Intelligent nozzle control method, system and device based on three-dimensional imaging

Abstract

The invention relates to an intelligent nozzle control method, system and device based on three-dimensional imaging, and the method comprises the steps: obtaining the three-dimensional point cloud data of the surface of an insulator, constructing a three-dimensional curved surface model of the surface of the insulator according to the three-dimensional point cloud data, and obtaining the curved surface shape features of the surface of the insulator; obtaining ice layer thickness data of the surface of the insulator according to the three-dimensional point cloud data, classifying the ice layer thickness data, and obtaining ice layer distribution areas of different thickness grades; the spraying direction and angle of the nozzle are controlled based on the curved surface shape characteristics of the insulator surface and the ice layer distribution area; and inputting the water outlet or air outlet parameter data at the current moment into the trained LSTM prediction model, outputting the predicted ice layer thickness and the ice accretion area, and controlling the water outlet or air outlet temperature, the flow velocity and the flow direction of the nozzle according to the predicted ice layer thickness and the ice accretion area. Accurate, efficient and safe deicing on the surface of the insulator is achieved, and stable operation of a power transmission line is effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent nozzle control technology, and in particular to an intelligent nozzle control method, system and device based on three-dimensional imaging. Background Art

[0002] In the case of ice disasters on transmission lines, the thickness of the ice layer on the surface of the insulator and the area of ​​ice accumulation are the key factors affecting the insulation performance of the insulator. At present, when cleaning and deicing robots remove the ice layer on the surface of the insulator, they mainly use high-pressure water guns or hot air guns, but there are the following technical problems: first, the spraying direction and angle of the water gun or hot air gun are difficult to accurately control, which easily causes spraying dead angles, resulting in incomplete removal of ice in local areas; second, during spraying, the temperature, flow rate, flow direction and other parameters of the water or air flow cannot be adjusted in real time according to the changes in the thickness of the ice layer and the area of ​​ice accumulation, making it difficult to achieve the best ice removal effect; at the same time, during the deicing process, the surface temperature of the insulator changes dramatically, which is easy to produce thermal stress, causing the insulator to crack or be damaged, affecting the safe and stable operation of the transmission line. Therefore, it is urgent to study an intelligent deicing method and system that can monitor the surface temperature of the insulator in real time and can intelligently adjust the nozzle temperature, flow rate, flow direction and spray shape according to the thickness of the ice layer and the area of ​​ice accumulation, so as to improve the efficiency of emergency disposal of ice disasters on transmission lines. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent nozzle control method, system and device based on three-dimensional imaging, so as to achieve accurate, efficient and safe deicing of the insulator surface and effectively ensure the stable operation of the transmission line.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] On the one hand, a smart nozzle control method based on three-dimensional imaging is provided, comprising:

[0006] Acquire three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and acquire the surface shape features of the insulator surface;

[0007] The ice thickness data on the insulator surface is obtained based on the 3D point cloud data, and the ice thickness data is classified to obtain the distribution areas of ice layers with different thickness levels;

[0008] Based on the curved shape characteristics of the insulator surface and the distribution area of ​​the ice layer, the spray direction and angle of the nozzle are controlled;

[0009] The water or wind outlet parameter data at the current moment is input into the trained LSTM prediction model, and the predicted ice thickness and ice accumulation area are output. The water or wind outlet parameters of the nozzle are controlled according to the predicted ice thickness and ice accumulation area, wherein the water or wind outlet parameters are temperature, flow rate and flow direction.

[0010] Optionally, constructing a three-dimensional surface model of the insulator surface includes:

[0011] Acquire 3D point cloud data of the insulator surface through a 3D scanner and pre-process the 3D point cloud data;

[0012] According to the pre-processed 3D point cloud data, a 3D mesh model is constructed using a triangular meshing algorithm;

[0013] The three-dimensional mesh model is simplified and smoothed to obtain a three-dimensional surface model.

[0014] Optionally, obtaining the curved surface shape feature of the insulator surface includes:

[0015] Segment the 3D surface model and extract the local surfaces of the insulator shed and disk;

[0016] Calculate the curvature distribution characteristics of the local surface and generate a gray value curvature distribution histogram;

[0017] The shape type of the local surface is determined based on the peak position and half-height width of the gray value curvature distribution histogram;

[0018] Based on the shape type of the local surface, the surface shape characteristics of the insulator surface are obtained.

[0019] Optionally, obtaining ice thickness data on the surface of the insulator according to the three-dimensional point cloud data includes:

[0020] According to the 3D point cloud data, calculate the normal vector and curvature value of each point;

[0021] The plane area where the point is located is determined according to the normal vector and the curvature value, and the point cloud data of the plane area on the surface of the insulator is extracted;

[0022] The region growing algorithm is used to perform plane segmentation on the plane area point cloud data to obtain point cloud data sets of different plane areas;

[0023] The RANSAC algorithm is used to fit the plane. For the point cloud data sets in different plane areas, the distance from the point to the fitting plane is calculated, and the distance is used as the ice thickness value to obtain the ice thickness data of each plane area.

[0024] Optionally, ice thickness data is classified to obtain ice distribution areas of different thickness levels, including:

[0025] According to the distribution characteristics of ice thickness data, K-means clustering algorithm is used to perform cluster analysis to obtain ice thickness data categories of different thickness levels;

[0026] According to the ice thickness data category of each thickness grade, the corresponding plane area point cloud data is extracted, and the distribution areas of ice layers of different thickness grades are obtained through the spatial position relationship of the point cloud data.

[0027] Optionally, based on the curved surface shape characteristics of the insulator surface and the ice layer distribution area, obtaining the spray direction and angle of the nozzle includes:

[0028] According to the surface shape characteristics and ice distribution area, a fuzzy control rule base is established;

[0029] Obtain the fuzzy set of the spray direction and angle of the nozzle, and use the membership function to describe the fuzzy set;

[0030] Based on the fuzzy control rule base and fuzzy sets, the spray direction and angle of the nozzle are dynamically calculated using the Mamdani fuzzy reasoning algorithm. The maximum-minimum synthesis method is used in the reasoning process, and the center of gravity method is used for defuzzification.

[0031] Optionally, the water or air discharge parameters of the nozzles are controlled according to the predicted ice thickness and ice accumulation area, including:

[0032] The predicted ice thickness and ice accumulation area are compared with the expected ice thickness and ice accumulation area thresholds. If the predicted value exceeds the threshold range, the optimal water or wind output parameters are learned and controlled by setting a reward function based on the reinforcement learning algorithm.

[0033] Optionally, obtain the trained LSTM prediction model including:

[0034] According to the outlet water or air temperature, flow velocity and flow direction, obtain the ice thickness and ice accumulation area data in the historical time period and obtain the training sample set;

[0035] By setting the number of neurons in the input layer, hidden layer and output layer, as well as the number of iterations and learning rate, the long short-term memory neural network algorithm is used to train the training sample set to obtain the trained LSTM prediction model.

[0036] To achieve the above-mentioned purpose, the present invention also provides an intelligent nozzle control system based on three-dimensional imaging, comprising: a surface shape analysis module, an ice thickness analysis module, a fuzzy control adjustment module, and a nozzle parameter optimization module;

[0037] The surface shape analysis module is used to obtain the three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and obtain the surface shape characteristics of the insulator surface;

[0038] The ice thickness analysis module is used to obtain the ice thickness data on the insulator surface based on the three-dimensional point cloud data, and classify the ice thickness data to obtain the distribution areas of ice layers with different thickness levels;

[0039] A fuzzy control adjustment module is used to control the spray direction and angle of the nozzle based on the curved shape characteristics of the insulator surface and the ice layer distribution area;

[0040] The nozzle parameter optimization module is used to input the current water or air outlet parameter data into the trained LSTM prediction model, output the predicted ice thickness and ice accumulation area, and control the nozzle's water or air outlet parameters according to the predicted ice thickness and ice accumulation area, where the water or air outlet parameters are temperature, flow rate and flow direction.

[0041] To achieve the above-mentioned purpose, the present invention also provides an intelligent nozzle control device based on three-dimensional imaging, including an intelligent nozzle control system based on three-dimensional imaging.

[0042] The beneficial effects of the present invention are as follows: by obtaining the three-dimensional point cloud data of the insulator surface to reconstruct the three-dimensional model, the surface shape characteristics and ice thickness data of the insulator surface are obtained. Based on this information, the present invention uses a fuzzy control algorithm to adjust the spray direction and angle of the water gun and the hot air gun to adapt to the complex surface of the insulator. At the same time, a mapping relationship model between ice layer characteristics and water flow parameters is established through a neural network algorithm to optimize the deicing effect in real time. The invention realizes accurate, efficient and safe deicing of the insulator surface, and effectively ensures the stable operation of the transmission line. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 A flowchart of an intelligent nozzle control method based on three-dimensional imaging according to an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of a nozzle system according to an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of an array nozzle model 1 according to an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of an array nozzle model 2 according to an embodiment of the present invention;

[0048] Figure 5A schematic diagram of an array nozzle model 3 according to an embodiment of the present invention;

[0049] Figure 6 A schematic diagram of an array nozzle model 4 according to an embodiment of the present invention;

[0050] Figure 7 A flow chart of constructing a three-dimensional curved surface model of an insulator surface according to an embodiment of the present invention;

[0051] Figure 8 A flow chart of obtaining the curved surface shape characteristics of an insulator surface according to an embodiment of the present invention;

[0052] Fig. 9 This is a flow chart of obtaining ice thickness data on the surface of an insulator according to three-dimensional point cloud data in an embodiment of the present invention;

[0053] Fig.10 A flow chart of obtaining ice distribution areas of different thickness levels according to an embodiment of the present invention;

[0054] Fig.11 A flow chart of obtaining the spray direction and angle of the nozzle according to an embodiment of the present invention;

[0055] Fig.12 A framework diagram of an intelligent nozzle control system based on three-dimensional imaging according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Embodiment 1:

[0059] like Figure 1 This embodiment provides an intelligent nozzle control method based on three-dimensional imaging, including:

[0060] Acquire three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and acquire the surface shape features of the insulator surface;

[0061] The ice thickness data on the insulator surface is obtained based on the 3D point cloud data, and the ice thickness data is classified to obtain the distribution areas of ice layers with different thickness levels;

[0062] Based on the curved shape characteristics of the insulator surface and the distribution area of ​​the ice layer, the spray direction and angle of the nozzle are controlled;

[0063] The water or wind output parameter data at the current moment is input into the trained LSTM prediction model, and the predicted ice thickness and ice accumulation area are output. The water or wind output parameters of the nozzle are controlled according to the predicted ice thickness and ice accumulation area, where the water or wind output parameters are temperature, flow rate and flow direction.

[0064] In this embodiment, a gas-liquid dual-purpose nozzle is used to achieve different nozzle wind patterns, covering a large range of working surfaces at one time. The nozzles are independently universally adjustable and can cover a large angle and range of insulators at one time, such as Figure 2 In addition, this embodiment also provides four array nozzle models, such as Figure 3-6 The left part of the picture shows the shape of the object sprayed, and the right part shows the position of each nozzle.

[0065] Further, such as Figure 7 As shown in the figure, the three-dimensional surface model of the insulator surface includes:

[0066] Acquire 3D point cloud data of the insulator surface through a 3D scanner and pre-process the 3D point cloud data;

[0067] According to the pre-processed 3D point cloud data, a 3D mesh model is constructed using a triangular meshing algorithm;

[0068] The three-dimensional mesh model is simplified and smoothed to obtain a three-dimensional surface model.

[0069] Specifically, the 3D point cloud data of the insulator surface is obtained by a 3D scanner, with a scanning accuracy of up to 1 mm and a point cloud density of 100 points per square centimeter. The obtained 3D point cloud data is preprocessed, and a statistical filtering algorithm is used to remove noise points. The mean of the 10 points around each point is used as the threshold, and outliers that deviate more than 2 times the threshold are removed to improve the quality of the point cloud data. Based on the preprocessed 3D point cloud data, the Delaunay triangulation algorithm is used to reconstruct the 3D mesh model of the insulator surface. The mesh simplification algorithm is used to reduce the number of mesh facets by 30%, and then the Laplacian smoothing algorithm is used for 5 iterations of optimization to obtain a streamlined and smooth 3D surface model.

[0070] Further, such as Figure 8 As shown, the surface shape features of the insulator surface include:

[0071] Segment the 3D surface model and extract the local surfaces of the insulator shed and disk;

[0072] Calculate the curvature distribution characteristics of the local surface and generate a gray value curvature distribution histogram;

[0073] The shape type of the local surface is determined based on the peak position and half-height width of the gray value curvature distribution histogram;

[0074] Based on the shape type of the local surface, the surface shape characteristics of the insulator surface are obtained.

[0075] Specifically, the reconstructed three-dimensional surface model of the insulator is segmented, and the local surfaces of key parts such as the insulator shed and shed plate are extracted by the region growing algorithm. For the segmented local surfaces, the principal curvature method is used to calculate their curvature distribution characteristics, and a 256-level grayscale curvature distribution map is generated. The grayscale value distribution histogram is statistically analyzed, and the shape type of the local surface, such as concave, convex, saddle, etc., is determined by the peak position and half-height width of the histogram. Based on the shape type of the local surface, the surface shape characteristics of the insulator surface are obtained.

[0076] Further, such as Fig. 9 As shown, obtaining ice thickness data on the insulator surface based on three-dimensional point cloud data includes:

[0077] According to the 3D point cloud data, calculate the normal vector and curvature value of each point;

[0078] The plane area where the point is located is determined according to the normal vector and the curvature value, and the point cloud data of the plane area on the surface of the insulator is extracted;

[0079] The region growing algorithm is used to perform plane segmentation on the plane area point cloud data to obtain point cloud data sets of different plane areas;

[0080] The RANSAC algorithm is used to fit the plane. For the point cloud data sets in different plane areas, the distance from the point to the fitting plane is calculated, and the distance is used as the ice thickness value to obtain the ice thickness data of each plane area.

[0081] Specifically, based on the point cloud data of the insulator surface, the least squares method is used to fit the point cloud data, the normal vector and curvature value of each point are calculated, and 30° is used as the normal vector angle threshold and 1 is used as the curvature threshold to extract the plane area point cloud data of the insulator surface. For the plane area point cloud data, the regional growing algorithm is used for plane segmentation, with the preset distance as the distance threshold and the preset angle as the normal vector angle threshold to obtain the point cloud data set of different plane areas on the insulator surface. For each plane area point cloud data, the RANSAC algorithm is used to fit the plane, calculate the distance from the point to the fitting plane, and use the distance as the ice thickness value to obtain the ice thickness data of each plane area.

[0082] Further, such as Fig.10 As shown in the figure, the ice thickness data is classified to obtain the distribution areas of ice layers with different thickness levels, including:

[0083] According to the distribution characteristics of ice thickness data, K-means clustering algorithm is used to perform cluster analysis to obtain ice thickness data categories of different thickness levels;

[0084] According to the ice thickness data category of each thickness grade, the corresponding plane area point cloud data is extracted, and the distribution areas of ice layers of different thickness grades are obtained through the spatial position relationship of the point cloud data.

[0085] Specifically, according to the distribution characteristics of ice thickness data, K-means clustering algorithm is used for cluster analysis, and the number of clusters k is set to 3 to obtain three types of ice thickness data: light ice (thickness <1cm), moderate ice (1cm≤thickness <2cm) and heavy ice (thickness≥2cm). For each ice thickness category, the corresponding plane area point cloud data is extracted, and the distribution areas of ice layers of different thickness levels are obtained through the spatial position relationship of the point cloud data. In this embodiment, the ice thickness distribution area is also mapped to the three-dimensional model of the insulator, and different thickness levels are represented by different colors. The distribution map of ice thickness on the surface of the insulator is generated to provide an intuitive decision-making basis for insulator deicing.

[0086] Further, such as Fig.11 As shown, based on the curved surface shape characteristics of the insulator surface and the ice layer distribution area, the spray direction and angle of the nozzle are obtained including:

[0087] According to the surface shape characteristics and ice distribution area, a fuzzy control rule base is established;

[0088] Obtain the fuzzy set of the spray direction and angle of the nozzle, and use the membership function to describe the fuzzy set;

[0089] Based on the fuzzy control rule base and fuzzy sets, the spray direction and angle of the nozzle are dynamically calculated using the Mamdani fuzzy reasoning algorithm. The maximum-minimum synthesis method is used in the reasoning process, and the center of gravity method is used for defuzzification.

[0090] Specifically, according to the three-dimensional model of the insulator surface and the distribution area of ​​the ice layer, a fuzzy control rule base is established, and the fuzzy set of the spray direction and angle of the water gun and the hot air gun is set, and the membership function is used to describe the fuzzy set. The surface shape characteristics of the insulator surface and the distribution area of ​​the ice layer are used as the input of the fuzzy reasoning algorithm. The optimal spray direction and angle of the water gun and the hot air gun are dynamically calculated using the Mamdani fuzzy reasoning algorithm. The reasoning process adopts the maximum-minimum synthesis method, and the defuzzification adopts the center of gravity method. The rotating motor of the water gun and the hot air gun is controlled to adjust the spray direction and angle in real time to adapt to the surface characteristics of the insulator surface. The angular resolution of the rotating motor is 1° and the speed is 60rpm. The deicing effect of the insulator surface is monitored in real time by machine vision technology. A high-speed camera is used to collect images at a speed of 200 frames / second. The image is edge detected and morphologically processed to identify the residual ice layer area. If a blind spot area is found, the fuzzy control rule base is further optimized by the neural network algorithm to improve the deicing effect.

[0091] Furthermore, the water or air discharge parameters of the nozzles are controlled according to the predicted ice thickness and ice accumulation area, including:

[0092] The predicted ice thickness and ice accumulation area are compared with the expected ice thickness and ice accumulation area thresholds. If the predicted value exceeds the threshold range, the optimal water or wind output parameters are learned and controlled by setting a reward function based on the reinforcement learning algorithm.

[0093] Specifically, a nonlinear mapping relationship model between ice thickness, ice accumulation area and water flow temperature, flow velocity and flow direction is constructed. According to the established LSTM prediction model, the change trend of ice thickness and ice accumulation area corresponding to different water gun water discharge parameter combinations under the current water flow state is predicted in real time. The model prediction results are compared with the expected ice thickness and ice accumulation area thresholds. If the predicted value exceeds the threshold range, the dynamic adjustment mechanism of the water gun parameters is triggered. The dynamic adjustment mechanism is based on the reinforcement learning algorithm. It guides the agent to learn through the reward function and searches for the optimal combination of water gun water pressure, flow rate, temperature and other parameters to quickly reduce the ice thickness and ice accumulation area to below the target value. The optimal water gun water discharge parameters obtained by the reinforcement learning algorithm are passed to the nozzle control system to adjust the flow velocity, flow rate, temperature and other parameters of the water gun or fan in real time to keep the ice clearing process in the best working state. The changes in ice thickness and ice accumulation area are continuously monitored. When the values ​​are stable within the target threshold and last for a certain period of time, the ice clearing process is judged to be completed, and the adjustment of the water gun parameters is stopped to complete the entire intelligent ice clearing operation.

[0094] Furthermore, obtaining the trained LSTM prediction model includes:

[0095] According to the outlet water or air temperature, flow velocity and flow direction, obtain the ice thickness and ice accumulation area data in the historical time period and obtain the training sample set;

[0096] By setting the number of neurons in the input layer, hidden layer and output layer, as well as the number of iterations and learning rate, the long short-term memory neural network algorithm is used to train the training sample set to obtain the trained LSTM prediction model.

[0097] Embodiment 2:

[0098] like Fig.12 As shown, an intelligent nozzle control system based on three-dimensional imaging includes: a surface shape analysis module, an ice thickness analysis module, a fuzzy control adjustment module, and a nozzle parameter optimization module;

[0099] The surface shape analysis module is used to obtain the three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and obtain the surface shape characteristics of the insulator surface;

[0100] The ice thickness analysis module is used to obtain the ice thickness data on the insulator surface based on the three-dimensional point cloud data, and classify the ice thickness data to obtain the distribution areas of ice layers with different thickness levels;

[0101] A fuzzy control adjustment module is used to control the spray direction and angle of the nozzle based on the curved shape characteristics of the insulator surface and the ice layer distribution area;

[0102] The nozzle parameter optimization module is used to input the current water or air outlet parameter data into the trained LSTM prediction model, output the predicted ice thickness and ice accumulation area, and control the nozzle's water or air outlet parameters according to the predicted ice thickness and ice accumulation area, where the water or air outlet parameters are temperature, flow rate and flow direction.

[0103] Embodiment three:

[0104] This embodiment provides an intelligent nozzle control device based on three-dimensional imaging, including an intelligent nozzle control system based on three-dimensional imaging.

[0105] The above embodiments are only descriptions of the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent nozzle control method based on three-dimensional imaging, characterized in that: include: Acquire three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and acquire the surface shape features of the insulator surface; The ice thickness data on the insulator surface is obtained based on the 3D point cloud data, and the ice thickness data is classified to obtain the distribution areas of ice layers with different thickness levels; Based on the curved shape characteristics of the insulator surface and the distribution area of ​​the ice layer, the spray direction and angle of the nozzle are controlled; The water or wind output parameter data at the current moment is input into the trained LSTM prediction model, and the predicted ice thickness and ice accumulation area are output. The water or wind output parameters of the nozzle are controlled according to the predicted ice thickness and ice accumulation area, where the water or wind output parameters are temperature, flow rate and flow direction.

2. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: The three-dimensional surface model of the insulator surface includes: Acquire 3D point cloud data of the insulator surface through a 3D scanner and pre-process the 3D point cloud data; According to the pre-processed 3D point cloud data, a 3D mesh model is constructed using a triangular meshing algorithm; The three-dimensional mesh model is simplified and smoothed to obtain a three-dimensional surface model.

3. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: The surface shape features of the insulator surface include: Segment the 3D surface model and extract the local surfaces of the insulator shed and disk; Calculate the curvature distribution characteristics of the local surface and generate a gray value curvature distribution histogram; The shape type of the local surface is determined based on the peak position and half-height width of the gray value curvature distribution histogram; Based on the shape type of the local surface, the surface shape characteristics of the insulator surface are obtained.

4. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: Acquiring ice thickness data on the surface of the insulator according to the three-dimensional point cloud data includes: calculating the normal vector and curvature value of each point according to the three-dimensional point cloud data; The plane area where the point is located is determined according to the normal vector and the curvature value, and the point cloud data of the plane area on the surface of the insulator is extracted; The region growing algorithm is used to perform plane segmentation on the plane area point cloud data to obtain point cloud data sets of different plane areas; The RANSAC algorithm is used to fit the plane. For the point cloud data sets in different plane areas, the distance from the point to the fitting plane is calculated, and the distance is used as the ice thickness value to obtain the ice thickness data of each plane area.

5. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: Classifying the ice thickness data to obtain the distribution areas of ice layers of different thickness levels includes: performing cluster analysis using the K-means clustering algorithm according to the distribution characteristics of the ice thickness data to obtain the categories of ice thickness data of different thickness levels; According to the ice thickness data category of each thickness grade, the corresponding plane area point cloud data is extracted, and the distribution areas of ice layers of different thickness grades are obtained through the spatial position relationship of the point cloud data.

6. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: Based on the curved surface shape characteristics and ice layer distribution area of ​​the insulator surface, obtaining the spray direction and angle of the nozzle includes: establishing a fuzzy control rule library according to the curved surface shape characteristics and ice layer distribution area; Obtain the fuzzy set of the spray direction and angle of the nozzle, and use the membership function to describe the fuzzy set; Based on the fuzzy control rule base and fuzzy sets, the spray direction and angle of the nozzle are dynamically calculated using the Mamdani fuzzy reasoning algorithm. The maximum-minimum synthesis method is used in the reasoning process, and the center of gravity method is used for defuzzification.

7. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: The water or air discharge parameters of the nozzles are controlled according to the predicted ice thickness and ice accumulation area, including: The predicted ice thickness and ice accumulation area are compared with the expected ice thickness and ice accumulation area thresholds. If the predicted value exceeds the threshold range, the optimal water or wind output parameters are learned and controlled by setting a reward function based on the reinforcement learning algorithm.

8. The intelligent nozzle control method based on three-dimensional imaging according to claim 1, characterized in that: Obtaining the trained LSTM prediction model includes: According to the outlet water or air temperature, flow velocity and flow direction, obtain the ice thickness and ice accumulation area data in the historical time period and obtain the training sample set; By setting the number of neurons in the input layer, hidden layer and output layer, as well as the number of iterations and learning rate, the long short-term memory neural network algorithm is used to train the training sample set to obtain the trained LSTM prediction model.

9. An intelligent nozzle control system based on three-dimensional imaging, characterized in that: include: Surface shape analysis module, ice thickness analysis module, fuzzy control adjustment module, nozzle parameter optimization module; The surface shape analysis module is used to obtain the three-dimensional point cloud data of the insulator surface, construct a three-dimensional surface model of the insulator surface according to the three-dimensional point cloud data, and obtain the surface shape characteristics of the insulator surface; The ice thickness analysis module is used to obtain the ice thickness data on the insulator surface based on the three-dimensional point cloud data, and classify the ice thickness data to obtain the distribution areas of ice layers with different thickness levels; A fuzzy control adjustment module is used to control the spray direction and angle of the nozzle based on the curved shape characteristics of the insulator surface and the ice layer distribution area; The nozzle parameter optimization module inputs the current water flow or wind flow temperature, flow velocity and flow direction data into the trained LSTM prediction model, outputs the predicted ice thickness and ice accumulation area, and controls the nozzle's water or wind output parameters according to the predicted ice thickness and ice accumulation area.

10. An intelligent nozzle control device based on three-dimensional imaging, characterized in that: It includes the intelligent nozzle control system based on three-dimensional imaging as described in claim 9.

Images