Obstacle avoidance method and system for electric power inspection robot

Abstract

The invention discloses an obstacle avoidance method and system of an electric power inspection robot, and relates to the technical field of electric power inspection robot obstacle avoidance, the system comprises an ice condition acquisition module, an icing prediction processing module, an icing risk assessment module and an obstacle avoidance execution module, the ice condition acquisition module is used for acquiring icing data of an obstacle avoidance environment; the icing prediction processing module is divided into an icing data processing unit, an icing feature extraction unit and an icing model construction unit; the icing data processing unit is used for converting echo time into ice layer thickness data and converting ice piton surface three-dimensional point cloud data into an ice piton three-dimensional model; the icing feature extraction unit is used for extracting ice piton length data; the icing model construction unit is used for obtaining obstacle avoidance results corresponding to different ice layers and ice pitons respectively, and constructing an ice layer obstacle avoidance model and an ice piton obstacle avoidance model; the icing risk assessment module is used for analyzing ice layer thickness data and ice piton length data; and the obstacle avoidance execution module receives the analysis result and executes corresponding operation.

Description

Technical Field

[0001] The present invention relates to the technical field of obstacle avoidance for electric power inspection robots, and in particular to an obstacle avoidance method and system for electric power inspection robots. Background Art

[0002] The obstacle avoidance method and system background technology of the power inspection robot originates from the growth of the inspection demand of the power system and the rapid development of robot technology. With the large-scale commissioning of power facilities, the inspection workload has increased dramatically. The traditional manual inspection method can no longer meet the needs of efficiency and safety. Therefore, the power inspection robot came into being to replace manual inspection tasks. In terms of obstacle avoidance, robot technology provides a variety of methods, such as obstacle avoidance based on sensor fusion, obstacle avoidance based on artificial intelligence, and obstacle avoidance based on map construction. These methods use information from multiple sensors such as lidar, ultrasonic sensors, and infrared sensors, combined with artificial intelligence technologies such as deep learning and fuzzy logic control, so that the robot can perceive the surrounding environment in real time, accurately identify obstacles, and autonomously plan inspection routes to achieve efficient and safe inspections. With the continuous advancement of technology, the obstacle avoidance method and system of the power inspection robot will become more intelligent and autonomous, providing more reliable guarantees for the safe and stable operation of the power system;

[0003] Although the obstacle avoidance methods and systems of existing power inspection robots have made great progress, there are still some problems that need to be optimized. When the power inspection robot is working in winter, due to the large ice cones and thick ice layers in some areas, which occupy most of the line space, the power inspection robot cannot avoid obstacles by bypassing measures. In addition, the icing conditions in different line sections are different. The power inspection robot cannot take corresponding obstacle avoidance measures according to the icing conditions, which affects its working performance. Summary of the invention

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: an obstacle avoidance system for a power inspection robot, comprising an ice condition collection module, an icing prediction and processing module, an icing risk assessment module and an obstacle avoidance execution module;

[0005] The ice condition collection module is used to obtain ice obstacle data in the power inspection environment;

[0006] The icing prediction and processing module is divided into an icing data processing unit, an icing feature extraction unit and an icing model construction unit, wherein the icing data processing unit is used to convert the echo time into ice layer thickness data, and convert the three-dimensional point cloud data of the ice cone surface into an ice cone three-dimensional model; the icing feature extraction unit is used to extract ice cone length data from the ice cone three-dimensional model; the icing model construction unit is used to construct an ice layer obstacle avoidance model and an ice cone obstacle avoidance model to obtain obstacle avoidance results corresponding to different ice layer data and ice cone data respectively.

[0007] The icing risk assessment module analyzes the ice thickness data and the ice cone length data to assess the risk of the robot operation and transmits the analysis results to the obstacle avoidance execution module;

[0008] The obstacle avoidance execution module receives the evaluation result of the operation risk of the inspection robot and executes corresponding operations to avoid obstacles.

[0009] Furthermore, the obstacle avoidance environment icing data is ice layer data and ice cone data in the moving area of ​​the power inspection robot, and the acquisition process thereof includes:

[0010] A1. Equip the power inspection robot with an ice detector and a 3D scanner, wherein the ice detector and the 3D scanner are both installed at the front end of the power inspection robot;

[0011] A2. Use an ice detector to send high-frequency electromagnetic waves to the target ice layer in real time, receive reflected signals, and record their echo time;

[0012] A3. Adjust the scanning parameters of the 3D scanner to fit the size of the ice cone. The 3D scanner projects a laser beam onto the surface of the ice cone and captures the reflected light signal. After multiple scans, the 3D point cloud data of the ice cone surface is obtained.

[0013] Furthermore, the ice data processing unit converts the echo time into ice thickness data by calculation, and converts the three-dimensional point cloud data of the ice cone surface into a three-dimensional model of the ice cone using a Delaunay triangulation algorithm. The specific process includes:

[0014] B1. The calculation formula for converting the echo signal into ice thickness data is:

[0015] D=ct / 2

[0016] Where D is the thickness of the ice layer, c is the propagation speed of electromagnetic waves in the ice layer, and t is the echo time;

[0017] B2. Preprocess the 3D point cloud data of the ice cone surface, take the preprocessed point cloud data as input, use the Delaunay triangulation algorithm to build a mesh model composed of triangles, connect adjacent points according to the definition of Delaunay edges, and form a 3D model of the ice cone.

[0018] Furthermore, in the icing feature extraction unit, the process of extracting ice cone length data from the ice cone three-dimensional model includes:

[0019] Input the 3D model of the ice cone into the Blender modeling software, adjust the viewing angle and position in the Blender 3D view, obtain the tip and bottom points of the ice cone, and calculate the length of the ice cone based on the geometric principle of 3D space. The calculation formula is:

[0020]

[0021] Among them, L is the length of the ice cone, the coordinates of the tip of the ice cone are M(x1,y1,z1), and the coordinates of the bottom of the ice cone are N(x2,y2,z2).

[0022] Furthermore, the process of the icing model building unit obtaining obstacle avoidance results corresponding to different ice layer data and different ice cone data includes:

[0023] C1. Clean and organize the collected ice thickness and ice cone length data;

[0024] C2. For ice thickness, preset ice thickness thresholds, which are 15mm for the first threshold and 100mm for the second threshold. Input the preprocessed ice thickness data into the clustering algorithm, set the number of clusters to 3, and divide the ice thickness data into different clusters according to the similarity of the ice data. According to the preset ice thickness threshold, judge the ice thickness data of each cluster. When the ice thickness is less than 15mm, it is the first cluster; when the ice thickness is between 15mm and 100mm, it is the second cluster; when the ice thickness is greater than 100mm, it is the third cluster. The generated clusters are divided into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. Output the clustering results to obtain obstacle avoidance results corresponding to different ice layers.

[0025] C3. For the length of ice cones, ice cone length thresholds are set, which are the first threshold of 5 cm and the second threshold of 15 cm. The preprocessed ice cone length data is input into the clustering algorithm, and the number of clusters is set to 3. The clustering algorithm divides the ice cone length data into different clusters according to the similarity of the ice cone length data. The ice cone data of each cluster is judged according to the preset threshold. When the ice cone length is less than 2 cm, it is the first cluster; when the ice cone length is between 5 and 15 cm, it is the second cluster; when the ice cone length is higher than 15 cm, it is the third cluster. The generated clusters are divided into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. The clustering results are output to obtain the obstacle avoidance results corresponding to different ice cones.

[0026] Furthermore, the process of constructing the ice layer obstacle avoidance model by the icing model construction unit includes:

[0027] The ice thickness data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, the processed ice thickness data and its obstacle avoidance result data are used as data sets, and the data sets are divided into training set and validation set. The training set divides the ice thickness characteristics on the root node according to the first ice thickness threshold and the second ice thickness threshold of the ice thickness, and creates corresponding leaf nodes, and records the corresponding obstacle avoidance results on each leaf node. When the corresponding value of the input root node is lower than the first ice thickness threshold of the ice thickness, the leaf node outputs a result with an obstacle avoidance difficulty of 0%; when the corresponding value of the input root node is between the first ice thickness threshold and the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 50%; when the corresponding value of the input root node is higher than the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 100%, and a preliminary ID3 decision tree model is constructed. The constructed ID3 decision tree model is then verified through the validation set, the threshold and model parameters are adjusted, the ID3 decision tree model is optimized, and an ice obstacle avoidance model is obtained.

[0028] Furthermore, the process of constructing the ice cone obstacle avoidance model by the icing model construction unit includes:

[0029] The ice cone length data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, the processed ice cone length data and its obstacle avoidance result data are used as data sets, and the data sets are divided into training set and validation set. The training set divides the ice cone length features on the root node according to the first ice cone length threshold and the second ice cone length threshold, and creates corresponding leaf nodes, and records the corresponding obstacle avoidance results on each leaf node. When the input root node corresponding ice cone length value is lower than the first ice cone length threshold of the ice cone length, the leaf node outputs the result that the obstacle avoidance difficulty is 0%; when the input root node corresponding ice cone length value is between the first ice cone length threshold and the second ice cone length threshold, the leaf node outputs the result that the obstacle avoidance difficulty is 50%; when the input root node corresponding ice cone length value is higher than the second ice cone length threshold, the leaf node outputs the result that the obstacle avoidance difficulty is 100%, and a preliminary ID3 decision tree model is constructed. Then, the constructed ID3 decision tree model is verified through the validation set, the threshold and model parameters are adjusted, the ID3 decision tree model is optimized, and the ice cone obstacle avoidance model is obtained.

[0030] Furthermore, the icing risk assessment module analyzes the ice thickness data and the ice cone length data in the following process:

[0031] S1. Analyze the collected ice thickness data and input them into the ice obstacle avoidance model. When the input value is lower than the first ice thickness threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice thickness threshold and the second ice thickness threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice thickness threshold, the output obstacle avoidance difficulty is 100%;

[0032] S2. Analyze the collected ice cone length data and input them into the ice cone obstacle avoidance model. When the input value is lower than the first ice cone length threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice cone length threshold and the second ice cone length threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice cone length threshold, the output obstacle avoidance difficulty is 100%;

[0033] S3. Wirelessly transmit the obstacle avoidance difficulty percentage result to the obstacle avoidance execution module.

[0034] Furthermore, the process of the obstacle avoidance execution module receiving the analysis result and executing the corresponding operation includes:

[0035] When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance; when the obstacle avoidance difficulty is 50%, a command is issued to mobilize the inspection robot to start the mechanical vibration de-icing device; when the obstacle avoidance difficulty is 100%, a command is issued to dispatch the inspection robot to stop moving forward, change the inspection route, and issue an alarm to the staff for handling.

[0036] Furthermore, an obstacle avoidance method for a power inspection robot is implemented based on the obstacle avoidance system of the power inspection robot described above, and is characterized by comprising the following steps:

[0037] Step 1: Use an ice detector to send electromagnetic waves to the ice layer, collect its echo time, and use a 3D scanner to collect 3D point cloud data of the ice cone;

[0038] Step 2: Convert the echo time into ice thickness data by calculation, convert the 3D point cloud data of the ice cone surface into a 3D model of the ice cone using the Delaunay triangulation algorithm, and then extract the ice cone length data from the 3D model of the ice cone using the Blender modeling software;

[0039] Step 3: Set the thresholds of ice thickness data and ice cone length data, and output obstacle avoidance results corresponding to different ice layer data and different ice cone data through clustering algorithm;

[0040] Step 4: Use ID3 decision tree algorithm to build ice layer obstacle avoidance model and ice cone obstacle avoidance model;

[0041] Step 5: Analyze the real-time ice thickness data and ice cone length data, and transmit the analysis results to the obstacle avoidance execution module;

[0042] Step 6. Receive the analysis results. When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance. When the obstacle avoidance difficulty is 50%, issue a command to mobilize the inspection robot to start the mechanical vibration de-icing device. When the obstacle avoidance difficulty is 100%, issue a command to dispatch the inspection robot to stop moving forward, change the inspection route, and send an alarm to the staff for handling.

[0043] Beneficial effects of the present invention: The obstacle avoidance method and system of the power inspection robot closely combines the three-dimensional data acquisition technology, three-dimensional modeling technology, feature extraction technology, clustering algorithm analysis technology, model building technology and modern information technology in the system of the present invention, integrates ice detection technology and three-dimensional scanning technology, accurately captures the ice layer data and ice cone data that the inspection robot needs to avoid obstacles, and achieves real-time and comprehensive monitoring of the centralized monitoring system of the inspection robot through the ID3 decision tree algorithm. Through the ice layer thickness data, ice cone length data, ice layer obstacle avoidance model and ice cone obstacle avoidance model obtained after data processing, the obstacle avoidance difficulty prediction of the inspection robot when working in an icing environment is analyzed in real time, and abnormal conditions are monitored in real time. It can timely respond to anomalies encountered during the operation of the system, and solve the problem that when the power inspection robot is working in winter, the ice cones are large and the ice layer is thick in some areas, occupying most of the line space, making it impossible for the power inspection robot to avoid obstacles by bypassing measures. In addition, the icing conditions in different line sections are different, and the power inspection robot cannot take corresponding obstacle avoidance measures according to the icing conditions, affecting its working performance. It ensures that the method in the present invention can refine the dynamic monitoring standard of the obstacle avoidance system of the power inspection robot within a more precise range, so that the monitored data becomes a more accurate indicator under the same conditions. The development and application of this method have significantly enhanced the intelligence of the inspection robot in the obstacle avoidance process. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A block diagram of an obstacle avoidance system of an electric power inspection robot according to the present invention;

[0045] Figure 2 Flow chart of an obstacle avoidance method for a power inspection robot of the present invention DETAILED DESCRIPTION

[0046] 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.

[0047] like Figure 1 and Figure 2As shown, the present invention provides a technical solution: an obstacle avoidance system for a power inspection robot, comprising an ice condition collection module, an icing prediction and processing module, an icing risk assessment module and an obstacle avoidance execution module;

[0048] Ice condition collection module, used to obtain ice obstacle data in power inspection environment;

[0049] The icing prediction processing module is divided into an icing data processing unit, an icing feature extraction unit and an icing model construction unit, wherein the icing data processing unit is used to convert the echo time into ice layer thickness data, and convert the three-dimensional point cloud data of the ice cone surface into an ice cone three-dimensional model; the icing feature extraction unit is used to extract ice cone length data from the ice cone three-dimensional model; the icing model construction unit is used to construct an ice layer obstacle avoidance model and an ice cone obstacle avoidance model to obtain obstacle avoidance results corresponding to different ice layer data and ice cone data respectively.

[0050] The icing risk assessment module analyzes the ice thickness data and ice cone length data to assess the risk of robot operation and transmits the analysis results to the obstacle avoidance execution module;

[0051] The obstacle avoidance execution module receives the assessment results of the inspection robot's operation risks and performs corresponding operations to avoid obstacles.

[0052] The obstacle avoidance environment icing data is the ice layer data and ice cone data of the moving area of ​​the power inspection robot. The acquisition process includes:

[0053] A1. Equip the power inspection robot with an ice detector and a 3D scanner, wherein the ice detector and the 3D scanner are both installed at the front end of the power inspection robot;

[0054] A2. Use an ice detector to send high-frequency electromagnetic waves to the target ice layer in real time, receive reflected signals, and record their echo time;

[0055] A3. Adjust the scanning parameters of the 3D scanner to fit the size of the ice cone. The 3D scanner projects a laser beam onto the surface of the ice cone and captures the reflected light signal. After multiple scans, the 3D point cloud data of the ice cone surface is obtained.

[0056] The ice data processing unit converts the echo time into ice thickness data by calculation, and converts the 3D point cloud data of the ice cone surface into a 3D model of the ice cone using the Delaunay triangulation algorithm. The specific process includes:

[0057] B1. The calculation formula for converting the echo signal into ice thickness data is:

[0058] D=ct / 2

[0059] Where D is the thickness of the ice layer, c is the propagation speed of electromagnetic waves in the ice layer, and t is the echo time;

[0060] B2. Preprocess the 3D point cloud data of the ice cone surface, take the preprocessed point cloud data as input, use the Delaunay triangulation algorithm to build a mesh model composed of triangles, connect adjacent points according to the definition of Delaunay edges, and form a 3D model of the ice cone.

[0061] In the icing feature extraction unit, the process of extracting ice cone length data from the ice cone three-dimensional model includes:

[0062] Input the 3D model of the ice cone into the Blender modeling software, adjust the viewing angle and position in the Blender 3D view, obtain the tip and bottom points of the ice cone, and calculate the length of the ice cone based on the geometric principle of 3D space. The calculation formula is:

[0063]

[0064] Among them, L is the length of the ice cone, the coordinates of the tip of the ice cone are M(x1,y1,z1), and the coordinates of the bottom of the ice cone are N(x2,y2,z2).

[0065] The process of the icing model building unit obtaining obstacle avoidance results corresponding to different ice layer data and different ice cone data includes:

[0066] C1. Clean and organize the collected ice thickness and ice cone length data;

[0067] C2. For ice thickness, preset ice thickness thresholds, which are 15 mm for the first threshold and 100 mm for the second threshold. Input the preprocessed ice thickness data into the clustering algorithm, set the number of clusters to 3, and divide the ice thickness data into different clusters based on the similarity of the ice data. According to the preset ice thickness threshold, judge the ice thickness data of each cluster. When the ice thickness is less than 15 mm, it is the first cluster; when the ice thickness is between 15 mm and 100 mm, it is the second cluster; when the ice thickness is greater than 100 mm, it is the third cluster. The generated clusters are divided into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. Output the clustering results to obtain obstacle avoidance results corresponding to different ice layers.

[0068] C3. For the length of ice cones, ice cone length thresholds are set, which are the first threshold of 5 cm and the second threshold of 15 cm. The preprocessed ice cone length data is input into the clustering algorithm, and the number of clusters is set to 3. The clustering algorithm divides the ice cone length data into different clusters according to the similarity of the ice cone length data. The ice cone data of each cluster is judged according to the preset threshold. When the ice cone length is less than 2 cm, it is the first cluster; when the ice cone length is between 5 and 15 cm, it is the second cluster; when the ice cone length is higher than 15 cm, it is the third cluster. The generated clusters are divided into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. The clustering results are output to obtain the obstacle avoidance results corresponding to different ice cones.

[0069] The icing model building unit builds an ice obstacle avoidance model in the following process:

[0070] The ice thickness data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, the processed ice thickness data and its obstacle avoidance result data are used as data sets, and the data sets are divided into training set and validation set. The training set divides the ice thickness characteristics on the root node according to the first ice thickness threshold and the second ice thickness threshold of the ice thickness, and creates corresponding leaf nodes, and records the corresponding obstacle avoidance results on each leaf node. When the corresponding value of the input root node is lower than the first ice thickness threshold of the ice thickness, the leaf node outputs a result with an obstacle avoidance difficulty of 0%; when the corresponding value of the input root node is between the first ice thickness threshold and the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 50%; when the corresponding value of the input root node is higher than the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 100%, and a preliminary ID3 decision tree model is constructed. The constructed ID3 decision tree model is then verified through the validation set, the threshold and model parameters are adjusted, the ID3 decision tree model is optimized, and an ice obstacle avoidance model is obtained.

[0071] The icing model building unit, the process of building the ice cone obstacle avoidance model includes:

[0072] The ice cone length data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, the processed ice cone length data and its obstacle avoidance result data are used as data sets, and the data sets are divided into training set and validation set. The training set divides the ice cone length features on the root node according to the first ice cone length threshold and the second ice cone length threshold, and creates corresponding leaf nodes, and records the corresponding obstacle avoidance results on each leaf node. When the input root node corresponding ice cone length value is lower than the first ice cone length threshold of the ice cone length, the leaf node outputs the result that the obstacle avoidance difficulty is 0%; when the input root node corresponding ice cone length value is between the first ice cone length threshold and the second ice cone length threshold, the leaf node outputs the result that the obstacle avoidance difficulty is 50%; when the input root node corresponding ice cone length value is higher than the second ice cone length threshold, the leaf node outputs the result that the obstacle avoidance difficulty is 100%, and a preliminary ID3 decision tree model is constructed. Then, the constructed ID3 decision tree model is verified through the validation set, the threshold and model parameters are adjusted, the ID3 decision tree model is optimized, and the ice cone obstacle avoidance model is obtained.

[0073] Icing risk assessment module, the process of analyzing ice thickness data and ice cone length data includes:

[0074] S1. Analyze the collected ice thickness data and input them into the ice obstacle avoidance model. When the input value is lower than the first ice thickness threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice thickness threshold and the second ice thickness threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice thickness threshold, the output obstacle avoidance difficulty is 100%;

[0075] S2. Analyze the collected ice cone length data and input them into the ice cone obstacle avoidance model. When the input value is lower than the first ice cone length threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice cone length threshold and the second ice cone length threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice cone length threshold, the output obstacle avoidance difficulty is 100%;

[0076] S3. Wirelessly transmit the obstacle avoidance difficulty percentage result to the obstacle avoidance execution module.

[0077] The obstacle avoidance execution module receives the analysis results and performs the corresponding operations, including:

[0078] When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance; when the obstacle avoidance difficulty is 50%, a command is issued to mobilize the inspection robot to start the mechanical vibration de-icing device; when the obstacle avoidance difficulty is 100%, a command is issued to dispatch the inspection robot to stop moving forward, change the inspection route, and issue an alarm to the staff for handling.

[0079] An obstacle avoidance method for a power inspection robot is implemented based on the obstacle avoidance system of the power inspection robot, and comprises the following steps:

[0080] Step 1: Use an ice detector to send electromagnetic waves to the ice layer, collect its echo time, and use a 3D scanner to collect 3D point cloud data of the ice cone;

[0081] Step 2: Convert the echo time into ice thickness data by calculation, convert the 3D point cloud data of the ice cone surface into a 3D model of the ice cone using the Delaunay triangulation algorithm, and then extract the ice cone length data from the 3D model of the ice cone using the Blender modeling software;

[0082] Step 3: Set the thresholds of ice thickness data and ice cone length data, and output obstacle avoidance results corresponding to different ice layer data and different ice cone data through clustering algorithm;

[0083] Step 4: Use ID3 decision tree algorithm to build ice layer obstacle avoidance model and ice cone obstacle avoidance model;

[0084] Step 5: Analyze the real-time ice thickness data and ice cone length data, and transmit the analysis results to the obstacle avoidance execution module;

[0085] Step 6. Receive the analysis results. When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance. When the obstacle avoidance difficulty is 50%, issue a command to mobilize the inspection robot to start the mechanical vibration de-icing device. When the obstacle avoidance difficulty is 100%, issue a command to dispatch the inspection robot to stop moving forward, change the inspection route, and send an alarm to the staff for handling.

[0086] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. Without further restrictions. The sentence "comprises an element defined by ... does not exclude the existence of other identical elements in the process, method, article or device including the element". Although embodiments of the present invention have been shown and described, it is understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the attached claims and their equivalents.

Claims

1. An obstacle avoidance system for a power inspection robot, comprising an ice condition collection module, an icing prediction processing module, an icing risk assessment module and an obstacle avoidance execution module, characterized in that: The ice condition collection module is used to obtain ice obstacle data in the power inspection environment; The icing prediction and processing module is divided into an icing data processing unit, an icing feature extraction unit and an icing model construction unit, wherein the icing data processing unit is used to convert the echo time into ice layer thickness data, and convert the three-dimensional point cloud data of the ice cone surface into an ice cone three-dimensional model; the icing feature extraction unit is used to extract ice cone length data from the ice cone three-dimensional model; the icing model construction unit is used to construct an ice layer obstacle avoidance model and an ice cone obstacle avoidance model to obtain obstacle avoidance results corresponding to different ice layer data and ice cone data respectively. The icing risk assessment module analyzes the ice thickness data and the ice cone length data to assess the risk of the robot operation and transmits the analysis results to the obstacle avoidance execution module; The obstacle avoidance execution module receives the evaluation result of the operation risk of the inspection robot and executes corresponding operations to avoid obstacles.

2. The obstacle avoidance system of a power inspection robot according to claim 1, characterized in that: The icing obstacle data of the power inspection environment is ice layer data and ice cone data of the moving area of ​​the power inspection robot, and the acquisition process includes: A1. Equip the power inspection robot with an ice detector and a 3D scanner, wherein the ice detector and the 3D scanner are both installed at the front end of the power inspection robot; A2. Use an ice detector to send high-frequency electromagnetic waves to the target ice layer in real time, receive reflected signals, and record their echo time; A3. Adjust the scanning parameters of the 3D scanner to fit the size of the ice cone. The 3D scanner projects a laser beam onto the surface of the ice cone and captures the reflected light signal. After multiple scans, the 3D point cloud data of the ice cone surface is obtained.

3. The obstacle avoidance system of a power inspection robot according to claim 2, characterized in that: The ice data processing unit converts the echo time into ice thickness data by calculation, and converts the three-dimensional point cloud data of the ice cone surface into a three-dimensional model of the ice cone using the Delaunay triangulation algorithm. The specific process includes: B1. The calculation formula for converting the echo signal into ice thickness data is: D=ct / 2 Where D is the thickness of the ice layer, c is the propagation speed of electromagnetic waves in the ice layer, and t is the echo time; B2. Preprocess the 3D point cloud data of the ice cone surface, take the preprocessed point cloud data as input, use the Delaunay triangulation algorithm to build a mesh model composed of triangles, connect adjacent points according to the definition of Delaunay edges, and form a 3D model of the ice cone.

4. The obstacle avoidance system of the power inspection robot according to claim 3 is characterized in that: In the icing feature extraction unit, the process of extracting ice cone length data from the ice cone three-dimensional model includes: Input the 3D model of the ice cone into the Blender modeling software, adjust the viewing angle and position in the Blender 3D view, obtain the tip and bottom points of the ice cone, and calculate the length of the ice cone based on the geometric principle of 3D space. The calculation formula is: Among them, L is the length of the ice cone, the coordinates of the tip of the ice cone are M(x1,y1,z1), and the coordinates of the bottom of the ice cone are N(x2,y2,z2).

5. The obstacle avoidance system of a power inspection robot according to claim 4, characterized in that: The process of the icing model building unit obtaining obstacle avoidance results corresponding to different ice layer data and different ice cone data includes: C1. Clean and organize the collected ice thickness and ice cone length data; C2. For ice thickness, preset ice thickness thresholds, namely the first ice thickness threshold and the second ice thickness threshold. Input the preprocessed ice thickness data into the clustering algorithm, set the number of clusters to 3, and divide the ice thickness data into different clusters according to the similarity of the ice data. According to the preset ice thickness threshold, judge the ice thickness data of each cluster and divide it into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. Output the clustering results to obtain obstacle avoidance results corresponding to different ice layers. C3. For the length of ice cones, preset ice cone length thresholds, namely the first ice cone length threshold and the second ice cone length threshold. Input the preprocessed ice cone length data into the clustering algorithm, set the number of clusters to 3, and the clustering algorithm divides the ice cone length data into different clusters according to the similarity of the ice cone lengths. According to the preset threshold, the ice cone data of each cluster is judged and divided into three clustering results: no impact on obstacle avoidance, impact on obstacle avoidance, and failure to avoid obstacles. Output the clustering results to obtain the obstacle avoidance results corresponding to different ice cones.

6. The obstacle avoidance system of a power inspection robot according to claim 5, characterized in that: The process of constructing the ice layer obstacle avoidance model by the icing model construction unit includes: The ice thickness data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, and the processed ice thickness data and its obstacle avoidance result data are used as data sets. The data sets are divided into training set and validation set. The training set divides the ice thickness characteristics on the root node according to the first ice thickness threshold and the second ice thickness threshold, and creates corresponding leaf nodes. The corresponding obstacle avoidance results are recorded on each leaf node. When the corresponding value of the input root node is lower than the first ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 0%; when the corresponding value of the input root node is between the first ice thickness threshold and the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 50%; when the corresponding value of the input root node is higher than the second ice thickness threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 100%. A preliminary ID3 decision tree model is constructed, and then the constructed ID3 decision tree model is verified through the validation set. The threshold and model parameters are adjusted to optimize the ID3 decision tree model and obtain an ice obstacle avoidance model.

7. The obstacle avoidance system of a power inspection robot according to claim 6, characterized in that: The process of constructing the ice cone obstacle avoidance model by the icing model construction unit includes: The ice cone length data and its obstacle avoidance result data are standardized in ID3 decision tree format, the decision tree is initialized, the processed ice cone length data and its obstacle avoidance result data are used as data sets, and the data sets are divided into training set and validation set. The training set divides the ice cone length features on the root node according to the first ice cone length threshold and the second ice cone length threshold, and creates corresponding leaf nodes, and records the corresponding obstacle avoidance results on each leaf node. When the input root node corresponding ice cone length value is lower than the first ice cone length threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 0%; when the input root node corresponding ice cone length value is between the first ice cone length threshold and the second ice cone length threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 50%; when the input root node corresponding ice cone length value is higher than the second ice cone length threshold, the leaf node outputs a result with an obstacle avoidance difficulty of 100%, and a preliminary ID3 decision tree model is constructed. The constructed ID3 decision tree model is then verified through the validation set, the threshold and model parameters are adjusted, the ID3 decision tree model is optimized, and the ice cone obstacle avoidance model is obtained.

8. The obstacle avoidance system of a power inspection robot according to claim 7, characterized in that: The icing risk assessment module analyzes the ice thickness data and the ice cone length data in the following process: S1. Analyze the collected ice thickness data and input them into the ice obstacle avoidance model. When the input value is lower than the first ice thickness threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice thickness threshold and the second ice thickness threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice thickness threshold, the output obstacle avoidance difficulty is 100%; S2. Analyze the collected ice cone length data and input them into the ice cone obstacle avoidance model. When the input value is lower than the first ice cone length threshold, the output obstacle avoidance difficulty is 0%; when the input value is between the first ice cone length threshold and the second ice cone length threshold, the output obstacle avoidance difficulty is 50%; when the input value is higher than the second ice cone length threshold, the output obstacle avoidance difficulty is 100%; S3. Wirelessly transmit the obstacle avoidance difficulty percentage result to the obstacle avoidance execution module.

9. The obstacle avoidance system of a power inspection robot according to claim 8, characterized in that: The process of the obstacle avoidance execution module receiving the analysis result and executing the corresponding operation includes: When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance; when the obstacle avoidance difficulty is 50%, a command is issued to mobilize the inspection robot to start the mechanical vibration de-icing device; when the obstacle avoidance difficulty is 100%, a command is issued to dispatch the inspection robot to stop moving forward, change the inspection route, and issue an alarm to the staff for handling.

10. An obstacle avoidance method for a power inspection robot, implemented based on the obstacle avoidance system for the power inspection robot according to any one of claims 1 to 9, characterized in that: It consists of the following steps: Step 1: Use an ice detector to send electromagnetic waves to the ice layer, collect its echo time, and use a 3D scanner to collect 3D point cloud data of the ice cone; Step 2: Convert the echo time into ice thickness data by calculation, convert the 3D point cloud data of the ice cone surface into a 3D model of the ice cone using the Delaunay triangulation algorithm, and then extract the ice cone length data from the 3D model of the ice cone using the Blender modeling software; Step 3: Set the thresholds of ice thickness data and ice cone length data, and output obstacle avoidance results corresponding to different ice layer data and different ice cone data through clustering algorithm; Step 4: Use ID3 decision tree algorithm to build ice layer obstacle avoidance model and ice cone obstacle avoidance model; Step 5: Analyze the real-time ice thickness data and ice cone length data, and transmit the analysis results to the obstacle avoidance execution module; Step 6. Receive the analysis results. When the obstacle avoidance difficulty is 0%, it has no effect on the inspection robot's obstacle avoidance. When the obstacle avoidance difficulty is 50%, issue a command to mobilize the inspection robot to start the mechanical vibration de-icing device. When the obstacle avoidance difficulty is 100%, issue a command to dispatch the inspection robot to stop moving forward, change the inspection route, and send an alarm to the staff for handling.

Images