Multi-information fusion wind power station intelligent fire-fighting fire verification and positioning method

Abstract

The invention discloses a multi-information fusion wind power station intelligent fire-fighting fire verification and positioning method, and belongs to the technical field of fire-fighting safety. The method comprises the following steps: establishing a three-dimensional dynamic model of a wind power station unit; a cloud platform is provided, the three-dimensional dynamic model is imported, and the sensor data and the unit operation state data displayed in real time are sent to the cloud platform, so that sensor data features and visual image features are identified and generated on the cloud platform, a feature matching library is established, and the unit operation state data are combined to obtain a feature matching library; and the fire risk of the key part is judged, and the fire risk of the wind power station unit is verified and positioned. According to the invention, by fusing various fire feature information and equipment operation state information, the false alarm rate of a single sensor is significantly reduced, and the accuracy and reliability of fire identification are improved. And by analyzing the spatial and logical association of the data of the multiple sensors, the specific part or area where the fire source occurs can be accurately judged.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of fire safety, and particularly relates to a multi-information fusion intelligent fire safety fire verification and positioning method for a wind power station. BACKGROUND

[0002] As a clean and renewable energy form, wind power plays a vital role in the transformation of global energy structure. With the rapid development of the wind power industry, the single machine capacity is continuously increasing, the scale of wind farms is expanding, and more and more wind farms are deployed in remote mountainous areas, coastal beaches, and even offshore areas with complex environments and few human traces. These development trends not only bring huge economic benefits, but also pose extremely severe challenges to the safe operation of wind farms, especially wind turbine generators. Fire is one of the most devastating accidents that threaten the safety of wind power stations. Once it occurs, it not only has a significant impact on the environment, but also poses a threat to human life. Therefore, early identification and diagnosis of fire hazards can reduce unnecessary losses.

[0003] Currently, the fire safety solutions commonly used in wind power stations are mainly automatic fire extinguishing systems based on traditional point-type smoke / temperature fire detectors, such as high-pressure water mist and gas fire extinguishing systems. In actual application, this system has significant defects, resulting in its "unintelligent" and unreliability. Specifically, the traditional system only relies on a single fire physical characteristic, such as smoke or temperature, which not only cannot cope with the interference of complex cabin environments, causing frequent false alarms, but also can only be alarmed passively and delayed after the occurrence of open flames or thick smoke. At the same time, due to the lack of linkage analysis of fire precursors such as bearing temperature and current, it cannot achieve early warning and risk prediction. More importantly, the existing system lacks fire source positioning capability, resulting in blind and inefficient fire extinguishing actions, which cannot timely save losses and poses great challenges to wind power stations in dealing with sudden fires. SUMMARY

[0004] The purpose of the present application is to overcome the shortcomings of the prior art and provide a multi-information fusion intelligent fire safety fire verification and positioning method for a wind power station.

[0005] The present application deeply fuses fire physical characteristic information (smoke, temperature, flame spectrum) with the operating state information (electrical parameters, mechanical temperature, vibration data) of the wind power station itself, realizes early and accurate warning (verification) of fire and precise positioning of the fire source through intelligent algorithms, and ultimately realizes the transformation of the intelligent fire safety mode from "passive alarm" to "active early warning and precise fire extinguishing", greatly improving the safety level of the wind power station.

[0006] Since the wind turbine is usually arranged in a complex environment and a rarely visited area, the present application aims to solve the problem of low intelligence and false alarm of the existing wind turbine fire extinguishing system, so as to reduce the labor and cost and improve the intelligence level of the fire extinguishing system.

[0007] In order to achieve the above object, the technical scheme adopted by the present application is as follows.

[0008] The present application provides a kind of multi-information fusion's wind power station wisdom fire extinguishing fire verification and positioning method, comprising the following steps: The CAD model of wind power station unit is constructed, and the sensor model of the key position in wind turbine is established, the installation coordinates of sensor model are labeled, the sensor data variable and unit operating state variable are set, and the sensor data variable and model attribute are associated to establish the three-dimensional dynamic model of wind power station unit.

[0009] A cloud platform is provided, the three-dimensional dynamic model of wind power station unit established is imported, and the real-time sensor data and unit operating state data displayed in wind power station unit are sent to cloud platform, to identify and generate sensor data features and visual image features in cloud platform, establish feature matching library, and judge the fire risk of key position in combination with unit operating state data, to verify and locate the fire risk of wind power station unit.

[0010] Preferably, the key position is the spatial position of the component that can cause fire in wind turbine.

[0011] In the present application, the key position is specifically the blade root, gear box and generator winding, etc.;The key position of sensor model can be determined according to the core components of wind power station unit to be modeled. In the present application, the three-dimensional dynamic model of wind power station unit mainly includes fire multi-type sensor, visual monitoring unit and unit operating state unit. By mapping the abstract sensor alarm signal to the specific spatial position of wind power station unit, the precise positioning of spatial coordinates is realized.

[0012] Preferably, the sensor data variable is temperature threshold and smoke concentration threshold, and the unit operating state variable is rotating speed and power. When the measured temperature of the key position exceeds the temperature threshold, the corresponding key position of the three-dimensional dynamic model is warned by color change.;Or, when the measured smoke concentration of the key position exceeds the smoke concentration threshold, the corresponding key position of the three-dimensional dynamic model is warned by color change.

[0013] The application mainly correlates the sensor data variables and the model attributes through the software parameterization function, and the model attributes can be color change early warning signals, for example, when the sensor real-time data detects whether a single sensor data reaches a threshold value, if yes, a pre-warning is triggered, and is marked as a to-be-verified state, and the early warning signal can be changed to a yellow flashing.

[0014] Preferably, the sensor corresponding to the sensor model is at least one of a temperature sensor, a smoke sensor, a flame sensor, an electric arc sensor, an oil mist sensor and an infrared flame sensor.

[0015] The application can set multiple types of sensors at corresponding key positions for real-time detection of sensor data, effectively improving the efficiency and accuracy of fire verification and positioning, thereby synchronously capturing traditional fire signals and equipment failure precursor information, laying a solid data cornerstone for extremely early verification and accurate positioning of fire.

[0016] Preferably, the sensor data is temperature and oil mist data of the key position; the unit operation state data is full load operation state data and over-maintenance state data; the sensor data features are temperature rising rate and smoke concentration change trend; the visual image features are flame color and smoke shape.

[0017] Preferably, the method for verifying the fire risk of the wind power station unit is as follows: A first-level verification is set, which is to compare the sensor data with the sensor data variables to determine the model attributes; when the sensor data reaches the sensor data variables, a pre-warning state is triggered, and the cloud platform marks the fire risk level of the corresponding key position as a to-be-verified state or yellow.

[0018] A second-level verification is set, which is that after triggering the pre-warning state, the cloud platform extracts the sensor data features and visual image features of the key position of the pre-warning, and combines the unit operation state data to obtain the number of abnormal unit operation state data, and when the number of abnormal unit operation state data is greater than or equal to 2, the cloud platform marks the fire risk level of the corresponding key position as a suspected fire state or orange.

[0019] A third-level verification is set, which is that in the suspected fire state, the cloud platform pushes the pre-warning information to the operation and maintenance personnel to confirm the fire risk level as a formal alarm or a released alarm.

[0020] Preferably, after the step of verifying the fire risk of the wind power station unit is executed, the step of positioning the fire risk of the wind power station unit is continuously executed; the method for positioning the fire risk of the wind power station unit is as follows: According to the installation coordinates of the sensor for fire verification, highlight the key parts for fire risk verification in the three-dimensional dynamic model to obtain the highlighted parts; according to the fire risk level of the highlighted parts, the fire risk of the wind power station unit is positioned, the actual coordinate data of the highlighted parts are obtained, and are synchronized to the monitoring center large screen and the emergency rescue terminal, so that the position of the fire occurrence is positioned.

[0021] The beneficial effects of the present application are as follows: 1. The present application significantly reduces the false alarm rate of a single sensor by fusing various fire feature information and equipment operating state information, and improves the accuracy and reliability of fire identification. And by analyzing the spatial and logical correlation of multi-sensor data, the specific components or areas where the fire source occurs are accurately judged to provide decision basis for directional fire extinguishing, that is, an intelligent fire fighting decision system is built, the fire fighting system and the fan main control system are deeply integrated, and the automation and intelligent fire fighting closed loop management from fire verification, intelligent judgment to precise action are realized; .2. The present application provides a wind power station intelligent fire fighting fire verification and positioning method based on multi-information fusion, which overcomes the false alarm problem caused by single sensor due to environmental interference through "multi-source information fusion" and "three-level verification mechanism", and greatly improves the credibility of judgment; 3. The present application monitors the equipment operating state information as a precursor feature of fire, discovers the fault fire hazard in advance through the abnormality of electrical and mechanical parameters (such as bearing overtemperature) before the appearance of traditional fire physical features (open fire, thick smoke), so as to better prevent, verify and locate the fire; 4. The positioning mechanism based on the "three-dimensional dynamic model" of the present application converts the abstract alarm signal into "precise spatial coordinates" inside the unit, which can not only inform that "a fire has occurred", but also accurately indicate the specific position, greatly improving the fire extinguishing efficiency and reducing the overall loss; 5. The present application changes the traditional process of relying on manual judgment and response into a smart closed loop of system autonomous judgment and accurate information pushing, significantly reducing the dependence on the experience and reaction speed of operation and maintenance personnel, and is particularly suitable for remote and difficult-to-inspect wind power stations, and improves the efficiency and level of safety management; 6. The system provided by the present application provides a feasible and effective solution, which not only has great help to the whole life cycle economic benefit of wind power station, but also has great potential for the application of fire in other scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is the specific flowchart of the wind turbine generator set fire verification and positioning method based on the hierarchical fusion algorithm of the embodiment of the present application.

[0023] Figure 2This is a flowchart of the three-level fire verification of a wind turbine generator set according to an embodiment of the present invention.

[0024] Figure 3 This is a flowchart illustrating the precise coordinate positioning and guidance for wind turbine generator fires according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0026] This invention primarily utilizes IoT, multi-sensor information fusion, and artificial intelligence technologies to provide a method for fire verification and location in wind power plants based on multi-information fusion. Multi-sensor information fusion technology comprehensively utilizes information from different sources and dimensions, processing it collaboratively through intelligent algorithms to obtain more accurate and reliable judgment results than any single information source.

[0027] The specific implementation scheme of this invention revolves around four core aspects of smart fire protection for wind power stations: "CAD modeling, sensor identification, multi-source data fusion, and verification and positioning".

[0028] Specifically, this invention utilizes CAD modeling methods to create a three-dimensional dynamic model of the wind turbine generator, lowering the barrier to understanding and using the system, thereby increasing the accessibility of the technology and encouraging a wider audience.

[0029] By combining commonly used fire alarm sensors, visual monitoring, and manual assisted monitoring, multi-source information fusion is achieved, and data synchronization and command reception are performed with the cloud platform. Data is received in real time, and a fire verification and location logic is set up to perform data judgment and analysis. Then, according to the rules, the fire risk level of each part of the unit is dynamically updated to achieve the goal of fire verification and location.

[0030] The system receives different information from sensor data and sets thresholds, uploading the data in real time to a cloud platform. The cloud stores a 3D model of the wind turbine unit, and simultaneously sets sensor data variables and unit operating status variables at corresponding locations on the model for auxiliary verification and location. When sensor data reaches the set threshold, logical judgments are made, and fire verification is performed in conjunction with monitoring vision and the 3D dynamic model. Through this design, the goal of intelligent fire protection and efficient fire verification and location for wind power stations is achieved.

[0031] The present invention will be further described in detail below with reference to the embodiments, but the scope of protection of the present invention is not limited to these embodiments.

[0032] A method for intelligent fire protection verification and location in wind power stations, incorporating multi-information fusion, includes the following steps: Step 1: Establish a three-dimensional dynamic model of the wind power station unit.

[0033] The specific method is as follows: construct a CAD model of the wind power station unit, establish sensor models for key parts of the wind turbine unit, mark the installation coordinates of the sensor models, set sensor data variables and unit operating status variables, and associate sensor data variables with model attributes to establish a three-dimensional dynamic model of the wind power station unit.

[0034] Specifically, the key components refer to the spatial locations of parts within a wind turbine generator that could trigger a fire. In this embodiment of the invention, these key components specifically include the blade roots, gearbox, and generator windings. The key locations for installing the sensor model can be determined based on the core components of the wind turbine generator to be modeled. In this invention, the three-dimensional dynamic model of the wind turbine generator mainly includes multiple types of fire sensors, a visual monitoring unit, and a unit operating status unit. By mapping abstract sensor alarm signals to the specific spatial location of the wind turbine generator, precise spatial coordinate positioning is achieved.

[0035] Specifically, the sensor data variables are temperature threshold and smoke concentration threshold; the unit operating status variables are speed and power; the method for associating sensor data variables with model attributes is as follows: When the measured temperature of a critical part exceeds the temperature threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning; or, when the measured smoke concentration of a critical part exceeds the smoke concentration threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning.

[0036] The main implementation of this invention is to associate sensor data variables with model attributes through software parameterization. The model attributes can be color-changing warning signals. For example, when the sensor detects whether the data of a single sensor reaches a threshold in real time, if so, a pre-alarm is triggered and marked as a state to be verified. The warning signal can be changed to a yellow flashing signal.

[0037] First, a CAD (Computer-Aided Design) model of the selected wind turbine generator is constructed, establishing a three-dimensional dynamic model of the wind turbine generator. The specific method is as follows: Step 1.1: Before modeling, identify the core components of the wind turbine generator to be modeled, including blades, hubs, gearboxes, generators, towers, and nacelle shells, and reserve modeling space for sensor installation locations in conjunction with fire monitoring functions.

[0038] Step 1.2: Collect basic data and information about the wind power station units, such as two-dimensional engineering drawings of the wind power station units and component assembly diagrams; of course, missing data can also be supplemented by on-site measurements based on specific existing wind power station units.

[0039] Step 1.3: Based on the collected basic data and information of the wind turbine generator set, create sensor models for key components of the wind turbine generator set (such as blade roots, gearbox, and generator windings). The sensor models can be simplified to cylinders or cubes, and their installation coordinates should be labeled to facilitate data variable binding on the cloud platform.

[0040] Step 1.4: Set sensor data variables (such as temperature and smoke concentration thresholds) and unit operating status variables (such as speed and power) at the corresponding locations in the 3D dynamic model, and associate the sensor data variables with model attributes (such as color change warning for the corresponding part of the model when the temperature exceeds the standard) through the software parameterization function.

[0041] Specifically, the sensor data variables are temperature threshold and smoke concentration threshold; the unit operating status variables are speed and power. The method for linking sensor data variables with model attributes through software parameterization is as follows: When the measured temperature of a critical part exceeds the temperature threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning; or, when the measured smoke concentration of a critical part exceeds the smoke concentration threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning.

[0042] Step 1.5: Optimize and adjust the 3D dynamic model, delete non-critical details, reduce the complexity of the 3D dynamic model, and facilitate subsequent loading and data synchronization on the cloud platform.

[0043] In this invention, the CAD model is the carrier for achieving precise spatial coordinate positioning and is the core technical means to map abstract sensor alarm signals (logical positions) to the specific physical positions (spatial positions) of wind power station units.

[0044] Step 2, sensor identification.

[0045] The sensor model corresponds to at least one of the following: a temperature sensor, a smoke sensor, a flame sensor, an arc sensor, an oil mist sensor, and an infrared flame sensor.

[0046] In this embodiment of the invention, the sensors are various types of sensors commonly used in fire monitoring, such as temperature sensors, smoke sensors, and flame sensors, to monitor temperature and smoke information at different locations in real time. This effectively improves the efficiency and accuracy of fire verification and location, and aims to simultaneously capture traditional fire signals and equipment failure precursor information, thereby laying a solid data foundation for the very early verification and accurate location of fires.

[0047] In addition, embodiments of the present invention can also be configured with differentiated monitoring sensors according to the fire risk characteristics of different parts of the unit, providing different physical parameters. The differentiated monitoring sensors are such as arc sensors, oil mist sensors and infrared flame sensors, which are used to assist in verifying and locating the occurrence of fire.

[0048] Step 3: Provide a cloud platform to import the established 3D dynamic model of the wind power station unit, and send the real-time sensor data and unit operation status data displayed in the wind power station unit to the cloud platform. The cloud platform will identify and generate sensor data features and visual image features, establish a feature matching library, and combine the unit operation status data to judge the fire risk of key parts.

[0049] Specifically, the sensor data includes temperature and oil mist data of key components; the unit operating status data includes full-load operating status data and over-maintenance status data; the sensor data features include the rate of temperature rise and the trend of smoke concentration change; and the visual image features include flame color and smoke shape.

[0050] The main implementation method of this invention is to fuse and identify multi-source data; the specific method is as follows: Step 3.1: Set up a cloud platform, import the established CAD model, and send the real-time unit operation status data to the cloud platform.

[0051] Step 3.2: After receiving data from multiple sources, the cloud platform uses a "layered fusion algorithm" to achieve information complementarity.

[0052] First, a weighted average is performed on the data from multiple sensors at the same location to eliminate errors from a single sensor, such as temporary drift from a temperature sensor. Specifically, the multiple sensor data includes the temperature and oil mist data of the gearbox; of course, the multiple sensor data can also be sensor data from other characteristic locations.

[0053] The next step is to identify sensor data features and visual image features, and establish a feature matching library. Specifically, sensor data features include the rate of temperature rise and the trend of smoke concentration changes; visual image features include flame color and smoke shape.

[0054] Furthermore, by combining the unit's operating status data (such as whether it is operating at full load or whether it has been maintained recently), a comprehensive assessment of fire risk is made to avoid false alarms caused by high operating temperatures (such as increased engine room temperature in summer). This allows for better identification and monitoring of the occurrence and location of fires, making the cloud platform not just a data dashboard, but also a data fusion and processing center.

[0055] The detailed flowchart of the fire verification and location method for wind turbine generator sets based on the hierarchical fusion algorithm is shown below. Figure 1 .

[0056] Step 4: Verify and locate the fire risks of wind power station units.

[0057] The embodiments of the present invention use different logic to verify and locate fires in wind power plants.

[0058] Step 4.1, the method for verifying the fire risk of wind power station units is as follows: The first level of verification is set up, which compares sensor data with sensor data variables to determine the model attributes. When the sensor data reaches the sensor data variable, a pre-alarm state is triggered, and the cloud platform marks the fire risk level of the corresponding key parts as pending verification or yellow.

[0059] A second level of verification is set up. After the pre-alarm state is triggered, the cloud platform extracts the sensor data features and visual image features of the key parts of the pre-alarm state, and combines them with the unit operation status data to obtain the number of abnormalities in the unit operation status data. When the number of abnormalities in the unit operation status data is ≥2, the cloud platform marks the fire risk level of the corresponding key parts as suspected fire state or orange.

[0060] A third level of verification is set up. In the event of a suspected fire, the cloud platform pushes an early warning message to the operation and maintenance personnel to confirm whether the fire risk level is a formal alarm or the alarm is deactivated.

[0061] Specifically, the embodiments of the present invention establish a "three-level verification mechanism" to ensure the accuracy of fire assessment, with the specific rules as follows: The first level of verification is sensor threshold triggering. Each sensor has a preset threshold. When a single sensor reaches the threshold, an "early alarm" state is triggered, and the cloud platform marks the area as "pending verification" or yellow.

[0062] Next, the second level of verification is carried out, namely cross-confirmation of multi-source data. After the pre-alarm, the system automatically retrieves other sensor data and visual images of the pre-alarm location. For example, if the temperature exceeds the threshold, it checks whether there is smoke or flame, whether the camera captures abnormal light spots, and combines the unit's operating status data analysis to detect whether there are any abnormalities in the data operation. If two or more data are abnormal, it is upgraded to "suspected fire".

[0063] Finally, there is the third level of verification, namely manual confirmation. In the event of a suspected fire, the system pushes an early warning message to the maintenance personnel, including a 3D model location screenshot and a real-time video link. The maintenance personnel confirm the fire by remotely viewing or on-site inspection. If a fire is confirmed, a "formal alarm" is triggered and emergency measures are taken; otherwise, the warning is lifted.

[0064] A flowchart of the three-level fire verification process for wind turbine generator sets, as follows: Figure 2 As shown.

[0065] Step 4.2: After verifying the fire risk of the wind turbine generator sets, proceed to locate the fire risk of the wind turbine generator sets. The method for locating the fire risk of the wind turbine generator sets is as follows: Based on the installation coordinates of the sensors used for fire verification, key areas for fire risk verification are highlighted in the 3D dynamic model to obtain the highlighted areas. Based on the fire risk level of the highlighted areas, the fire risk of the wind power station units is located, the actual coordinate data of the highlighted areas is obtained, and the data is synchronized to the monitoring center's large screen and emergency rescue terminal to locate the fire location.

[0066] For the fire location logic, the implementation method of this invention is mainly based on a three-dimensional dynamic model to achieve "precise coordinate positioning". The core is to use the three-dimensional model to realize the conversion from "logical alarm" to "spatial coordinates".

[0067] First, key components are located. Based on the installation location of alarm sensors and the location of each component of the unit, i.e., the sensor IDs and coordinates are preset in the model, the component is highlighted in the 3D model. According to the fire risk level of different parts on the model, and combined with the detailed location information of the part displayed on the monitoring, the positioning results are converted into actual geographic coordinates and synchronized to the monitoring center screen and emergency rescue terminal to provide accurate guidance for fire fighting and rescue, and achieve precise positioning of the fire location within 1 meter.

[0068] Meanwhile, since existing technologies do not have the ability to accurately locate based on three-dimensional models, this invention can also adopt a visual presentation, that is, on a three-dimensional dynamic model, fire risk level and location information can be displayed in real time and dynamically through color changes (such as green-yellow-red) and icon flashing, which can greatly reduce the threshold for information understanding.

[0069] The flowchart of the precise coordinate positioning and guidance for wind turbine fire according to an embodiment of the present invention is as follows: Figure 3 As shown.

[0070] The embodiments of this invention solve long-standing pain points in the field of fire protection for wind power plants through fundamental innovation in technical architecture, and achieve comprehensive breakthroughs in four core dimensions: reliability, timeliness, accuracy and intelligence.

[0071] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent fire protection verification and location in wind power stations using multi-information fusion, characterized in that, Includes the following steps: A CAD model of the wind power station unit is constructed, and sensor models are established for key parts of the wind turbine unit. The installation coordinates of the sensor models are marked, sensor data variables and unit operating status variables are set, and sensor data variables are associated with model attributes to establish a three-dimensional dynamic model of the wind power station unit. A cloud platform is provided to import the established 3D dynamic model of the wind power station unit and send the real-time sensor data and unit operation status data displayed in the wind power station unit to the cloud platform. The cloud platform identifies and generates sensor data features and visual image features, establishes a feature matching library, and combines the unit operation status data to judge the fire risk of key parts, thereby realizing the verification and location of fire risks of wind power station units.

2. The method for intelligent fire protection and fire location in wind power stations based on multi-information fusion according to claim 1, characterized in that, The critical parts refer to the spatial locations of components in a wind turbine generator that could trigger a fire.

3. The method for intelligent fire protection and fire location verification in wind power stations based on multi-information fusion according to claim 1, characterized in that, The sensor data variables are temperature threshold and smoke concentration threshold; the unit operating status variables are speed and power; the method for associating sensor data variables with model attributes is as follows: When the measured temperature of a critical part exceeds the temperature threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning; or, when the measured smoke concentration of a critical part exceeds the smoke concentration threshold, the critical part corresponding to the 3D dynamic model will change color to issue a warning.

4. The method for intelligent fire protection and fire location verification in wind power stations based on multi-information fusion according to claim 1, characterized in that, The sensor model corresponds to at least one of the following: a temperature sensor, a smoke sensor, a flame sensor, an arc sensor, an oil mist sensor, and an infrared flame sensor.

5. The method for intelligent fire protection and fire location verification in wind power stations based on multi-information fusion according to claim 1, characterized in that, The sensor data includes temperature and oil mist data of key components; the unit operating status data includes full-load operating status data and over-maintenance status data; the sensor data features include the rate of temperature rise and the trend of smoke concentration change; the visual image features include flame color and smoke shape.

6. The method for intelligent fire protection and fire location in wind power stations based on multi-information fusion according to claim 1, characterized in that, The method for verifying the fire risk of wind power station units is as follows: The first level of verification is set up, which compares sensor data with sensor data variables to determine the model attributes. When the sensor data reaches the sensor data variable, a pre-alarm state is triggered, and the cloud platform marks the fire risk level of the corresponding key parts as pending verification or yellow. The second level of verification is set up. After the pre-alarm state is triggered, the cloud platform extracts the sensor data features and visual image features of the key parts of the pre-alarm state, and combines them with the unit operation status data to obtain the number of abnormalities in the unit operation status data. When the number of abnormalities in the unit operation status data is ≥2, the cloud platform marks the fire risk level of the corresponding key parts as suspected fire state or orange. A third level of verification is set up. In the event of a suspected fire, the cloud platform pushes an early warning message to the operation and maintenance personnel to confirm whether the fire risk level is a formal alarm or the alarm is deactivated.

7. The method for intelligent fire protection and fire location in wind power stations based on multi-information fusion according to claim 6, characterized in that, After performing the fire risk verification steps for the wind power station units, the fire risk positioning for the wind power station units will continue. The method for locating the fire risk of wind power station units is as follows: Based on the installation coordinates of the sensors used for fire verification, the key areas for fire risk verification are highlighted in the 3D dynamic model to obtain the highlighted areas. Based on the fire risk level of the highlighted areas, the fire risk of the wind power station units is located, the actual coordinate data of the highlighted areas is obtained, and the data is synchronized to the monitoring center's large screen and emergency rescue terminal to locate the location of the fire.

Images