification: Measure the projected length of the cylindrical object in the image and determine whether it is greater than or equal to a preset length threshold to exclude small tubular objects.
[0044] 3. Connection verification (one end): Determine whether one end of the object has a visual connection with a cylindrical object (i.e., the fan casing).
[0045] 4. Connection verification (other end): Determine whether the other end of the object extends into a circular or near-circular hole or structure (i.e., ventilation shaft).
[0046] It should be noted that the verification of connectivity can be achieved by determining whether the edge contours of two objects intersect or whether the minimum distance is less than a preset pixel threshold. For ventilation shafts, prior location information in the hydropower station environment (such as their typical location on a wall or ground) can be used for auxiliary judgment.
[0047] In addition, each verification step in S203 above can be assigned a weighted score. Finally, a total confidence score is calculated. If the total score exceeds the success threshold, the ventilation duct is determined to exist; if the score is below the failure threshold, it is determined to not exist; if the score is between the two, it can be marked as suspected, triggering the drone to retake the picture from a different angle or upload the image for manual review, in order to improve the reliability of the system.
[0048] S204, if the judgment result of the object is yes, then it is determined that the ventilation duct exists in the image.
[0049] The ventilation duct includes a spiral-patterned ventilation duct and a cylindrical fan.
[0050] Specifically, step S204 synthesizes the results of all verification conditions and gives the final conclusion as to whether a ventilation duct exists.
[0051] For example, if all four judgments for a certain object in step S203 are true, then it is determined that a target ventilation device exists in the image. The cylindrical object with spiral patterns is the ventilation duct, and the cylindrical object connected to it is the fan. If any condition is not met, it is determined that there is no suitable ventilation duct in the image. That is, if a non-suitable ventilation duct is determined, the image needs to be re-captured using a drone.
[0052] S300, if the ventilation duct exists in the image, then according to the minimum circumscribed rectangle area of the ventilation duct, the image is divided into multiple regions along the spiral direction of the spiral line outside the ventilation duct, and the crease edge points in each region are obtained.
[0053] In step S300, when the ventilation duct is large enough in the image, its surface is subjected to detailed segmentation analysis. The segmentation is based on the spiral structure of the ventilation duct itself, and the extracted features are the crease edge points that reflect its surface condition.
[0054] Specifically, obtaining the crease edge points in each region may include the following steps: S301, if the ventilation duct exists in the image, calculate the area of the ventilation duct in the image to obtain the area of the smallest bounding rectangle of the ventilation duct in the image.
[0055] Specifically, step S301 is used to quantify the size of the identified ventilation ducts in the image, providing a basis for determining whether to perform detailed analysis.
[0056] For example, the pixel region occupied by the identified ventilation duct in the image is determined. A rectangle with the smallest area that completely encloses the duct pixel region is found; this rectangle is called the minimum bounding rectangle. Its sides are parallel to the coordinate axes of the image. Next, the area of this minimum bounding rectangle is calculated, which is the product of its width (pixel value) and height (pixel value). This area value characterizes the visual size of the duct in the image.
[0057] S302, if the area of the minimum circumscribed rectangle is greater than or equal to a preset area threshold, then the area is divided into multiple regions based on the spiral direction of the spiral line on the outer surface of the ventilation duct to obtain the analysis region.
[0058] Specifically, in step S302, when the pipe size meets the conditions, its surface image is divided into multiple local regions for fine feature analysis.
[0059] For example, the area of the smallest bounding rectangle obtained in step S301 is compared with a preset area threshold. This preset area threshold is used to ensure that the ventilation duct image has sufficient resolution for meaningful subdivision. Subsequent partitioning operations are only performed when the area of the smallest bounding rectangle is greater than or equal to this preset area threshold. Next, the direction of the spiral extension is determined based on the direction of the spiral pattern on the surface of the ventilation duct.
[0060] Subsequently, along a defined spiral direction, the outer surface image of the ventilation duct (typically corresponding to the portion containing the duct body within the smallest bounding rectangle) is divided into several continuous, strip-shaped analysis regions. These regions cover the visible surface of the duct along the spiral path.
[0061] S303, using an edge detection algorithm, calculate the image edge points of each analysis region to form spiral crease features, and obtain the crease edge points in each region.
[0062] Specifically, in step S303, feature points that characterize the geometric shape of the spiral pattern are precisely located within each analysis area.
[0063] For example, an edge detection algorithm is used to detect edges in each analysis region. This algorithm calculates the intensity (gradient) of grayscale change in different directions for each pixel in each analysis region of the image, and identifies edge points at locations with drastic grayscale changes (i.e., the edges of objects). Subsequently, crease feature points are filtered out from all detected edge points, selecting points located on the ridges (tops of raised portions) or valleys (bottoms of recessed portions) of the spiral pattern. These points constitute the crease features of the spiral pattern and are called crease edge points. The filtering can be based on the geometric relationship between the gradient direction of the edge point and the direction of the spiral pattern. That is, the edge characteristics of a spiral crease: the edge line of the spiral pattern (whether a raised peak or a recessed valley) extends primarily along the spiral direction. Thus, the angle between the gradient direction of each edge point and the direction of the spiral pattern in its region is calculated. If this angle is close to 90 degrees (allowing for a certain error range, such as 85-95 degrees), then this edge point is very likely the feature point of the spiral crease we are looking for. Conversely, if the angle deviates significantly from 90 degrees, then this point may belong to other irrelevant edges in the image (such as scratches, shadow boundaries, or the edges of other objects).
[0064] S400, map the crease edge points in each region to Hough space to obtain the effective set of lines for each region.
[0065] In step S400, the Hough transform is used to map points belonging to the same straight line in the image space to the same peak point in the parameter space, thereby accurately identifying meaningful straight line features from messy crease edge points.
[0066] Specifically, obtaining the effective set of lines for each region may include the following steps: S401, for the pixel coordinates of the crease edge points in each region, the corresponding extreme radius value is calculated using the Hough transform formula under a preset set of θ angles, and each crease edge point is mapped to the Hough parameter space composed of the set of θ angles and the extreme radius value, to obtain the Hough parameter space point set of each region.
[0067] Specifically, in step S401, each crease edge point in the image space is transformed into the Hough parameter space for representation, in preparation for statistical analysis of collinear crease edge points.
[0068] For example, a set of angles θ is preset, such as from 0 degrees to 180 degrees, taking values in steps of 45 degrees. This covers all possible straight line directions. Next, a coordinate transformation is performed, where for each crease edge point in the current analysis region, its pixel coordinates are (x, y). For each angle θ in the preset set, it is substituted into the Hough transform formula: ; Where θ represents the angle between the normal to the line and the x-axis, typically ranging from 0 to 180 degrees, for example, with a step size of 45 degrees, calculating the perpendicular distances corresponding to 0, 45, 90, 135, and 180 degrees; (x, y) represents the pixel coordinates of a crease edge point in image space; ρ represents the perpendicular distance from the origin (0, 0) of the image coordinate system to this line. A line is defined by this formula, and in image space, any line can be uniquely determined by two parameters θ and ρ. θ determines the direction of the line (i.e., the direction of its normal), and ρ determines the distance of the line from the origin.
[0069] The formula serves to bridge the gap between image space and Hough parameter space (θ-ρ space). In image space, a crease edge point (x, y) can exist on countless straight lines passing through it. In Hough parameter space, this crease edge point (x, y) corresponds to a sine curve. This curve is formed by all points that satisfy the formula. This means that every possible straight line passing through point (x, y) in image space corresponds to a point on the curve in parameter space. After traversing all crease edge points and all preset θ angles within the region, a set of points in Hough parameter space corresponding to that region is obtained, i.e., the Hough parameter space point set.
[0070] S402, perform statistics on the Hough parameter space point set of each region, count the number of mapping points in each parameter unit, and construct the accumulator matrix corresponding to each region.
[0071] The parameter unit includes a set of θ angles and polar radius values.
[0072] Specifically, in step S402, the number of crease edge points in the image space that support each point (θ, ρ) in the quantization parameter space is determined to find the strongest straight line feature.
[0073] For example, the continuous Hough parameter space (θ, ρ) is discretized by dividing it into small grids based on the step size of θ and the accuracy requirements of ρ. Each grid is called a parameter cell. Next, the point set of the Hough parameter space obtained in step S401 is traversed. For each point, its corresponding parameter cell is found, and the count value of that cell is incremented by 1. Subsequently, a matrix is constructed, organizing the count values of all parameter cells into a two-dimensional table (matrix), which is the accumulator matrix. The rows and columns of the matrix typically correspond to the discrete values of ρ and θ, respectively. The value of each cell in the accumulator matrix intuitively reflects how many crease edge points in the image space lie on the line defined by the parameters of that cell (θ, ρ).
[0074] S403, in the accumulator matrix, identify local extreme points whose number exceeds a preset threshold, and obtain a candidate peak set for each region.
[0075] Specifically, in step S403, significant straight-line features supported by a sufficient number of image crease edge points are identified from the accumulator matrix.
[0076] For example, a minimum count value is set as a preset threshold to filter out weak responses caused by noise or sporadic crease edge points. In the accumulator matrix, points whose count values are not only greater than the preset threshold but also significantly higher than the count values of their surrounding neighboring cells are identified. These points are called local extrema. All these qualified local extrema (θ, ρ) constitute a candidate peak set. Each peak in the set corresponds to a potential straight line in the image space. These straight lines are candidates for valid straight lines identified from the spiral creases in that region.
[0077] S404, calculate the corresponding straight line equation based on the candidate peaks in the candidate peak set of each region, and determine the straight line formed by the edge points that conform to the straight line equation as the valid straight line, thereby obtaining the valid straight line set of each region.
[0078] Specifically, in step S404, the candidate peaks (i.e. potential straight line features) in the Hough parameter space are transformed back into specific straight lines in the image space, and the pixels that constitute these straight lines are identified, thereby obtaining the final set of valid straight lines identified in the region.
[0079] For example, each candidate peak in the candidate peak set is itself a set of parameters. This parameter pair directly defines a straight line in the image space. The equation of this straight line is determined by the parameters θ and ρ. Next, the edge points constituting the straight line are filtered. In the image space, all crease edge points in the current region are traversed. For each crease edge point (x, y), its distance to the straight line defined by (θ, ρ) is calculated. If this distance is less than a preset tolerance threshold, the crease edge point is considered to be a point that conforms to the equation of this straight line, that is, it lies on this straight line. Finally, the valid straight lines are determined and a set is generated. The straight line formed by all crease edge points that conform to the equation of this straight line is confirmed as a valid straight line. By traversing all candidate peaks in the candidate peak set and repeating the calculation of the straight line equation and the filtering of the edge points constituting the straight line, the set of all valid straight lines in the region can be obtained, that is, the set of valid straight lines.
[0080] S500, if the set of valid straight lines in each region is less than a preset number threshold, then determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to a preset angle threshold, and obtain the determination result.
[0081] In step S500, the ventilation duct will fold when not ventilated, resulting in a reduction of the regular straight lines of the spiral pattern on its surface, and a noticeable bend at the wellhead inlet. By comprehensively evaluating these two characteristics, its working status can be reliably determined.
[0082] Specifically, obtaining the judgment result may include the following steps: S501, if the total number of valid straight lines in the set of valid straight lines in all the regions is less than the preset number threshold, then calculate the tangent angle of the curved profile of the ventilation duct at the wellhead to obtain the folding angle of the ventilation duct.
[0083] Specifically, in step S501, when the number of straight features indicates an anomaly (possible folding), the degree of bending of the pipeline at the critical part (wellhead) is further quantified to provide a basis for condition judgment.
[0084] It should be noted that at the bend where the ventilation duct enters the manhole, tangent lines are drawn to the contours of the upstream section (pointing towards the main body of the duct) and the downstream section (pointing towards the inside of the manhole). These two tangent lines will form an included angle. This application defines the supplementary angle of the included angle (i.e., 180 degrees minus the included angle value) as the fold angle. Therefore, the gentler the bend of the duct (in ventilation mode), the larger this fold angle; the sharper the bend of the duct (in fold mode), the smaller this fold angle.
[0085] For example, the total count and judgment are as follows: First, all valid straight lines obtained from all analysis areas are summarized, and their total number is calculated. This total number is compared with a preset quantity threshold. Only when the total number is less than the preset quantity threshold (indicating that the pipe surface lacks regular long straight line characteristics due to folds), the calculation of the fold angle is triggered. Otherwise, the process may directly lead to the judgment of "ventilation status".
[0086] Locating the curved section of the manhole: In the image, locate the area where the end of the ventilation duct enters the circular manhole. This area will exhibit a distinct curved profile due to duct collapse when not ventilated.
[0087] Calculating the folding angle: Using image analysis techniques, such as edge detection algorithms (like the Canny operator), the contour line of the curved section is extracted. Then, near the inflection point (the point of maximum curvature) of the curved contour, tangents are drawn to the upstream and downstream contours of the curved section, respectively. These two tangents naturally form an angle, which is the folding angle of the ventilation duct. This folding angle visually reflects the severity of the duct's curvature.
[0088] Further, calculate the folding angle: In the area where the end of the ventilation duct connects to the wellhead, as identified in the image, perform the following operations: Contour extraction utilizes edge detection algorithms (such as the Canny operator) to specifically extract the contour lines of the curved region, resulting in a set of ordered contour pixels.
[0089] Inflection point localization involves calculating the curvature of each contour pixel on the contour line and determining the point with the maximum curvature as the inflection point of the curve.
[0090] Tangent fitting involves selecting a segment on the contour line upstream of the inflection point (pointing towards the main body of the pipeline) and fitting a straight line using the least squares method as the upstream tangent line; similarly, a segment on the contour line downstream of the inflection point (pointing towards the inside of the wellhead) is fitted with a straight line as the downstream tangent line.
[0091] Angle calculation: Calculate the included angle between the two tangents mentioned above. This included angle is the folding angle of the ventilation duct.
[0092] S502, compare the folding angle with a preset angle threshold to generate the judgment result.
[0093] Specifically, the folding angle calculated in step S501 is compared with a preset angle threshold set in advance based on safety standards and a large amount of experimental data. After the comparison, the generated judgment result has three possibilities: greater than, less than, and equal to.
[0094] For example, if the folding angle is greater than or equal to the preset angle threshold, the result is that the ventilation device is in a ventilation state (i.e., an extended state).
[0095] If the folding angle is less than the preset angle threshold, the result is that the ventilation equipment is in a non-ventilated state (i.e., folded state).
[0096] The preset quantity threshold and preset angle threshold can be set based on the statistical analysis results of historical normal (ventilated) and abnormal (non-ventilated) image samples.
[0097] S600, if the judgment result is greater than or equal to the preset angle threshold, it is determined that the ventilation duct is in a ventilation state.
[0098] The S600 step is the logical endpoint of the ventilation equipment status monitoring. It outputs the final signal based on the judgment results generated by the aforementioned steps.
[0099] Specifically, determining that the ventilation duct is in a ventilation state may include the following steps: S601, if the folding angle in the judgment result is greater than or equal to the preset angle threshold, then the state of the ventilation duct is marked as ventilation state, and a normal ventilation state signal is generated.
[0100] S602, if the folding angle in the judgment result is less than the preset angle threshold, then the state of the ventilation duct is marked as non-ventilation state, and a ventilation state abnormality signal is generated.
[0101] Specifically, in steps S601 to S602, the status of ventilation ducts determined to be normal or abnormal is confirmed, and corresponding indication signals are generated.
[0102] For example, based on the judgment result generated in S502, the comparison conclusion between the folding angle and the preset angle threshold is explicitly indicated. When the conclusion is "the folding angle is greater than or equal to the preset angle threshold," the system marks the current operating status of the monitored ventilation duct as "ventilation status" on the data layer or user interface. At the same time, the system generates a normal ventilation status signal. Conversely, when the conclusion is "the folding angle is less than the preset angle threshold," it is marked as non-ventilation status, and an abnormal ventilation status signal is generated.
[0103] Example like Figure 2 As shown, this application provides a method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs), including: 1. Acquire images taken by the drone while it is operating in the target space.
[0104] 2. Convert the captured image to grayscale. Enhance the contrast of the grayscale image. Extract the edges of all objects in the enhanced image.
[0105] 3. Determine whether there is a spiral-patterned object in the image region at the edge. If there is a spiral-patterned object in the image region at the edge, further determine whether the overall shape of the spiral-patterned object is cylindrical. If the overall shape is cylindrical and the length is greater than a preset length threshold, further determine whether one end of the object is connected to a cylindrical object, and at the same time determine whether the other end of the object enters a circular well opening.
[0106] If the other end of the object enters the circular well opening, then ventilation equipment is confirmed to exist in the image. The cylindrical object is the ventilation duct, and the cylindrical object is the fan.
[0107] 4. Determine the minimum bounding rectangle area corresponding to the ventilation equipment. The image taken is the image of the ventilation equipment taken by the drone from the side. If the minimum bounding rectangle area is greater than the preset area threshold, the ventilation duct is divided into multiple areas along the spiral direction.
[0108] 5. Extract the crease edge points in each region. Map each crease edge point to Hough space and determine the valid lines. Specifically, based on the coordinates of each crease edge point and parameter A, calculate parameter B using the following formula: A; Where B represents the perpendicular distance from the origin to the line in the Hough coordinate system, and A represents the angle between the perpendicular line and the x-axis. The range of A is set to 0-180 degrees, with a step size of 45 degrees, i.e., calculating the perpendicular distances corresponding to 0 degrees, 45 degrees, 90 degrees, 135 degrees, and 180 degrees. An accumulator matrix is constructed, as shown in Table 1. Obtain significant peak values from the accumulator matrix (e.g., 5 in Table 1). Based on these peak values, calculate the equation of the effective line. Specifically, select a target peak value that is greater than a preset peak value from among multiple peak values, and substitute the target peak value into the equation. A obtains the equation of the valid straight line and filters out the edge points that conform to this equation. The Hough transform can then accurately identify the true straight line features from the chaotic edge points.
[0109]
[0110] Table 1 6. If the number of effective straight lines is less than the preset threshold, the folding angle of the ventilation duct when it enters the wellhead is determined. If the folding angle is greater than the preset angle threshold, the ventilation equipment is determined to be in ventilation mode.
[0111] Secondly, this application provides a drone-based hydropower station ventilation equipment status monitoring system, applied to the aforementioned drone-based hydropower station ventilation equipment status monitoring method, comprising: The acquisition unit is used to acquire images taken by the UAV while it is operating in the target space. A feature recognition unit is used to perform feature recognition on the image to determine whether a ventilation duct exists in the image; The region division unit is used to divide the ventilation duct into multiple regions along the spiral direction of the spiral line outside the ventilation duct based on the minimum bounding rectangle area of the ventilation duct if the ventilation duct exists in the image, and obtain the crease edge points in each region. A mapping unit is used to map the crease edge points in each region to Hough space to obtain the effective set of lines for each region; The judgment unit is used to determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to the preset angle threshold if the effective straight line set of each region is less than a preset number threshold, and to obtain the judgment result. The result unit is used to determine that the ventilation duct is in a ventilation state if the judgment result is greater than or equal to a preset angle threshold.
[0112] Thirdly, this application provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the aforementioned method.
[0113] Fourthly, this application provides a computer program that, when executed by a processor, implements the steps of the aforementioned method.
[0114] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0115] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.
[0116] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs), characterized in that... Acquire images taken by drones while they are operating in the target space; The image is subjected to feature recognition to determine whether a ventilation duct exists in the image; If the ventilation duct exists in the image, then based on the minimum circumscribed rectangle area of the ventilation duct, the image is divided into multiple regions along the spiral direction of the outer spiral of the ventilation duct, and the crease edge points in each region are obtained. Map the crease edge points in each region to Hough space to obtain the effective set of lines for each region; If the set of valid straight lines in each region is less than a preset number threshold, then determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to a preset angle threshold, and obtain the determination result. If the judgment result is greater than or equal to the preset angle threshold, the ventilation duct is determined to be in a ventilation state.
2. The method for monitoring the status of ventilation equipment in hydropower stations based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, The step of acquiring images taken by the UAV during its operation in the target space includes: Construct a flight path for the unmanned aerial vehicle (UAV), wherein the flight path includes at least one target space operation point; The images were obtained by taking pictures using a drone while operating in the target space at the target space work site.
3. The method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, The step of performing feature recognition on the image to determine whether a ventilation duct exists in the image includes: The image is subjected to grayscale processing to obtain the grayscale processed image; The image after grayscale processing is subjected to edge information feature recognition to determine whether there is an object with a spiral pattern. If an object with a spiral pattern exists, determine whether the overall shape of the object conforms to a cylindrical shape, whether the length of the cylindrical shape is greater than or equal to a preset length threshold, whether one end is connected to a cylindrical object, and whether the other end extends to a circular well opening, to obtain the judgment result of the object. If the judgment result of the object is yes, then it is determined that the ventilation duct exists in the image, wherein the ventilation duct includes a spiral-patterned ventilation duct and a cylindrical fan.
4. The method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, If the ventilation duct exists in the image, the step of dividing the image into multiple regions along the spiral direction of the outer spiral of the ventilation duct based on the minimum circumscribed rectangle area of the ventilation duct, and obtaining the crease edge points in each region, includes: If the ventilation duct exists in the image, calculate the area of the ventilation duct in the image to obtain the area of the smallest bounding rectangle of the ventilation duct in the image; If the area of the minimum bounding rectangle is greater than or equal to a preset area threshold, the area is divided into multiple regions based on the spiral direction of the spiral line on the outer surface of the ventilation duct to obtain the analysis region. Using an edge detection algorithm, the image edge points of each analysis region are calculated to form spiral crease features, thus obtaining the crease edge points in each region.
5. The method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, The step of mapping the crease edge points in each region to Hough space to determine the valid straight lines includes: For the pixel coordinates of the crease edge points in each region, the corresponding extreme radius value is calculated using the Hough transform formula under a preset set of θ angles, and each crease edge point is mapped to the Hough parameter space composed of the set of θ angles and the extreme radius value, thus obtaining the Hough parameter space point set for each region. The Hough parameter space point set of each region is statistically analyzed, and the number of mapping points in each parameter unit is counted to construct the accumulator matrix corresponding to each region. The parameter unit includes a set of θ angles and polar radius values. In the accumulator matrix, local extreme points with a number exceeding a preset threshold are identified to obtain a candidate peak set for each region; Based on the candidate peaks in the candidate peak set of each region, the corresponding straight line equation is calculated, and the straight line formed by the edge points that conform to the straight line equation is determined as the valid straight line, thus obtaining the valid straight line set of each region.
6. The method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, If the number of valid straight lines in each region is less than a preset threshold, then the step of determining whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to a preset angle threshold, and obtaining the determination result, includes: If the total number of valid lines in the set of valid lines in all the regions is less than the preset number threshold, then the tangent angle of the curved profile of the ventilation duct at the wellhead is calculated to obtain the folding angle of the ventilation duct. The folding angle is compared with a preset angle threshold to generate the judgment result.
7. The method for monitoring the status of ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs) as described in claim 1, characterized in that, The step of determining that the ventilation duct is in a ventilation state if the judgment result is greater than or equal to a preset angle threshold includes: If the folding angle is greater than or equal to the preset angle threshold in the judgment result, the state of the ventilation duct is marked as ventilation state, and a normal ventilation state signal is generated. If the folding angle is less than the preset angle threshold in the judgment result, the ventilation duct is marked as non-ventilated and an abnormal ventilation signal is generated.
8. A condition monitoring system for ventilation equipment in a hydropower station based on unmanned aerial vehicles (UAVs), characterized in that, The method for monitoring the status of unmanned aerial vehicle (UAV)-based ventilation equipment in hydropower stations, as described in any one of claims 1-7, comprises: The acquisition unit is used to acquire images taken by the UAV while it is operating in the target space. A feature recognition unit is used to perform feature recognition on the image to determine whether a ventilation duct exists in the image; The region division unit is used to divide the ventilation duct into multiple regions along the spiral direction of the spiral line outside the ventilation duct based on the minimum bounding rectangle area of the ventilation duct if the ventilation duct exists in the image, and obtain the crease edge points in each region. A mapping unit is used to map the crease edge points in each region to Hough space to obtain the effective set of lines for each region; The judgment unit is used to determine whether the folding angle of the ventilation duct when it enters the wellhead is greater than or equal to the preset angle threshold if the effective straight line set of each region is less than a preset number threshold, and to obtain the judgment result. The result unit is used to determine that the ventilation duct is in a ventilation state if the judgment result is greater than or equal to a preset angle threshold.
9. A computer storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.
10. A computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.