A wind farm booster station fire alarm linkage monitoring method and system

Abstract

The present application relates to the technical field of fire monitoring, and discloses a wind power plant booster station fire alarm linkage monitoring method and system, multi-source sensing signals of a booster station equipment area are collected; the multi-source sensing signals are subjected to space-time domain fusion processing to obtain a fusion feature spectrum; a fire evolution characteristic vector is extracted based on the fusion feature spectrum, representing the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate and electrical parameter abnormal fluctuation amplitude; fire risk grade determination is carried out according to the fire evolution characteristic vector, and a graded alarm instruction is generated based on the fire risk grade determination, the response intensity of the graded alarm instruction is positively correlated with the fire risk grade determination result, accurate identification of the combustion type and dynamic determination of the risk grade are realized, and in combination with the graded alarm response and the sensitivity self-adaptive adjustment mechanism, the identification ability for complex fire scenes is improved while the false alarm rate is reduced, and the fire safety of the wind power plant booster station is ensured.

Description

Technical Field

[0001] This invention relates to the field of fire monitoring technology, and more specifically, to a fire alarm linkage monitoring method and system for wind farm substations. Background Technology

[0002] Fire detection and alarm at wind farm substations refers to the technology that monitors signals such as temperature, smoke, and electrical parameters within the substation to promptly detect fire hazards and issue alarms. It is a key component in ensuring the safe operation of wind farms.

[0003] Existing fire detection systems for substations primarily employ threshold alarm methods using single sensors (such as smoke or heat detectors). When the value of a sensor exceeds a fixed threshold, an audible and visual alarm is triggered. However, current technologies struggle to integrate multi-source heterogeneous signals for cross-verification, cannot distinguish between different combustion types such as electrical fires and combustible material fires, and lack adaptability to complex electromagnetic environments and meteorological interference. Furthermore, existing systems often use fixed sensitivity parameters, failing to adaptively adjust based on on-site interference characteristics. This results in high false alarm rates in scenarios such as sandstorms, strong light, and welding operations, or missed alarms under new combustion modes, making it difficult to meet the high reliability and safety protection requirements of wind farm substations. Summary of the Invention

[0004] This invention provides a fire alarm linkage monitoring method and system for wind farm substations. By integrating multi-source sensing signals in the spatiotemporal domain and extracting fire evolution features, it achieves accurate identification of combustion types and dynamic determination of risk levels. Combined with a graded alarm response and sensitivity adaptive adjustment mechanism, and supporting self-learning updates for unknown combustion types, it reduces false alarm rates while improving the ability to identify complex fire scenarios, thus ensuring the fire safety of wind farm substations.

[0005] To achieve the above objectives, the present invention provides a fire alarm linkage monitoring method for wind farm booster stations, comprising: Collect multi-source sensing signals from the equipment area of ​​the booster station. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. The multi-source sensing signals are subjected to spatiotemporal fusion processing to obtain a fused feature map; Fire evolution feature vectors are extracted based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; Fire risk level is determined based on the fire evolution feature vector, and graded alarm commands are generated based on the fire risk level determination. The response intensity of the graded alarm commands is positively correlated with the fire risk level determination result.

[0006] Further, the multi-source sensing signals are subjected to spatiotemporal fusion processing to obtain a fused feature map, including: The sampling times of the infrared thermal radiation signal, visible light image signal, gaseous suspended particulate matter concentration signal and electrical insulation status signal are aligned to establish a multi-channel data frame with a unified time reference. Calculate the geometric registration offset of each channel's data frame in the spatial coordinate system, perform pixel-level spatial resampling based on the geometric registration offset, and generate a spatiotemporally aligned multi-source observation matrix; Based on the multi-source observation matrix, feature correlation mapping between channels is performed to construct the fused feature map, which includes a temperature-smoke joint distribution layer, an electrical-thermal field coupling layer, and a visible light texture verification layer.

[0007] Further, based on the fused feature map, fire evolution feature vectors are extracted, including: A dynamic tracking window is set up in the fused feature map to monitor the cumulative growth trend of the temperature field diffusion rate within a preset time segment; Extract the spatial distribution pattern of the smoke concentration gradient change rate, and identify the dominant direction of smoke diffusion and the topological structure of the diffusion front; The high-frequency pulse component in the electrical insulation status signal is monitored for synchronization correlation with the infrared thermal radiation signal. When the synchronization correlation exceeds a preset coupling threshold, it is marked as an electrical thermal runaway associated region. The cumulative growth trend, the spatial distribution pattern, and the synchronous correlation are vectorized to construct the fire evolution feature vector.

[0008] Furthermore, the fire risk level is determined based on the fire evolution characteristic vector, including: The parameters of each dimension in the fire evolution feature vector are matched with the preset combustion type discrimination boundary, which includes the electrical equipment overheating type boundary, the combustible material smoldering type boundary and the mixed combustion type boundary; The combustion type is determined based on the matching results; Based on the combustion type classification and the magnitude of the fire evolution characteristic vector, the fire risk level determination is generated by mapping. The fire risk level determination includes the initial risk level, the diffusion risk level, and the critical risk level.

[0009] Furthermore, based on the matching results, the combustion type corresponding to the current fire evolution characteristics is determined, including: When the matching degree between the fire evolution feature vector and any combustion type discrimination boundary is lower than the preset confidence threshold, it is determined as an unconfirmed suspected event; For the unconfirmed suspected events, a multi-condition cross-validation mechanism is initiated, which includes: verifying the continuous upward trend of the concentration signal of the gaseous suspended particulate matter, verifying the persistence of texture abrupt changes in the visible light image signal, and verifying the transmission consistency of infrared thermal radiation signals in adjacent areas. When the output of the multi-condition cross-validation mechanism meets the preset confirmation conditions, the unconfirmed suspected event is confirmed as a real fire event and the fire risk level is upgraded. If the preset confirmation conditions are not met, suppress the generation of the current alarm command and mark it as an environmental interference event.

[0010] Furthermore, after suppressing the generation of the current alarm command and marking it as an environmental interference event, it also includes: The frequency of occurrence of the aforementioned environmental disturbance events was statistically analyzed.

[0011] Furthermore, it also includes: When the frequency of occurrence exceeds a preset interference threshold, the sensitivity parameter of the combustion type discrimination boundary is adjusted according to the multidimensional transient characteristics of the environmental interference event. The sensitivity parameter includes the trigger threshold of the temperature field diffusion rate and the response coefficient of the smoke concentration gradient change rate. The feature extraction weights of the fused feature map are updated based on the adjusted sensitivity parameters, reducing the weight ratio of the high-frequency interference channel and enhancing the response gain of the stable feature channel.

[0012] Furthermore, based on the fire risk level determination, a graded alarm command is generated, including: When the initial risk level is determined, an equipment inspection early warning command and an enhanced monitoring frequency command are generated, triggering continuous acquisition of visible light images of the local area. When the risk level of diffusion is determined, an audible and visual alarm activation command and a fire pre-activation preparation command are generated to activate the pre-pressurization state of the automatic fire extinguishing device. When the critical risk level is determined, an emergency evacuation command for the entire station, a power cut-off interlock command, and an automatic fire extinguishing spray command are generated to cut off the electrical connection of the fault area and start the smoke exhaust of the entire area.

[0013] Furthermore, it generates station-wide emergency evacuation commands, power cut-off interlock commands, and automatic fire suppression spray commands, cuts off electrical connections in the fault area, and initiates full-area smoke extraction, including: Based on the spatial distribution pattern in the fire evolution characteristic vector, the main combustion area and smoke diffusion path are located; Based on the main combustion area and the smoke diffusion path, calculate the optimal evacuation route and smoke avoidance passage for personnel, and embed the optimal route into the station-wide emergency evacuation command; Based on the distribution of the electrical thermal runaway associated areas, the order of electrical disconnection is determined, prioritizing the disconnection of upstream switching equipment electrically connected to the main combustion area, while preserving the power supply circuits for the monitoring system and emergency lighting; Based on the wind direction compensation coefficient of the smoke diffusion path, the spray angle and coverage of the automatic fire extinguishing spray command are adjusted to ensure that the fire extinguishing medium covers the leading edge of the smoke diffusion path.

[0014] To achieve the above objectives, the present invention also provides a fire alarm linkage monitoring system for wind farm substations, comprising: The signal acquisition module is used to acquire multi-source sensing signals from the equipment area of ​​the substation. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. The map fusion module is used to perform spatiotemporal domain fusion processing on the multi-source sensing signals to obtain a fused feature map; The vector construction module is used to extract fire evolution feature vectors based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; The fire monitoring module is used to determine the fire risk level based on the fire evolution feature vector, and to generate graded alarm commands based on the fire risk level determination. The response intensity of the graded alarm commands is positively correlated with the fire risk level determination result.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention discloses a fire alarm linkage monitoring method and system for wind farm substations. The method involves collecting multi-source sensing signals from the substation equipment area; performing spatiotemporal fusion processing on the multi-source sensing signals to obtain a fused feature map; extracting fire evolution feature vectors based on the fused feature map to characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; determining the fire risk level based on the fire evolution feature vectors; generating graded alarm commands based on the fire risk level determination; and ensuring the fire safety of wind farm substations by positively correlated response intensity with the fire risk level determination result, thereby achieving accurate identification of combustion types and dynamic risk level determination. Combined with a graded alarm response and sensitivity adaptive adjustment mechanism, this method reduces false alarm rates while improving the ability to identify complex fire scenarios. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a fire alarm linkage monitoring method for a wind farm booster station according to an embodiment of the present invention is shown. Figure 2 A schematic diagram of a fire alarm linkage monitoring system for a wind farm substation is shown in an embodiment of the present invention. Detailed Implementation

[0017] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] The following is a description of preferred embodiments of the present invention in conjunction with the accompanying drawings.

[0022] like Figure 1 As shown, an embodiment of the present invention discloses a fire alarm linkage monitoring method for a wind farm booster station, comprising: S110: Collect multi-source sensing signals from the equipment area of ​​the booster station. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. S120: Perform spatiotemporal domain fusion processing on the multi-source sensing signals to obtain a fused feature map; S130: Extract fire evolution feature vectors based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; S140: Determine the fire risk level based on the fire evolution characteristic vector, and generate a graded alarm command based on the fire risk level determination. The response intensity of the graded alarm command is positively correlated with the fire risk level determination result.

[0023] In this embodiment, the substation equipment area refers to the primary equipment area (main transformer, circuit breaker, disconnector), secondary equipment area (protection panel, control cabinet), and cable trench area within the wind farm substation. Multi-source sensing signals are collected through a heterogeneous sensor network: infrared thermal radiation signals are acquired using infrared thermal imagers or point-type infrared temperature sensors to reflect the surface temperature distribution of the equipment; visible light image signals are acquired using high-definition network cameras for visual confirmation; gaseous particulate matter concentration signals are acquired using laser scattering smoke detectors or particulate matter sensors to monitor smoke particle concentration; and electrical insulation status signals are acquired using partial discharge detectors and leakage current monitoring devices to reflect the degree of insulation degradation of the equipment.

[0024] In some embodiments of this application, the multi-source sensing signals are subjected to spatiotemporal fusion processing to obtain a fused feature map, including: The sampling times of the infrared thermal radiation signal, visible light image signal, gaseous suspended particulate matter concentration signal and electrical insulation status signal are aligned to establish a multi-channel data frame with a unified time reference. Calculate the geometric registration offset of each channel's data frame in the spatial coordinate system, perform pixel-level spatial resampling based on the geometric registration offset, and generate a spatiotemporally aligned multi-source observation matrix; Based on the multi-source observation matrix, feature correlation mapping between channels is performed to construct the fused feature map, which includes a temperature-smoke joint distribution layer, an electrical-thermal field coupling layer, and a visible light texture verification layer.

[0025] In this embodiment, sampling time alignment adopts a network time protocol or a precise time protocol to ensure that the data acquisition timestamp deviation of the four types of sensors is less than 100 milliseconds, establishing a unified time reference. A multi-channel data frame refers to a data set composed of four types of data at the same time. Geometric registration offset refers to the difference in coordinates of the same physical point in the image caused by different sensor installation positions (such as the physical distance between an infrared camera and a visible light camera), calculated using affine transformation parameters or homography matrices obtained through calibration. Pixel-level spatial resampling refers to performing geometric transformations (such as perspective transformation or bilinear interpolation) on the image based on the offset, so that all sensor data correspond to a unified substation plane coordinate system (with the main transformer center as the origin, the vertical axis as the X-axis, the horizontal axis as the Y-axis, and a resolution of 5 cm per pixel). The multi-source observation matrix is ​​a four-dimensional data structure (height × width × channel × time), such as 512 pixels × 512 pixels × 4 channels × 1 time. Inter-channel feature correlation mapping refers to calculating the spatial correlation between different physical quantities, such as whether a high-temperature area simultaneously has high smoke concentration and strong electrical discharge. The temperature-smoke joint distribution layer characterizes the spatial overlap between the temperature and smoke fields; for example, the overlap of a high-temperature zone and a smoke zone indicates an open flame, while the absence of smoke in a high-temperature zone indicates overheating. The electrical-thermal coupling layer characterizes the correlation between electrical anomalies and temperature anomalies; for example, the overlap of a partial discharge point and a hot spot indicates an electrical fire. The visible light texture verification layer provides visual confirmation information, such as flame texture and smoke color, to eliminate interference.

[0026] The beneficial effects of the above technical solution are: time alignment eliminates data misalignment in asynchronous sampling by multiple sensors; geometric registration achieves spatial unification of heterogeneous data, enabling different sensors to align with the same physical area; and through the hierarchical construction of fused feature maps, a multi-physics coupling relationship between temperature, smoke, electrical, and vision is established, providing a complete spatiotemporal data foundation for subsequent feature extraction and fire identification, and improving the accuracy of fire location.

[0027] In some embodiments of this application, extracting fire evolution feature vectors based on the fused feature map includes: A dynamic tracking window is set up in the fused feature map to monitor the cumulative growth trend of the temperature field diffusion rate within a preset time segment; Extract the spatial distribution pattern of the smoke concentration gradient change rate, and identify the dominant direction of smoke diffusion and the topological structure of the diffusion front; The high-frequency pulse component in the electrical insulation status signal is monitored for synchronization correlation with the infrared thermal radiation signal. When the synchronization correlation exceeds a preset coupling threshold, it is marked as an electrical thermal runaway associated region. The cumulative growth trend, the spatial distribution pattern, and the synchronous correlation are vectorized to construct the fire evolution feature vector.

[0028] In this embodiment, the dynamic tracking window is a movable spatial area (e.g., 5m x 5m) that slides across the fused feature map to track the movement and expansion of the high-temperature area. The preset time segment is set to 5 minutes, and the cumulative growth trend is calculated as the average temperature rise in the window area within 5 minutes (e.g., from 40℃ to 60℃, with a trend of 20℃ every 5 minutes). The spatial distribution of the smoke concentration gradient change rate is obtained by calculating the spatial gradient vector of the smoke concentration field, identifying the dominant direction of smoke diffusion (e.g., northeast) and the topological structure of the diffusion front (e.g., circular diffusion, directional spray, layered accumulation). The high-frequency pulse component in the electrical insulation status signal refers to the high-frequency current pulse (frequency range 1 MHz to 100 MHz) generated by partial discharge. Synchronous correlation is calculated through cross-correlation analysis with the infrared thermal radiation signal. If the correlation coefficient is greater than 0.8 (preset coupling threshold), it indicates that electrical discharge and temperature rise occur simultaneously, marking the area as an electrical thermal runaway associated area, indicating an electrical fire. Vector splicing refers to arranging the feature values ​​of the above three dimensions in a fixed order, such as [temperature growth rate value, smoke dominance direction encoding, smoke topology complexity, electrical-thermal coupling coefficient, spatial coordinates], to construct a fire evolution feature vector for subsequent level determination.

[0029] The beneficial effects of the above technical solution are as follows: the dynamic tracking window enables the movement tracking and extended monitoring of the fire area; the smoke gradient analysis identifies the diffusion direction and leading edge structure, predicting the fire spread path; the electrical-thermal coupling analysis distinguishes electrical fires from external fire sources; and the vector splicing forms a standardized fire feature description, which includes fire intensity information (temperature increase), fire type information (electrical coupling), and spatial evolution information (diffusion direction), providing multi-dimensional input for accurate risk level determination.

[0030] In some embodiments of this application, determining the fire risk level based on the fire evolution feature vector includes: The parameters of each dimension in the fire evolution feature vector are matched with the preset combustion type discrimination boundary, which includes the electrical equipment overheating type boundary, the combustible material smoldering type boundary and the mixed combustion type boundary; The combustion type is determined based on the matching results; Based on the combustion type classification and the magnitude of the fire evolution characteristic vector, the fire risk level determination is generated by mapping. The fire risk level determination includes the initial risk level, the diffusion risk level, and the critical risk level.

[0031] In this embodiment, the combustion type discrimination boundary is a classification hyperplane or decision region divided in the feature space, trained based on historical fire case data. The characteristics of the electrical equipment overheating type boundary are a high electrical-thermal coupling coefficient (greater than 0.7), moderate temperature increase (10 to 30°C every 5 minutes), and low smoke concentration (due to less smoke in the initial stage of insulation material overheating). The characteristics of the combustible smoldering type boundary are low electrical-thermal coupling (less than 0.3), low temperature but high smoke gradient (low smoldering temperature but large smoke). The mixed combustion type boundary lies between the two. The magnitude of the fire evolution feature vector is calculated by taking the square root of the sum of the squares of each dimension, representing the overall energy release intensity of the fire. When mapping to generate risk levels, a magnitude less than the first threshold (e.g., low energy) is the initial risk level; a magnitude between the first and second thresholds with a clearly defined combustion type is the diffusion risk level; and a magnitude greater than the second threshold or a mixture of multiple combustion types is the critical risk level. Different combustion types affect the risk judgment weight; for example, even with a moderate magnitude, the electrical overheating type may be upgraded to a higher risk level due to its high hazard.

[0032] The beneficial effects of the above technical solution are as follows: the nature of the fire is classified and identified by the combustion type discrimination boundary, which distinguishes electrical fires from ordinary combustible fires, providing a basis for subsequent targeted treatment (such as power outage fire extinguishing vs. foam fire extinguishing); the fire intensity is quantified by the feature vector modulus; and the risk is refined by the graded judgment, avoiding misjudgment caused by a single threshold, and improving the accuracy of alarms and the appropriateness of response.

[0033] In some embodiments of this application, determining the combustion type attribution corresponding to the current fire evolution characteristics based on the matching results includes: When the matching degree between the fire evolution feature vector and any combustion type discrimination boundary is lower than the preset confidence threshold, it is determined as an unconfirmed suspected event; For the unconfirmed suspected events, a multi-condition cross-validation mechanism is initiated, which includes: verifying the continuous upward trend of the concentration signal of the gaseous suspended particulate matter, verifying the persistence of texture abrupt changes in the visible light image signal, and verifying the transmission consistency of infrared thermal radiation signals in adjacent areas. When the output of the multi-condition cross-validation mechanism meets the preset confirmation conditions, the unconfirmed suspected event is confirmed as a real fire event and the fire risk level is upgraded. If the preset confirmation conditions are not met, suppress the generation of the current alarm command and mark it as an environmental interference event.

[0034] In this embodiment, the preset threshold is typically set to 0.6 or 0.7 (range 0 to 1). When the classifier's discrimination probability for each combustion type is lower than this value, it indicates that the feature vector is in a classification blind zone, which may be a new type of fire or environmental interference, and is judged as an unconfirmed suspected event. The multi-condition cross-validation mechanism improves the reliability of confirmation through triple verification: First, verify the continuous upward trend of smoke concentration, requiring at least three consecutive sampling periods (e.g., 15 minutes) of monotonous increase to exclude instantaneous dust interference; Second, verify the persistence of texture abrupt changes in visible light images, detecting moving flame textures through inter-frame difference or optical flow methods, requiring the presence of flame flickering features in multiple consecutive frames to exclude strong light irradiation or reflection interference; Third, verify the consistency of infrared thermal radiation signal transmission in adjacent areas, checking whether the high-temperature area expands to the surrounding area according to the heat conduction law (e.g., reasonable temperature gradient), excluding single-point sensor failure or external heat source projection interference. The preset confirmation conditions are set to at least two of the triple verifications being passed, or a specific combination (e.g., continuous smoke and continuous texture). When the conditions are met, the alarm is upgraded to a real fire, and the risk level is determined according to the characteristic vector magnitude. When the conditions are not met, the alarm is suppressed and marked as environmental interference (such as strong direct sunlight, welding operations, sandstorms) to avoid false alarms.

[0035] The beneficial effects of the above technical solution are: by setting a confidence threshold, the diversity and uncertainty of fire types are acknowledged, avoiding misjudgments caused by forced classification; by using a multi-condition cross-validation mechanism, the false alarm rate caused by single sensor failure or environmental interference is significantly reduced; and by using an alarm suppression and marking mechanism, the system's reliability and availability are improved by learning and eliminating interference events.

[0036] In some embodiments of this application, after suppressing the generation of the current alarm command and marking it as an environmental interference event, the method further includes: The frequency of occurrence of the aforementioned environmental disturbance events was statistically analyzed.

[0037] In some embodiments of this application, it also includes: When the frequency of occurrence exceeds a preset interference threshold, the sensitivity parameter of the combustion type discrimination boundary is adjusted according to the multidimensional transient characteristics of the environmental interference event. The sensitivity parameter includes the trigger threshold of the temperature field diffusion rate and the response coefficient of the smoke concentration gradient change rate. The feature extraction weights of the fused feature map are updated based on the adjusted sensitivity parameters, reducing the weight ratio of the high-frequency interference channel and enhancing the response gain of the stable feature channel.

[0038] In this embodiment, the frequency statistics refer to the number of times marked as environmental interference within a unit of time (e.g., within a week), and the multidimensional transient characteristics refer to the characteristics of the interference event, such as strong sunlight interference manifesting as high temperature but no smoke or electrical abnormalities, and welding operations manifesting as localized high temperature but no smoke diffusion. Sensitivity parameter adjustment refers to modifying the discrimination boundary according to the type of interference: if sunlight interference occurs frequently, the trigger threshold of the temperature field diffusion rate is increased (e.g., from 10°C per minute to 15°C per minute) to reduce the sensitivity to temperature abrupt changes; if dust interference occurs frequently, the response coefficient of the smoke concentration gradient change rate is reduced (e.g., the weight is reduced from 0.4 to 0.2) to reduce the contribution of the smoke channel. The feature extraction weight of the fused feature map refers to the fusion weight of each sensor channel when constructing the map, such as reducing the weight of the visible light channel (to avoid strong light interference) and increasing the weight of the electrical channel (to highlight electrical fire characteristics). Through dynamic adjustment, the system gradually adapts to the characteristics of the on-site environment, achieving a self-learning effect of "becoming more accurate with use".

[0039] The beneficial effects of the above technical solution are as follows: by statistically analyzing the spatiotemporal patterns of interference events, environmental noise patterns are identified; by adaptively adjusting sensitivity parameters, the detection system is optimized for specific environments; and by redistributing weights, interference channels are suppressed, effective channels are enhanced, repeatability false alarms are significantly reduced, and the long-term stability and reliability of the detection system in complex industrial environments are improved.

[0040] In some embodiments of this application, generating graded alarm instructions based on the fire risk level determination includes: When the initial risk level is determined, an equipment inspection early warning command and an enhanced monitoring frequency command are generated, triggering continuous acquisition of visible light images of the local area. When the risk level of diffusion is determined, an audible and visual alarm activation command and a fire pre-activation preparation command are generated to activate the pre-pressurization state of the automatic fire extinguishing device. When the critical risk level is determined, an emergency evacuation command for the entire station, a power cut-off interlock command, and an automatic fire extinguishing spray command are generated to cut off the electrical connection of the fault area and start the smoke exhaust of the entire area.

[0041] In this embodiment, the initial risk level is primarily addressed through observation and confirmation: equipment inspection warnings are pushed to maintenance personnel via SMS or mobile terminals, indicating suspected anomalies in a certain area requiring on-site verification; the enhanced monitoring frequency command increases the infrared and visible light sampling frequency from once per minute to once per second, enabling continuous video monitoring. The spread risk level is addressed primarily through warning and preparation: the audible and visual alarm activation command drives the audible and visual alarm to emit a high-decibel alarm and flashing bright light, alerting on-site personnel; the fire pre-start preparation command controls the automatic fire extinguishing system (such as a gas extinguishing device or water spray system) to enter a pre-pressurization state, with valves opening and pressure building up, ready to spray at any time, shortening the formal fire extinguishing activation time.

[0042] The beneficial effects of the above technical solution are as follows: by accurately matching the three-level alarm with the three-level response, the fire response is made gradient and appropriate, avoiding overreaction in low-risk situations (such as false alarms leading to unnecessary power outages) and insufficient response in high-risk situations; the pre-start preparation mechanism shortens the fire extinguishing response time; and by using power cut-off interlocking and smoke exhaust linkage, both personnel safety and equipment protection are taken into account, thereby improving the overall efficiency of the substation fire emergency response.

[0043] In some embodiments of this application, generating a station-wide emergency evacuation command, a power cut-off interlock command, and an automatic fire suppression spray command, cutting off the electrical connection of the fault area and initiating full-area smoke extraction, includes: Based on the spatial distribution pattern in the fire evolution characteristic vector, the main combustion area and smoke diffusion path are located; Based on the main combustion area and the smoke diffusion path, calculate the optimal evacuation route and smoke avoidance passage for personnel, and embed the optimal route into the station-wide emergency evacuation command; Based on the distribution of the electrical thermal runaway associated areas, the order of electrical disconnection is determined, prioritizing the disconnection of upstream switching equipment electrically connected to the main combustion area, while preserving the power supply circuits for the monitoring system and emergency lighting; Based on the wind direction compensation coefficient of the smoke diffusion path, the spray angle and coverage of the automatic fire extinguishing spray command are adjusted to ensure that the fire extinguishing medium covers the leading edge of the smoke diffusion path.

[0044] In this embodiment, the spatial distribution pattern is determined through contour analysis of the smoke concentration field and temperature field. The main combustion area refers to the high-temperature, high-smoke core area, and the smoke diffusion path refers to the channel through which the smoke diffuses with the airflow. Optimal path calculation is based on a graph search algorithm (such as the A* algorithm), prioritizing smoke-avoidance channels (paths where the smoke concentration is below the safety threshold) to avoid combustion areas and diffusion paths, generating dynamic routes from each post to the safety exit, and embedding evacuation instructions for guidance via broadcasts and indicator lights. The order of electrical disconnection is determined according to the single-line diagram of the substation, prioritizing the disconnection of upstream circuit breakers in the fault area (such as the low-voltage side circuit breaker of the main transformer) to isolate the fault point, but maintaining power supply to the station's monitoring system, emergency lighting, and fire-fighting facilities (through independent circuits or UPS) to ensure uninterrupted fire extinguishing and evacuation command. The wind direction compensation coefficient is obtained in real time through anemometers. For example, if the east wind is 3 meters per second, the spray angle is adjusted to deflect westward by a certain angle (such as 15 degrees) to compensate for the wind's deviation of the spray medium, ensuring that the extinguishing agent accurately covers the smoke diffusion front (downwind area) and improving fire extinguishing efficiency.

[0045] The beneficial effects of the above technical solution are: by combining the combustion zone and smoke path to plan evacuation routes, personnel are prevented from accidentally entering dense smoke areas or fire scenes; the hierarchical power outage strategy achieves a balance between fault isolation and necessary power supply, preventing the fire from spreading along electrical lines while ensuring the operation of the emergency system; and wind direction compensation enables precise control of fire extinguishing spray, ensuring that the fire extinguishing medium reaches the effective area, thereby improving the success rate of automatic fire extinguishing and the safety of personnel evacuation.

[0046] To further illustrate the technical concept of this invention, the technical solution of this invention will now be described in conjunction with specific application scenarios.

[0047] Correspondingly, such as Figure 2 As shown, this application also provides a fire alarm linkage monitoring system for wind farm substations, including: The signal acquisition module is used to acquire multi-source sensing signals from the equipment area of ​​the substation. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. The map fusion module is used to perform spatiotemporal domain fusion processing on the multi-source sensing signals to obtain a fused feature map; The vector construction module is used to extract fire evolution feature vectors based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; The fire monitoring module is used to determine the fire risk level based on the fire evolution feature vector, and to generate graded alarm commands based on the fire risk level determination. The response intensity of the graded alarm commands is positively correlated with the fire risk level determination result.

[0048] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0049] Although the invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the embodiments disclosed in this invention can be combined with each other in any way. The fact that not all of these combinations are described in this specification is merely for the sake of brevity and resource conservation.

[0050] It will be understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., 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 fire alarm linkage monitoring of a wind farm substation, characterized in that, include: Collect multi-source sensing signals from the equipment area of ​​the booster station. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. The multi-source sensing signals are subjected to spatiotemporal fusion processing to obtain a fused feature map; Fire evolution feature vectors are extracted based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; Fire risk level is determined based on the fire evolution feature vector, and graded alarm commands are generated based on the fire risk level determination. The response intensity of the graded alarm commands is positively correlated with the fire risk level determination result.

2. The fire alarm linkage monitoring method for wind farm substations according to claim 1, characterized in that, The multi-source sensing signals are subjected to spatiotemporal fusion processing to obtain a fused feature map, including: The sampling times of the infrared thermal radiation signal, visible light image signal, gaseous suspended particulate matter concentration signal and electrical insulation status signal are aligned to establish a multi-channel data frame with a unified time reference. Calculate the geometric registration offset of each channel's data frame in the spatial coordinate system, perform pixel-level spatial resampling based on the geometric registration offset, and generate a spatiotemporally aligned multi-source observation matrix; Based on the multi-source observation matrix, feature correlation mapping between channels is performed to construct the fused feature map, which includes a temperature-smoke joint distribution layer, an electrical-thermal field coupling layer, and a visible light texture verification layer.

3. The fire alarm linkage monitoring method for wind farm substations according to claim 1, characterized in that, Based on the fused feature map, fire evolution feature vectors are extracted, including: A dynamic tracking window is set up in the fused feature map to monitor the cumulative growth trend of the temperature field diffusion rate within a preset time segment; Extract the spatial distribution pattern of the smoke concentration gradient change rate, and identify the dominant direction of smoke diffusion and the topological structure of the diffusion front; The high-frequency pulse component in the electrical insulation status signal is monitored for synchronization correlation with the infrared thermal radiation signal. When the synchronization correlation exceeds a preset coupling threshold, it is marked as an electrical thermal runaway associated region. The cumulative growth trend, the spatial distribution pattern, and the synchronous correlation are vectorized to construct the fire evolution feature vector.

4. The fire alarm linkage monitoring method for wind farm substations according to claim 1, characterized in that, Determining the fire risk level based on the fire evolution characteristic vector includes: The parameters of each dimension in the fire evolution feature vector are matched with the preset combustion type discrimination boundary, which includes the electrical equipment overheating type boundary, the combustible material smoldering type boundary and the mixed combustion type boundary; The combustion type is determined based on the matching results; Based on the combustion type classification and the magnitude of the fire evolution characteristic vector, the fire risk level determination is generated by mapping. The fire risk level determination includes the initial risk level, the diffusion risk level, and the critical risk level.

5. The fire alarm linkage monitoring method for wind farm substations according to claim 4, characterized in that, Based on the matching results, the combustion type corresponding to the current fire evolution characteristics is determined, including: When the matching degree between the fire evolution feature vector and any combustion type discrimination boundary is lower than the preset confidence threshold, it is determined as an unconfirmed suspected event; For the unconfirmed suspected events, a multi-condition cross-validation mechanism is initiated, which includes: verifying the continuous upward trend of the concentration signal of the gaseous suspended particulate matter, verifying the persistence of texture abrupt changes in the visible light image signal, and verifying the transmission consistency of infrared thermal radiation signals in adjacent areas. When the output of the multi-condition cross-validation mechanism meets the preset confirmation conditions, the unconfirmed suspected event is confirmed as a real fire event and the fire risk level is upgraded. If the preset confirmation conditions are not met, suppress the generation of the current alarm command and mark it as an environmental interference event.

6. The fire alarm linkage monitoring method for wind farm booster stations according to claim 5, characterized in that, After suppressing the generation of the current alarm command and marking it as an environmental interference event, the following also includes: The frequency of occurrence of the aforementioned environmental disturbance events was statistically analyzed.

7. The fire alarm linkage monitoring method for wind farm substations according to claim 6, characterized in that, Also includes: When the frequency of occurrence exceeds a preset interference threshold, the sensitivity parameter of the combustion type discrimination boundary is adjusted according to the multidimensional transient characteristics of the environmental interference event. The sensitivity parameter includes the trigger threshold of the temperature field diffusion rate and the response coefficient of the smoke concentration gradient change rate. The feature extraction weights of the fused feature map are updated based on the adjusted sensitivity parameters, reducing the weight ratio of the high-frequency interference channel and enhancing the response gain of the stable feature channel.

8. The fire alarm linkage monitoring method for wind farm substations according to claim 1, characterized in that, Based on the fire risk level determination, a graded alarm command is generated, including: When the initial risk level is determined, an equipment inspection early warning command and an enhanced monitoring frequency command are generated, triggering continuous acquisition of visible light images of the local area. When the risk level of diffusion is determined, an audible and visual alarm activation command and a fire pre-activation preparation command are generated to activate the pre-pressurization state of the automatic fire extinguishing device. When the critical risk level is determined, an emergency evacuation command for the entire station, a power cut-off interlock command, and an automatic fire extinguishing spray command are generated to cut off the electrical connection of the fault area and start the smoke exhaust of the entire area.

9. The fire alarm linkage monitoring method for wind farm substations according to claim 8, characterized in that, Generates station-wide emergency evacuation commands, power cut-off interlock commands, and automatic fire suppression spray commands; cuts off electrical connections in the fault area and initiates full-area smoke extraction, including: Based on the spatial distribution pattern in the fire evolution characteristic vector, the main combustion area and smoke diffusion path are located; Based on the main combustion area and the smoke diffusion path, calculate the optimal evacuation route and smoke avoidance passage for personnel, and embed the optimal route into the station-wide emergency evacuation command; Based on the distribution of the electrical thermal runaway associated areas, the order of electrical disconnection is determined, prioritizing the disconnection of upstream switching equipment electrically connected to the main combustion area, while preserving the power supply circuits for the monitoring system and emergency lighting; Based on the wind direction compensation coefficient of the smoke diffusion path, the spray angle and coverage of the automatic fire extinguishing spray command are adjusted to ensure that the fire extinguishing medium covers the leading edge of the smoke diffusion path.

10. A fire alarm linkage monitoring system for a wind farm substation, applied to the fire alarm linkage monitoring method for a wind farm substation as described in any one of claims 1-9, characterized in that, include: The signal acquisition module is used to acquire multi-source sensing signals from the equipment area of ​​the substation. The multi-source sensing signals include infrared thermal radiation signals, visible light image signals, gaseous suspended particulate matter concentration signals, and electrical insulation status signals. The map fusion module is used to perform spatiotemporal domain fusion processing on the multi-source sensing signals to obtain a fused feature map; The vector construction module is used to extract fire evolution feature vectors based on the fused feature map, wherein the fire evolution feature vectors characterize the joint distribution of temperature field diffusion rate, smoke concentration gradient change rate, and abnormal fluctuation amplitude of electrical parameters; The fire monitoring module is used to determine the fire risk level based on the fire evolution feature vector, and to generate graded alarm commands based on the fire risk level determination. The response intensity of the graded alarm commands is positively correlated with the fire risk level determination result.

Images