Wind power plant current collection line tower inspection method based on multi-sensor data fusion

Abstract

The invention provides a multi-sensor data fusion wind power plant current collection line tower inspection method, which comprises the following steps: when an inclination monitoring terminal detects that a tower is abnormally inclined or exceeds a preset safety threshold, a system immediately triggers an early warning mechanism, immediately analyzes and locates the tower number and geographic position of the abnormal tower, and automatically generates an unmanned aerial vehicle inspection task; the unmanned aerial vehicle flies to the position of the tower according to the flight path, hovers at a preset detection position after reaching a target point, verifies the inclination degree and potential risks of the tower in a photographing and real-time video mode, and timely formulates and takes countermeasures. According to the invention, all-weather and omnibearing inspection (supporting night infrared inspection) can be carried out on all current collection lines and towers in a wind power plant area, a traditional manual inspection mode is replaced, the inspection operation is completed with high efficiency and high precision, the state of the towers is stably and efficiently monitored, the safety of the current collection lines is ensured, and the operation and maintenance intelligence level of the wind power plant is greatly enhanced.

Description

Technical Field

[0001] The invention provides a method for inspecting wind farm collector line towers by using a drone, and relates to the field of wind farm operation and maintenance. Background Art

[0002] The collector line of a wind farm is an important part that connects wind turbines and substations. Its main function is to collect the electricity generated by wind turbines and then transmit it to the substation for voltage boosting. The collector line tower of a wind farm is one of the main media for wind turbines to transmit electricity. In order to ensure the safe and stable operation of the equipment, and to maximize the utilization efficiency of the equipment, reduce the equipment failure rate, and extend the service life of the equipment, it is necessary to inspect the collector line of the wind farm during the operation of the wind farm. The collector line tower of a wind farm is usually located in a large and complex geographical environment. At present, most of the inspections are still carried out by manpower. Due to the harsh operating environment and long-term exposure to high temperature, high cold, wind and sand, humidity, ice and other environmental factors, it is a systematic and complex project. The traditional manual inspection method has problems such as high cost, high risk and low efficiency.

[0003] With the development of drone technology, its application in the field of power inspection has gradually been promoted. The existing drone inspection methods are mainly used for the inspection of conventional power lines, but the inspection technology for wind farm collector line towers is still in its infancy. Wind farm tower inspections have unique challenges, such as large inspection areas, harsh climate environments, and complex tower structures. Therefore, a new drone inspection method is urgently needed, which can improve inspection efficiency, reduce costs, and improve inspection accuracy through drone inspection data processing and analysis based on deep learning. Summary of the invention

[0004] The present invention provides a wind farm collector line tower inspection method with multi-sensor data fusion. The method adopts an inspection mode combining tilt sensors deployed on towers with drone intelligent inspection (image sensors collect images and videos). The method has the capabilities of unified scheduling, data management, data analysis and real-time early warning, and realizes all-weather and all-round inspection of all collector lines and towers in the wind farm area (supporting infrared inspection at night), replacing the traditional manual inspection mode, completing the inspection work with high efficiency and high precision, stably and efficiently monitoring the tower status, ensuring the safety of collector lines, and greatly enhancing the intelligent level of operation and maintenance of wind farms.

[0005] The technical solution adopted to achieve the above-mentioned purpose of the present invention is:

[0006] A wind farm collector line tower inspection method with multi-sensor data fusion includes the following steps: (1) installing a tilt sensor terminal on a tower of a wind farm collector line, numbering the tilt sensor terminal, binding the number of the tilt sensor terminal with the tower number of the tower, and storing the resultant numbers in a data storage unit of a control center, wherein the data storage unit stores geographic location data and airspace data of each tower;

[0007] (2) The tilt sensor terminal obtains the tilt data of the tower and wirelessly transmits the collected tilt data of the tower to the control center;

[0008] (3) The control center calculates and analyzes the received tower tilt data to determine the tilt status of the tower. When it is calculated that the tower is abnormally tilted or the tilt exceeds the preset safety threshold, the early warning mechanism is triggered and the drone inspection task is automatically generated;

[0009] (4) The control center retrieves the geographic location data and airspace data of the corresponding target tower according to the number of the tilt sensor terminal corresponding to the trigger warning mechanism, and generates the UAV flight path and initial hovering detection position information in combination with the geographic location of the UAV smart airport;

[0010] (5) The drone is equipped with an image sensor, a laser, and a meteorological sensor. It flies to the target tower according to the flight path and hovers at the preset initial hovering detection position;

[0011] (6) The drone collects and obtains the status image and video data of the target tower, and uses the laser to emit a laser beam to quickly scan the target tower to obtain the three-dimensional point cloud data of its surface; the image and video data and the three-dimensional point cloud data are transmitted to the control center, and the drone returns after completing the collection task;

[0012] (7) The control center receives the status image and video data of the target tower and the 3D point cloud data, and further verifies the tilt degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner;

[0013] (8) The control center processes and analyzes the three-dimensional point cloud data through a model matching algorithm and a semi-automatic editing method, quickly constructs a three-dimensional model and coordinate data of the target tower, and imports the three-dimensional model and coordinate data of the target tower into a data storage unit; and optimizes and stores the UAV flight path and hovering detection position information corresponding to the target tower in combination with the three-dimensional model and coordinate data of the target tower;

[0014] (9) In the subsequent inspection process, if the target tower that triggers the early warning mechanism is a tower that has not been inspected before, repeat steps (4)-(8) to perform the inspection; if the target tower that triggers the early warning mechanism is a tower that has been inspected before, perform the inspection according to the following steps: the control center retrieves the optimized UAV flight path and hovering detection position information and controls the UAV to fly to the target tower according to the optimized flight path and hover at the optimized hovering detection position; the UAV collects and obtains the status image and video data of the target tower, transmits the image and video data to the control center, and the UAV returns after completing the collection task; the control center receives the status image and video data of the target tower, and further verifies the tilt degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner.

[0015] Furthermore, the tilt sensing terminal in step (1) includes a tilt sensor, an adjustment bracket, a battery and a wireless communication module, wherein the tilt sensor and the wireless communication module are installed at the upper position of the tower through the adjustment bracket, and the battery is installed at the lower position of the tower, and the battery is connected to the tilt sensor and the wireless communication module and supplies power.

[0016] Furthermore, the wireless communication module adopts a 4G communication module to transmit the tower tilt data collected by the tilt sensor terminal to the control center through the cloud server.

[0017] Furthermore, the tilt sensor selects a high-precision tilt sensor with high resolution, low drift and strong anti-interference ability, improves the accuracy and stability of the tilt sensor through temperature compensation, noise filtering and calibration algorithms, uses an adaptive filter to filter out noise, and ensures the accuracy of tilt angle measurement through multi-point calibration, and uses a high-frequency sampling rate to capture and record the data of the tilt sensor to ensure the real-time and continuity of the tower tilt data.

[0018] Furthermore, in step (3), the inclination state of the tower is determined by using a Kalman filter and a fast Fourier transform (FFT) to smooth the tower inclination data and eliminate noise, and then using a trend analysis and anomaly detection algorithm.

[0019] Furthermore, the image sensor in step (5) uses a high-resolution optical camera and a thermal imager to obtain clear images and videos under different lighting conditions to capture the details of the target tower.

[0020] Furthermore, in steps (7) and (9), image enhancement, defect detection and classification image processing algorithms are used to process the real-time image, and a neural network model is trained to automatically identify defects or abnormalities on the target tower in the image.

[0021] Furthermore, a deep learning-based pole tower defect detection image system is set up inside the control center. The pole tower defect detection image system first preprocesses the image data set containing pole tower defects to enhance the generalization ability of the model. The data set contains two parts: positive samples and negative samples; then the deep learning model YOLO is combined with the adversarial neural network to train it so that it can automatically identify and analyze the defect features in the pole tower images taken by the drone.

[0022] Furthermore, the drone intelligent airport adopts a distributed fixed airport deployment mode.

[0023] Furthermore, the UAV adopts a method combining a real-time dynamic differential global positioning system and an inertial navigation system to achieve centimeter-level positioning accuracy of the UAV;

[0024] During the flight of the drone, real-time position feedback and path tracking algorithms are combined to automatically generate the optimal path according to mission requirements and environmental conditions to ensure that the drone can fly stably along the predetermined route;

[0025] During the flight, the drone combines lidar, ultrasonic sensors and image sensors to fuse data from different sensors to detect and avoid obstacles on the flight path in real time;

[0026] During the flight of the UAV, PID or MPC control algorithm is used to achieve precise attitude control and stable flight of the UAV.

[0027] Compared with the prior art, the wind farm collector line tower inspection method with multi-sensor data fusion provided by the present invention has the following advantages: 1. The tilt state of the tower is detected by the tilt sensor to achieve safe, stable, high-speed data transmission and real-time monitoring; when the horizontal tilt angle of the wind farm collector line tower exceeds the set angle threshold or the sudden acceleration exceeds the set acceleration threshold, an early warning is immediately issued. Considering the balance between transmission distance and power consumption, the tilt data can be stably and timely transmitted to the control center through the wireless communication module and a reliable data transmission protocol. Monitoring by the tilt sensor can preliminarily establish the target tower in the drone inspection in the subsequent steps, thereby reducing the workload of the drone inspection.

[0028] 2. After the target pole tower with abnormal tilt is identified, the drone is used for automated inspection. The drone inspection can cover a larger area and reduce high-risk operations in manual inspections, thereby improving the safety of inspections and work efficiency. The line laser scanning equipment carried by the drone scans the pole tower point by point and line by line to obtain precise pole tower structure information and generate a fine three-dimensional model, thereby further optimizing the later drone inspection path and hovering point coordinates, further improving the accuracy of the inspection. At the same time, the line laser scanning equipment can provide high-precision three-dimensional point cloud data, capture the slight deformation and structural anomalies of the pole tower, and significantly improve the inspection accuracy.

[0029] In summary, the solution provided by the present invention uses online tilt sensors deployed on the pole towers to monitor the tilt status of the pole towers in real time, and combines drone intelligent inspection to achieve regular inspection and online tilt warning monitoring. It can accurately locate the fault point, and the pole tower tilt monitoring terminal is combined with drone inspection to quickly check and review the fault status. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a general architecture diagram of the wind farm collector line tower inspection system provided in this embodiment;

[0031] Figure 2 This is a system architecture diagram of the tilt sensing terminal;

[0032] Figure 3 is a signal transmission path diagram;

[0033] Figure 4 Decomposition diagram of GAN adversarial network;

[0034] Figure 5 A flowchart for constructing a 3D model of a tower based on 3D point cloud data;

[0035] Figure 6 Provide a 3D model of the target tower and a schematic diagram of the drone waypoint optimization;

[0036] Figure 7 This is a module schematic diagram of the wind farm collector line tower inspection system provided in this embodiment. DETAILED DESCRIPTION

[0037] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited to the following embodiments.

[0038] The wind farm collector line tower inspection method using multi-sensor data fusion provided by the present invention comprises the following steps:

[0039] (1) Installing the tilt sensor terminal on the tower of the wind farm collection line, numbering the tilt sensor terminal, binding the number of the tilt sensor terminal with the tower number of the tower, and storing it in a data storage unit of the control center, wherein the data storage unit stores the geographical location data and airspace data of each tower; Figure 1 The following is a general architecture diagram of an intelligent inspection system established based on the inspection method provided by the present invention. Figure 1 The data layer in is the data storage unit, which stores tower number information, mission information, airspace information data, geographic information, data result information, and point cloud modeling information.

[0040] The tilt sensing terminal includes a tilt sensor, an adjustment bracket, a battery and a wireless communication module, wherein the tilt sensor and the wireless communication module are installed at the upper position of the tower through the adjustment bracket, and the battery is installed at the lower position of the tower. The battery is connected to the tilt sensor and the wireless communication module and supplies power. Figure 2 This is the system architecture diagram of the tilt sensing terminal, such as Figure 2 As shown, the wireless communication module adopts a 4G communication module, and transmits the tower tilt data collected by the tilt sensor terminal to the control center through the cloud server.

[0041] The tilt sensor selects a high-precision tilt sensor with high resolution, low drift and strong anti-interference ability, improves the accuracy and stability of the tilt sensor through temperature compensation, noise filtering and calibration algorithms, uses an adaptive filter to filter out noise, and ensures the accuracy of tilt angle measurement through multi-point calibration. A high-frequency sampling rate is used to capture and record the data of the tilt sensor to ensure the real-time and continuity of the tower tilt data.

[0042] (2) The tilt sensor terminal obtains the tilt data of the tower and transmits the collected tilt data of the tower wirelessly to the control center; the signal transmission path is as follows: Figure 3 As shown, the tilt data is finally transmitted to the control center.

[0043] (3) The control center calculates and analyzes the received tower tilt data to determine the tilt state of the tower. When it is calculated that the tower is abnormally tilted or the tilt degree exceeds the preset safety threshold, the early warning mechanism is triggered and the drone inspection task is automatically generated. In step (3), the tower tilt data is smoothed and noise is eliminated using Kalman filtering and fast Fourier transform (FFT), and then the tilt state of the tower is determined through trend analysis and anomaly detection algorithms.

[0044] (4) The control center retrieves the geographic location data and airspace data of the corresponding target tower according to the number of the tilt sensor terminal corresponding to the triggering warning mechanism, and generates the UAV flight path and initial hovering detection position information in combination with the geographic location of the UAV smart airport; the UAV smart airport adopts the deployment mode of a distributed fixed airport.

[0045] (5) The UAV is equipped with an image sensor, a laser and a meteorological sensor, flies to the target pole tower according to the flight path, and hovers at a preset initial hovering detection position; the image sensor in step (5) uses a high-resolution optical camera and a thermal imager to obtain clear images and videos under different lighting conditions to capture the details of the target pole tower.

[0046] (6) The drone collects and obtains the state image and video data of the target tower, and uses a laser to emit a laser beam to quickly scan the target tower to obtain the three-dimensional point cloud data of its surface; the image and video data and the three-dimensional point cloud data are transmitted to the control center, and the drone returns after completing the collection task; the drone adopts a method combining a real-time dynamic differential global positioning system and an inertial navigation system to achieve centimeter-level positioning accuracy of the drone;

[0047] During the flight of the drone, real-time position feedback and path tracking algorithms are combined to automatically generate the optimal path according to mission requirements and environmental conditions to ensure that the drone can fly stably along the predetermined route;

[0048] During the flight, the drone combines lidar, ultrasonic sensors and image sensors to fuse data from different sensors to detect and avoid obstacles on the flight path in real time;

[0049] During the flight of the UAV, PID or MPC control algorithms are used to achieve precise attitude control and stable flight of the UAV. (7) The control center receives the status image and video data of the target tower and the three-dimensional point cloud data, and further verifies the inclination degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner; uses image enhancement, defect detection and classification image processing algorithms to process real-time images, and trains neural network models to automatically identify defects or abnormalities on the target tower in the image. A tower defect detection image system based on deep learning is set up inside the control center. The tower defect detection image system first pre-processes the image data set containing tower defects to enhance the generalization ability of the model. The data set contains positive samples and negative samples; then uses the deep learning model yolo combined with the adversarial neural network to train it so that it can automatically identify and analyze the defect features in the tower images taken by the UAV. The process of the GAN adversarial network decomposition diagram is as follows: Figure 4As shown, the above-mentioned machine learning algorithm module is an existing technical solution in the field of artificial intelligence and will not be elaborated in detail in this patent.

[0050] (8) The control center processes and analyzes the 3D point cloud data through model matching algorithms and semi-automatic editing methods to quickly construct the 3D model and coordinate data of the target tower. Figure 5 The flowchart of building the 3D model of the pole tower in this step is that acquiring 3D point cloud data based on laser scanning and building a 3D model based on the 3D point cloud data is a common operation in the field of image processing and will not be elaborated in detail in this patent. The 3D model and coordinate data of the target pole tower are imported into the data storage unit; and the flight path and hovering detection position information of the drone corresponding to the target pole tower are optimized and stored in combination with the 3D model and coordinate data of the target pole tower; refer to Figure 6 , Figure 6 The 3D model of the target tower and the schematic diagram of the drone waypoint optimization.

[0051] (9) In the subsequent inspection process, if the target tower that triggers the early warning mechanism is a tower that has not been inspected before, repeat steps (4)-(8) to perform the inspection; if the target tower that triggers the early warning mechanism is a tower that has been inspected before, perform the inspection according to the following steps: the control center retrieves the optimized UAV flight path and hovering detection position information and controls the UAV to fly to the target tower according to the optimized flight path and hover at the optimized hovering detection position; the UAV collects and obtains the status image and video data of the target tower, transmits the image and video data to the control center, and the UAV returns after completing the collection task; the control center receives the status image and video data of the target tower, and further verifies the tilt degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner.

[0052] Based on the above inspection method, this embodiment also provides a wind farm collector line tower inspection system. The overall logical architecture of the system is shown in FIG. Figure 1 As shown, from top to bottom, it includes the display layer, business layer, technical layer, data layer and perception layer. Figure 7 This is a module diagram of the wind farm collector line tower inspection system.

Claims

1. A wind farm collector line tower inspection method based on multi-sensor data fusion, characterized in that The following steps are involved: (1) Installing the tilt sensor terminal on the tower of the wind farm collection line, numbering the tilt sensor terminal, binding the number of the tilt sensor terminal with the tower number of the tower, and storing it in a data storage unit of the control center, wherein the data storage unit stores the geographical location data and airspace data of each tower; (2) The tilt sensor terminal obtains the tilt data of the tower and wirelessly transmits the collected tilt data of the tower to the control center; (3) The control center calculates and analyzes the received tower tilt data to determine the tilt status of the tower. When it is calculated that the tower is abnormally tilted or the tilt exceeds the preset safety threshold, the early warning mechanism is triggered and the drone inspection task is automatically generated; (4) The control center retrieves the geographic location data and airspace data of the corresponding target tower according to the number of the tilt sensor terminal corresponding to the trigger warning mechanism, and generates the UAV flight path and initial hovering detection position information in combination with the geographic location of the UAV smart airport; (5) The drone is equipped with an image sensor, a laser, and a meteorological sensor. It flies to the target tower according to the flight path and hovers at the preset initial hovering detection position; (6) The drone collects and obtains the status image and video data of the target tower, and uses the laser to emit a laser beam to quickly scan the target tower to obtain the three-dimensional point cloud data of its surface; the image and video data and the three-dimensional point cloud data are transmitted to the control center, and the drone returns after completing the collection task; (7) The control center receives the status image and video data of the target tower and the 3D point cloud data, and further verifies the tilt degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner; (8) The control center processes and analyzes the three-dimensional point cloud data through a model matching algorithm and a semi-automatic editing method, quickly constructs a three-dimensional model and coordinate data of the target tower, and imports the three-dimensional model and coordinate data of the target tower into a data storage unit; and optimizes and stores the UAV flight path and hovering detection position information corresponding to the target tower in combination with the three-dimensional model and coordinate data of the target tower; (9) During the subsequent inspection process, if the target tower that triggers the early warning mechanism is a tower that has not been inspected before, repeat steps (4) to (8) to perform the inspection; If the target tower that triggers the early warning mechanism is a tower that has been inspected before, the inspection will be carried out according to the following steps: the control center retrieves the optimized UAV flight path and hovering detection position information and controls the UAV to fly to the target tower according to the optimized flight path and hover at the optimized hovering detection position; the UAV collects and obtains the status image and video data of the target tower, transmits the image and video data to the control center, and the UAV returns after completing the collection task; the control center receives the status image and video data of the target tower, and further verifies the inclination degree and potential risks of the target tower based on the status image and video data, so as to formulate and take countermeasures in a timely manner.

2. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1 is characterized by: The tilt sensing terminal in step (1) includes a tilt sensor, an adjustment bracket, a battery and a wireless communication module, wherein the tilt sensor and the wireless communication module are installed at the upper position of the tower through the adjustment bracket, and the battery is installed at the lower position of the tower, and the battery is connected to the tilt sensor and the wireless communication module and supplies power.

3. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 2 is characterized in that: The wireless communication module adopts a 4G communication module, and transmits the tower tilt data collected by the tilt sensor terminal to the control center through the cloud server.

4. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 2 is characterized by: The tilt sensor selects a high-precision tilt sensor with high resolution, low drift and strong anti-interference ability, improves the accuracy and stability of the tilt sensor through temperature compensation, noise filtering and calibration algorithms, uses an adaptive filter to filter out noise, and ensures the accuracy of tilt angle measurement through multi-point calibration. A high-frequency sampling rate is used to capture and record the data of the tilt sensor to ensure the real-time and continuity of the tower tilt data.

5. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1 is characterized by: In step (3), the inclination state of the tower is determined by using Kalman filtering and fast Fourier transform (FFT) to smooth the data and eliminate noise, and then using trend analysis and anomaly detection algorithms.

6. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1, characterized in that: The image sensor described in step (5) uses a high-resolution optical camera and a thermal imager to obtain clear images and videos under different lighting conditions to capture the details of the target tower.

7. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1 is characterized by: In steps (7) and (9), image enhancement, defect detection and classification image processing algorithms are used to process real-time images, and a neural network model is trained to automatically identify defects or abnormalities on the target tower in the image.

8. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 7 is characterized by: A deep learning-based pole tower defect detection image system is set up inside the control center. The pole tower defect detection image system first preprocesses the image data set containing pole tower defects to enhance the generalization ability of the model. The data set contains two parts: positive samples and negative samples; then the deep learning model YOLO is combined with the adversarial neural network to train it so that it can automatically identify and analyze the defect features in the pole tower images taken by drones.

9. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1, characterized in that: The UAV intelligent airport adopts the deployment mode of distributed fixed airports.

10. The wind farm collector line tower inspection method based on multi-sensor data fusion according to claim 1, characterized in that: The UAV adopts a method combining a real-time dynamic differential global positioning system and an inertial navigation system to achieve centimeter-level positioning accuracy of the UAV; During the flight of the drone, real-time position feedback and path tracking algorithms are combined to automatically generate the optimal path according to mission requirements and environmental conditions to ensure that the drone can fly stably along the predetermined route; During the flight, the drone combines lidar, ultrasonic sensors and image sensors to fuse data from different sensors to detect and avoid obstacles on the flight path in real time; During the flight of the UAV, PID or MPC control algorithm is used to achieve precise attitude control and stable flight of the UAV.

Images