Intelligent monitoring and early warning system for fire behavior of wind power generation field
By integrating technologies such as dual-spectral fusion smoke and fire recognition, multi-point temperature measurement and early warning, and panoramic infrared thermal imaging analysis, the problem of early monitoring of fire hazards in wind power plants has been solved, achieving efficient and accurate fire identification and location, and supporting automated management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-04
- Publication Date
- 2026-03-27
AI Technical Summary
Wind farms pose numerous fire hazards that are difficult to detect in their early stages. Traditional monitoring technologies suffer from issues such as missed reports, false alarms, and incomplete coverage, making it particularly difficult to achieve comprehensive monitoring in complex environments.
The system employs a dual-spectrum fusion smoke and fire identification module, a multi-point temperature measurement/over-temperature early warning module, a panoramic infrared thermal imaging analysis module, and an intelligent interference source shielding module. Combined with DEM elevation data, it achieves intelligent fire monitoring and early warning. Through technologies such as SURF feature point extraction and affine transformation, SIFT feature matching algorithm, and Kalman filter trajectory prediction, it realizes fire source identification, location, and false alarm shielding.
It enables accurate identification and early warning of fires in wind power plants, reduces the rate of missed and false alarms, improves monitoring coverage and positioning accuracy, supports automatic patrol and unattended operation, and provides intelligent fire decision support.
Smart Images
Figure CN121747261A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to an intelligent monitoring and early warning system for fires in wind power plants, belonging to the field of power plant fire monitoring. Background Technology
[0002] A wind farm is a power plant that converts the kinetic energy of wind and solar energy into electrical energy. A wind farm within a certain area is usually managed and operated by an operating and management entity that manages and operates its wind turbine generators, supporting transmission and transformation equipment, building facilities, operation and maintenance personnel, etc.
[0003] Most wind farms are built on mountain ridges, grasslands, and deserts, with wind turbine components distributed in forest and grassland environments. During operation, overheating of wind turbine generators, inverters, and data collection lines can cause fires, leading to forest and grassland fires and causing significant losses. The following are potential hazards during the operation of various equipment in wind farms: First, the wind turbine nacelles contain important electrical equipment such as transformers, relay protection systems, and automatic devices, which contain a variety of flammable materials and have a high fire load density, posing a fire hazard. Second, due to prolonged operation of the wind turbines and weather conditions, the turbines and blade tips are prone to high temperatures, also creating a fire hazard. Third, poor contact at data collection line joints can cause arcing and overheating, creating a fire hazard. Fourth, inverters and other equipment generally operate under high voltage and high current conditions; aging components can cause performance degradation, potentially leading to short circuits and overcurrents, and excessively high temperatures can ignite surrounding forest fires. The key to preventing and fighting fires at wind farms and minimizing losses lies in early detection and early intervention.
[0004] Due to the complex environment and road conditions of power plants, fires are difficult to extinguish once they spread. To extinguish fires promptly and minimize losses caused by fires involving power generation equipment, early detection and early warning are paramount for prevention. Currently, traditional infrared monitoring of wind turbines relies on strong ventilation to lower the actual temperature of hot spots, potentially masking early fires and leading to missed or delayed alarms. Furthermore, infrared sensors need to "see" the target; the confined space and dense equipment inside the nacelle / hub make comprehensive coverage difficult; obscured areas behind or inside equipment become blind spots; and common internal features of wind turbines, such as dust, oil vapor from gearbox vents, and oil mist from hydraulic system leaks, are highly similar to smoke, easily triggering false alarms with traditional video analysis. Rotating components (such as the main shaft), swaying cables, and the movement of maintenance personnel can also be misinterpreted as flickering flames or drifting smoke.
[0005] Therefore, it is essential to strengthen the research on fire monitoring and early warning technology for wind power plants and to develop fire monitoring and early warning systems for wind power plants, which has significant social and economic benefits. Summary of the Invention
[0006] In view of the actual situation of wind power plant monitoring, such as large monitoring range, many monitoring points, strong wind turbine vibration, large amount of data transmission, no wireless signal coverage, high construction and installation difficulty, and relatively harsh environment, the following measures should be taken. The purpose of this invention is to provide an intelligent fire monitoring and early warning system that integrates intelligent video monitoring of wind power plants with fire monitoring and early warning. The system is designed to realize functions such as intelligent video monitoring of wind power plants, accurate fire identification, and precise fire location, thus meeting the actual operation and management needs of wind power plants.
[0007] This invention is achieved through the following technical solutions; A smart monitoring and early warning system for fires in wind power plants includes a dual-spectrum fusion smoke and fire identification module, a multi-point temperature measurement / over-temperature early warning module, a panoramic infrared thermal imaging analysis module, an intelligent interference source shielding module, and a fire location module.
[0008] The dual-spectrum fusion smoke and fire recognition module is used to achieve pixel-level registration of thermal imaging and visible light images through SURF feature point extraction and affine transformation, output fused images and identify temperature; it can identify and alarm fire and heat sources around the clock, and thermal imaging + visible light imaging can make targets nowhere to hide.
[0009] The thermal imaging images (resolution 640×480@250) were fused using the SURF feature point extraction and affine transformation registration algorithm. Hz ) and visible light images (resolution 1920×1080@30 fps It enables automatic tracking and linked capture of fire source targets with "same target, same center"; when the target temperature is >150℃ and the visible light flame feature matching degree is >90%, an alarm is triggered, and the recognition delay is ≤2s.
[0010] The multi-point temperature measurement / over-temperature early warning module is used to divide the monitoring area into multiple independent sub-areas with sides ≤100m, and to perform non-contact wide-area temperature monitoring and over-temperature threshold early warning for each sub-area. It can issue over-temperature warnings within the monitoring area, and can also divide the monitoring area into multiple independent temperature measurement and over-temperature warning zones. Within the dual-light fusion visual scanning range, it achieves zero-miss detection of fire and over-temperature targets at ultra-long distances; and it can identify fire and over-temperature events within a range of 0-3km. When a dangerous alarm is detected, it can be reported to relevant personnel for handling, achieving automatic patrol and unattended operation.
[0011] The monitoring area is divided into independent sub-areas of 100m×100m, with non-contact wide-area temperature measurement (accuracy ±1.5℃), zone threshold early warning, and support for 8 levels of custom over-temperature thresholds (default threshold 80℃); the over-temperature event false alarm rate is <0.1%.
[0012] The panoramic infrared thermal imaging analysis module uses the SIFT feature matching algorithm to perform distortion correction and seamless stitching on multiple frames of infrared thermal images, outputting a panoramic monitoring view fused with DEM elevation data. The panoramic image stitching misalignment rate is less than 0.1 pixels to generate a panoramic monitoring view fused with DEM elevation data (resolution ≥ 4K). It realizes the panoramic infrared thermal imaging analysis function of seamless intelligent stitching of infrared thermal images, obtains all environmental information at the front-end site, and quickly realizes comprehensive fire situation decision analysis. The panoramic image makes it easy to know the location and alarm targets, making it more convenient for administrators to operate and manage.
[0013] The intelligent interference source shielding module, based on the pre-stored coordinates of the static heat source database and the Kalman filter trajectory prediction of the dynamic target, shields heat sources with a speed >30km / h or a temperature <100℃. The static shielding of fixed equipment heat sources has a temperature of (40-100℃), and the dynamic shielding of moving targets has a speed >30km / h (reducing the vehicle false alarm rate to 2.1%). By identifying known heat sources in the monitoring area, it shields known heat sources, minimizing human false alarms to the greatest extent.
[0014] The fire location module is equipped with a DEM elevation map, allowing for fire location at a single site. When a fire occurs, the system automatically identifies and alarms, locating the fire source's geographical location on the map for convenient emergency dispatch and command. Based on the pixel coordinates in the seamless panoramic monitoring view, combined with pre-set wind farm digital elevation model (DEM) data, the spatial coordinate transformation formula is as follows. ; ; ; Output the 3D geographic coordinates of the fire, where X is the 3D geodetic longitude coordinate of the target point, X0 is the geodetic longitude coordinate of the camera origin, U is the x-coordinate of the target point in the image, U0 is the x-coordinate of the image center point, and GSD is the ground resolution. θ The pitch angle is denoted by Y; where Y is the three-dimensional geodetic latitude coordinate of the target point, Y0 is the geodetic latitude coordinate of the camera origin, V is the pixel ordinate of the target point in the image, V0 is the pixel ordinate of the image center point, and Z is the elevation of the target point; three-dimensional positioning is achieved by combining DEM elevation data (1m resolution).
[0015] The long-distance fire identification module of the present invention uses an infrared thermal imager and a 300mm telephoto lens to achieve a thermal scanning frame rate of ≥15fps within a range of 0-3km.
[0016] The beneficial effects of this invention are: (1) By collecting fire information, a dual-spectrum high-definition video camera is used. The dual-spectrum high-definition camera (operating distance 2-3km) can cover an area with a single device radius of 3km, reducing the deployment density by 80%. By fusing thermal imaging and visible light features, the false alarm rate is <0.1% under a fire source scale of 0.5m², which improves the early identification rate by 40% compared with the traditional solution.
[0017] (2) Using a trajectory prediction model based on Kalman filter, the false alarm rate for vehicles is >92% (targets with speed >30km / h), and the accuracy of identifying intruders is 99.2% (temperature 36-42℃ + speed <10km / h). (3) Through intelligent smoke and fire recognition technology, the fire location module combines DEM elevation (accuracy 1m) with spatial coordinate transformation to trigger audible and visual alarms in real time, and automatically controls the pan-tilt-zoom camera to accurately track and locate suspected fire sources, greatly improving the confirmation efficiency of on-duty personnel. After the fire is confirmed, the system immediately transmits multi-dimensional data back to the command center, integrating high-precision geographic positioning, dynamic fire analysis, and meteorological situation assessment functions to achieve closed-loop management of "monitoring-early warning-decision". The innovative integration of path planning algorithms and real-time meteorological data interfaces provides a new paradigm of intelligent decision support for the field of fire prevention. Attached Figure Description
[0018] Figure 1 This is an example of the dual-spectrum fusion smoke recognition and monitoring function of the present invention.
[0019] Figure 2 This is a schematic diagram of the configuration of the multi-point temperature measurement / over-temperature early warning module of the present invention.
[0020] Figure 3 This is a schematic diagram of the monitoring screen for the multi-point temperature measurement / over-temperature early warning module of the present invention.
[0021] Figure 4 This is a schematic diagram of the device management of the multi-point temperature measurement / over-temperature early warning module of the present invention.
[0022] Figure 5 This is a schematic diagram of historical data curves for the multi-point temperature measurement / over-temperature early warning module of the present invention.
[0023] Figure 6 This is a schematic diagram of the multi-point temperature measurement / over-temperature early warning module device monitoring of the present invention.
[0024] Figure 7 This is a schematic diagram of the alarm and warning function of the multi-point temperature measurement / over-temperature warning module of the present invention.
[0025] Figure 8 This is an example of the multi-point temperature measurement / over-temperature warning function of the present invention, wherein Figure (a) shows multi-point temperature measurement and Figure (b) shows the over-temperature warning function.
[0026] Figure 9 This is an example diagram of the installation of the box-type variable thermal imaging temperature measurement module for the panoramic infrared thermal imaging analysis module of the present invention.
[0027] Figure 10 This is an example diagram of the installation of the panoramic infrared thermal imaging analysis module cable branch box for thermal imaging temperature measurement according to the present invention.
[0028] Figure 11 This is an example diagram of the installation of the panoramic infrared thermal imaging analysis module cable docking box for thermal imaging temperature measurement according to the present invention.
[0029] Figure 12 This is an example of the long-distance fire detection and temperature measurement function of the present invention.
[0030] Figure 13 This is an example of the panoramic image stitching function of the present invention.
[0031] Figure 14 This is an example of the intelligent interference source shielding function of the present invention, wherein part of Figure (a) is intelligent shielding of a known fire source; part of Figure (b) is intelligent shielding of a moving fire source.
[0032] Figure 15 This is an example of the fire location function of the present invention. Detailed Implementation
[0033] The present invention will be further described in detail below with reference to specific embodiments, but the scope of protection of the present invention is not limited thereto. Example 1
[0034] A smart monitoring and early warning system for fires in wind power plants includes a dual-spectrum fusion smoke and fire identification module, a multi-point temperature measurement / over-temperature early warning module, a panoramic infrared thermal imaging analysis module, an intelligent interference source shielding module, and a fire location module.
[0035] The dual-spectrum fusion smoke and fire recognition module is used to achieve pixel-level registration of thermal imaging and visible light images through SURF feature point extraction and affine transformation, outputting a fused image and identifying the temperature. It provides all-weather identification and alarm for fire and heat sources; the combination of thermal imaging and visible light imaging makes targets impossible to hide.
[0036] The thermal imaging images (resolution 640×480@250) were fused using the SURF feature point extraction and affine transformation registration algorithm. Hz ) and visible light images (resolution 1920×1080@30 fpsThis system enables automatic tracking and synchronized image capture of fire sources, achieving "same target, same center" alignment. An alarm is triggered when the target temperature exceeds 150℃ and the visible light flame feature matching degree exceeds 90%, with a recognition delay of ≤2 seconds. The dual-spectrum fusion smoke and fire recognition module (S1) simultaneously acquires images through a thermal imaging camera (FLIR A700) and a visible light camera (HikvisionDS-2DF8236I). After extracting feature points using the SURF algorithm, pixel-level registration is achieved through affine transformation. Tracking is triggered when the target temperature exceeds 150℃ and the visible light flame feature matching degree exceeds 90%. Figure 1 The image shows an example of dual-spectrum fusion smoke recognition and monitoring.
[0037] The multi-point temperature measurement / over-temperature early warning module has the ability to collect, monitor and analyze the temperature and humidity of electrical equipment in the monitored area online, and supports functions such as system configuration, large screen monitoring, equipment management, historical data curve query, real-time equipment monitoring and alarm early warning (see schematic diagrams for details). Figures 2 to 7 ).
[0038] This module also supports dividing the monitoring area into multiple independent sub-areas (side length ≤ 100m) and performing non-contact wide-area temperature monitoring and over-temperature threshold early warning for each sub-area. The system defaults to dividing the monitoring area into 100m×100m sub-areas to achieve non-contact wide-area temperature measurement with a measurement accuracy of ±1.5℃. It supports 8 levels of custom over-temperature thresholds (default threshold 80℃) and the over-temperature event false alarm rate is less than 0.1%.
[0039] Within the dual-light fusion visual scanning range, the system can achieve ultra-long-distance identification of fires and overheated targets, covering a range of 0–3 km, with zero missed detections. Once a dangerous alarm is detected, the system can automatically push information to relevant personnel, enabling automatic patrols and unmanned operation. This module is suitable for large-scale, wide-angle wind farm monitoring scenarios, effectively capturing early warning targets, detecting fire or heat sources, and issuing warnings. Figure 8 This is an example of multi-point temperature measurement / over-temperature warning.
[0040] The panoramic infrared thermal imaging analysis module is widely used in wind power generation for online temperature monitoring of many key equipment, including: various components of wind turbines, high-voltage cable adapters and cables, low-voltage switchgear contacts and cable joints, capacitors, circuit breakers and disconnect switches, motor outlet box cable joints, as well as cable surfaces, cable joints and cable interlayers in cable tunnels, and electrical cable heads in substations or important outdoor circuits.
[0041] In prefabricated substations, infrared thermal imaging sensors are deployed on the top of the enclosure to monitor cable head temperature changes in real time. Wireless mesh nodes are located outside the substation, forming a self-organizing network with the wireless mesh gateway at the fan location, enabling long-distance transmission of monitoring data. A schematic diagram of the prefabricated substation thermal imaging temperature measurement installation is shown below. Figure 9 As shown in the diagram. For cable branch boxes, infrared thermal imaging sensors are installed on the box wall to monitor the cable head temperature in real time; wireless mesh nodes are deployed next to the branch boxes, also connected to the wind turbine-side gateway via a self-organizing network to complete remote data transmission. The installation diagram of the thermal imaging temperature measurement in the cable branch box is shown below. Figure 10 As shown in the diagram. In the cable junction box, an infrared thermal imaging sensor is positioned at the bottom of the box to monitor the cable head temperature in real time; a wireless Mesh node is located next to the junction box, forming a self-organizing network with the wind turbine-side gateway to ensure stable remote transmission of monitoring data. The installation diagram of the thermal imaging temperature measurement in the cable junction box is shown below. Figure 11 As shown.
[0042] This module uses the SIFT feature matching algorithm to perform distortion correction and seamless stitching on multiple frames of infrared thermal images, outputting a panoramic monitoring view that integrates DEM elevation data. The panoramic image stitching misalignment rate is less than 0.1 pixels, generating a panoramic monitoring view that integrates DEM elevation data (resolution ≥ 4K). Its wide coverage area and accurate temperature measurement feature enable the system to be widely used in vast open areas such as wind farms. Figure 12 The image shows an example of long-distance fire detection and temperature measurement. The panoramic infrared thermal imaging analysis module enables seamless intelligent stitching of infrared thermal images for panoramic infrared thermal imaging analysis. It obtains all environmental information from the front-end site, quickly achieving comprehensive fire situation decision analysis. The panoramic image provides a clear and easy-to-understand view of the location and alarm targets, making it more convenient for administrators to operate and manage. Figure 13 The image shown is an example of panoramic image stitching; it realizes the panoramic infrared thermal image analysis function of seamless intelligent stitching of infrared thermal images, obtains all environmental information at the front end, and quickly realizes comprehensive fire situation decision analysis. The panoramic image makes it easy to know the location and alarm target, making it more convenient for administrators to operate and manage.
[0043] The intelligent interference source shielding module, based on the pre-stored coordinates of the static heat source database and the Kalman filter trajectory prediction of the dynamic target, shields heat sources with a speed >30km / h or a temperature <100℃. The static shielding of fixed equipment heat sources has a temperature of (40-100℃), and the dynamic shielding of moving targets has a speed >30km / h (reducing the vehicle false alarm rate to 2.1%). By identifying known heat sources in the monitoring area, it shields known heat sources, minimizing human false alarms to the greatest extent.
[0044] Through its intelligent shielding function, it can distinguish the control functions of multiple different areas, and divide the control areas of fire sources, personnel and vehicles within the wind power plant to avoid false alarms of heat sources such as high-speed vehicles. Figure 14 The image shows an example of the intelligent interference source shielding function.
[0045] The fire location module is equipped with a DEM elevation map, allowing for fire location at a single site. When a fire occurs, the system automatically identifies and alarms, locating the fire source's geographical location on the map for convenient emergency dispatch and command. Based on the pixel coordinates in the seamless panoramic monitoring view, combined with pre-set wind farm digital elevation model (DEM) data, the spatial coordinate transformation formula is as follows. ; ; ; Output the 3D geographic coordinates of the fire, where X is the 3D geodetic longitude coordinate of the target point, X0 is the geodetic longitude coordinate of the camera origin, U is the x-coordinate of the target point in the image, U0 is the x-coordinate of the image center point, and GSD is the ground resolution. θ The pitch angle is denoted by Y; where Y is the three-dimensional geodetic latitude coordinate of the target point, Y0 is the geodetic latitude coordinate of the camera origin, V is the pixel ordinate of the target point in the image, V0 is the pixel ordinate of the image center point, and Z is the elevation of the target point; three-dimensional positioning is achieved by combining DEM elevation data (1m resolution).
[0046] The fire location module is equipped with a DEM elevation map, allowing for fire location at a single station. When a fire occurs, the system automatically identifies and alarms, locating the fire source on the map for convenient emergency dispatch and command. Figure 15 The image shows an example of the fire location function.
[0047] The long-distance fire detection module described in this embodiment uses an infrared thermal imager and a 300mm telephoto lens to achieve a thermal scanning frame rate of ≥15fps within a range of 0-3km.
Claims
1. A smart monitoring and early warning system for fires in wind power plants, characterized in that: The system includes a dual-spectrum fusion smoke and fire recognition module, a multi-point temperature measurement / over-temperature early warning module, a panoramic infrared thermal imaging analysis module, an intelligent interference source shielding module, a fire location module, and a long-distance fire recognition module.
2. The intelligent monitoring and early warning system for fires in wind power farms according to claim 1, characterized in that, In the dual-spectrum fusion smoke recognition module, pixel-level registration of thermal imaging and visible light images is achieved through SURF feature point extraction and affine transformation, outputting a fused image and identifying the temperature.
3. The intelligent monitoring and early warning system for fires in wind power plants according to claim 1, characterized in that, The multi-point temperature measurement / over-temperature early warning module is used to divide the monitoring area into multiple independent sub-areas with a side length ≤100m, and to perform non-contact wide-area temperature monitoring and over-temperature threshold early warning for each sub-area. It has the ability to collect, monitor and analyze the temperature and humidity of electrical equipment in the monitoring area online, and supports system configuration, large screen monitoring, equipment management, historical data curve query, real-time equipment monitoring and alarm early warning functions.
4. The intelligent monitoring and early warning system for fires in wind power farms according to claim 1, characterized in that, The panoramic infrared thermal imaging analysis module uses the SIFT feature matching algorithm to perform distortion correction and seamless stitching on multiple frames of infrared thermal images, outputting a panoramic monitoring view that integrates DEM elevation data. This enables online monitoring of key equipment such as wind turbine components, high-voltage cable adapters and cables, low-voltage switchgear contacts and cable joints, capacitors, circuit breakers and disconnect switches, motor outlet boxes and cable joints, as well as cable surfaces, cable joints and cable interlayers in cable tunnels, and electrical cable heads in substations or important outdoor circuits.
5. The intelligent monitoring and early warning system for fires in wind power farms according to claim 1, characterized in that, In the intelligent interference source shielding module, based on the pre-stored coordinates of the static heat source database and the Kalman filter trajectory prediction of the dynamic target, heat sources with a speed > 30 km / h or a temperature < 100℃ are shielded.
6. The intelligent monitoring and early warning system for fires in wind power farms according to claim 1, characterized in that, In the fire location module, the pixel coordinates in the seamless panoramic monitoring view are combined with the preset wind farm digital elevation model (DEM) data, and the three-dimensional geographic coordinates of the fire are output through spatial coordinate transformation.
7. The intelligent monitoring and early warning system for fires in wind power farms according to claim 6, characterized in that, The calculation expression for outputting the three-dimensional geographic coordinates of the fire situation through spatial coordinate transformation is as follows: ； Where X represents the three-dimensional geographic coordinates of the fire. U represents the geodetic longitude coordinates of the camera origin, and U represents the x-coordinate of the target point in the image (in pixels). The pixel x-coordinate of the image center point, where GSD is the ground resolution. θ It is the pitch angle.
8. The intelligent monitoring and early warning system for fires in wind power plants according to claim 1, characterized in that, The system includes a dual-spectrum fusion smoke and fire recognition module, a multi-point temperature measurement / overheating early warning module, a panoramic infrared thermal imaging analysis module, an intelligent interference source shielding module, a fire location module, and a long-distance fire recognition module. The dual-spectrum fusion smoke and fire recognition module and the multi-point temperature measurement / overheating early warning module work collaboratively, specifically including: The dual-spectrum fusion smoke and fire recognition module achieves pixel-level registration of the 640×480@25Hz thermal imaging image and the 1920×1080@30fps visible light image through SURF feature point extraction and affine transformation, outputting a fused image and identifying the initial fire source temperature. The multi-point temperature measurement / over-temperature warning module receives the fused image data and divides the monitoring range of 0-3km into independent sub-regions with a side length of ≤100m. It performs non-contact wide-area temperature scanning with an accuracy of ±1.5℃ for each sub-region. When any sub-region detects a temperature >150℃ and a visible light flame feature matching degree >90%, it triggers a linkage alarm with a delay of ≤2s and simultaneously starts the fire location process of the panoramic infrared thermal imaging analysis module.
9. The intelligent monitoring and early warning system for fires in wind power plants according to claim 1, characterized in that, The long-range fire detection module uses an infrared thermal imager and a 300mm telephoto lens to achieve a thermal scanning frame rate of ≥15fps within a range of 0-3km.
10. The intelligent monitoring and early warning system for fires in wind power farms according to claim 1, characterized in that, In the panoramic infrared thermal imaging analysis module, an infrared thermal imaging sensor is arranged at the bottom of the cable docking box to monitor the temperature of the cable head in real time; a wireless Mesh node is set up next to the docking box to form a self-organizing network with the wind turbine-side gateway to ensure stable remote transmission of monitoring data.