Acousto-optic-thermal combined sensing forest fire real-time monitoring and early warning device and method

Abstract

The invention discloses an acousto-optic-thermal combined sensing forest fire real-time monitoring and early warning device and method, and relates to the technical field of fire monitoring and early warning, and the method comprises the steps: monitoring forest state information in real time through acousto-optic-thermal combined sensing equipment on a distributed control tower, extracting the state information of a certain control tower point position, and inputting the state information into a fire research and judgment device; and obtaining a fire research and judgment result. And when the result shows that a fire sign (according with a predetermined condition) exists, starting a support verification mechanism to calculate the comprehensive support degree. If the support degree reaches a preset threshold, an early warning signal is activated, and fire early warning is given out to the control tower point location. The technical problem that the early fire is difficult to recognize timely and accurately due to the fact that the existing forest fire monitoring and early warning technology is insufficient in the aspects of single data dimension and result credibility improvement is solved, and the technical effect of recognizing the early fire more timely and accurately is achieved by enriching the data dimension and enhancing the result credibility.

Description

Technical Field

[0001] The present application relates to the technical field of fire monitoring and early warning, and in particular to a real-time forest fire monitoring and early warning device and method using combined acoustic, optical and thermal sensing. Background Art

[0002] With global climate change and the increase in extreme weather events, the frequency and severity of forest fires are increasing. Traditional forest fire monitoring mostly relies on a single type of sensor or manual patrols. This method is difficult to capture early fire characteristics in an environment with complex terrain and dense vegetation, resulting in increasingly prominent problems of information lag and inaccurate identification. At the same time, the forest environment is changeable and the fire signals are unevenly distributed in space and time, which increases the difficulty of monitoring and early warning. Therefore, with the increasing demand for fire prevention, how to effectively deal with monitoring difficulties and uncertainties and reduce the interference of environmental factors on data collection and analysis has become a challenge that needs to be solved urgently.

[0003] Among the current relevant technologies, forest fire monitoring and early warning technologies have shortcomings in terms of single data dimension and lack of credibility in results, which leads to technical problems such as difficulty in timely and accurate identification of early fire conditions. Summary of the invention

[0004] This application solves the technical problem that the existing forest fire monitoring and early warning technology has shortcomings in terms of single data dimension and improvement of result credibility, which makes it difficult to identify early fire conditions in a timely and accurate manner by providing a real-time forest fire monitoring and early warning device and method with combined acoustic, optical and thermal perception.

[0005] This application provides a forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing, including:

[0006] A target real-time status information monitoring module, the target real-time status information monitoring module is used to obtain the target real-time status information of the target forest through dynamic monitoring by an acoustic, optical and thermal joint sensing device, wherein the acoustic, optical and thermal joint sensing device is mounted on a distributed tower of the target forest; a first status information extraction module, the first status information extraction module is used to extract the first status information corresponding to the first tower point in the target real-time status information, the first tower point refers to any one of the distributed towers; an output data acquisition module, the output data acquisition module is used to use the first status information as input data of a fire analysis device, and through the fire analysis device Output data is obtained, wherein the output data includes a first fire assessment result of the first tower point; a predetermined support verification mechanism reading module, wherein the predetermined support verification mechanism reading module is used to read a predetermined support verification mechanism when the first fire assessment result meets the predetermined verification constraint; a first comprehensive support acquisition module, wherein the first comprehensive support acquisition module is used to obtain a first comprehensive support of the first fire assessment result according to the predetermined support verification mechanism; and a fire warning module, wherein the fire warning module is used to activate a warning signal to issue a fire warning to the first tower point of the target forest when the first comprehensive support reaches a predetermined support threshold.

[0007] This application provides a real-time forest fire monitoring and early warning method based on combined acoustic, optical and thermal perception, including:

[0008] The target real-time status information of the target forest is obtained by dynamic monitoring using an acoustic, optical and thermal combined sensing device, wherein the acoustic, optical and thermal combined sensing device is mounted on a distributed tower of the target forest; the first status information corresponding to the first tower point in the target real-time status information is extracted, and the first tower point refers to any one of the distributed towers; the first status information is used as input data of a fire analysis device, and output data is obtained through the fire analysis device, wherein the output data includes a first fire analysis result of the first tower point; when the first fire analysis result meets a predetermined verification constraint, a predetermined support verification mechanism is read; according to the predetermined support verification mechanism, a first comprehensive support degree of the first fire analysis result is obtained; when the first comprehensive support degree reaches a predetermined support degree threshold, an early warning signal is activated to issue a fire warning to the first tower point of the target forest.

[0009] The real-time monitoring and early warning device and method for forest fires using the combined acoustic, optical and thermal sensing proposed in this application first monitors forest status information in real time through the combined acoustic, optical and thermal sensing equipment on the distributed tower, extracts status information of a certain tower point and inputs it into the fire analysis device to obtain the fire analysis result. When the result shows signs of fire (meets the predetermined conditions), the support verification mechanism is enabled to calculate the comprehensive support. If the support reaches the preset threshold, the warning signal is activated and a fire warning is issued for the tower point. By enriching the data dimension and enhancing the credibility of the results, the technical effect of more timely and accurate identification of early fire conditions is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the device according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed precisely in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0011] Figure 1 A schematic diagram of the structure of a forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing provided in an embodiment of the present application;

[0012] Figure 2 A flow chart of a method for real-time monitoring and early warning of forest fires using combined acoustic, optical and thermal sensing provided in an embodiment of the present application.

[0013] Explanation of the accompanying drawings: target real-time status information monitoring module 10, first status information extraction module 20, output data acquisition module 30, predetermined support verification mechanism reading module 40, first comprehensive support acquisition module 50, fire warning module 60. DETAILED DESCRIPTION

[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0015] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0016] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, device, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0017] The present application embodiment provides a forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing, such as Figure 1 As shown, the device comprises:

[0018] The target real-time status information monitoring module 10 is used to obtain the target real-time status information of the target forest through dynamic monitoring of the acoustic, optical and thermal joint sensing device, wherein the acoustic, optical and thermal joint sensing device is mounted on the distributed tower of the target forest. Specifically, in the target forest fire monitoring system, the target real-time status information monitoring module 10 achieves the dynamic monitoring task based on the acoustic, optical and thermal joint sensing device distributed in the tower. The process is: first determine the tower site and place the equipment according to the terrain, vegetation, meteorology and fire history data, and calibrate the acoustic, optical and temperature sensors. During acoustic monitoring, the high sampling rate acquisition signal is used to identify the sound through time-frequency analysis, distinguish the ambient sound from the fire-related abnormal sound according to the frequency range, extract the characteristic parameters to locate the sound source, and classify by machine learning; in terms of optical monitoring, infrared and visible light images are collected, and after pre-processing, the deep learning model is used to identify the visual features and fire parameters such as smoke, flames, and vegetation discoloration; temperature monitoring is a matrix acquisition to construct a three-dimensional temperature field, integrate meteorological elements to simulate heat exchange, locate abnormal areas, set threshold warnings, generate data sets and visualize. The back-end data center receives multi-source data, processes it based on the fusion algorithm after spatiotemporal registration, and combines it with the geographic information system to generate visual comprehensive status information, providing accurate data support for forest fire prevention and control, building a complete monitoring architecture, and improving fire warning efficiency.

[0019] The first state information extraction module 20 is used to extract the first state information corresponding to the first tower point in the target real-time state information, and the first tower point refers to any tower in the distributed tower. Specifically, the first state information extraction module 20 extracts the first tower point exclusive information from the target real-time state information. First, the received data is preprocessed to unify the acoustic, optical, and temperature data formats and remove noise. When extracting acoustic features, MFCC and LPC are used to extract spectral characteristics, and HMM is used to mine time series changes. The acoustic propagation model is used to compensate for attenuation and capture the acoustic characteristics of the fire; in terms of optical feature extraction, image segmentation and CNN are used to analyze the color change characteristics of smoke, flames and vegetation; temperature feature extraction relies on the time series analysis model to disassemble the rules and anomalies, and the temperature field distribution is calculated using a spatial interpolation algorithm. The thermal physical parameters are calculated according to the laws of thermodynamics to evaluate thermal hazards. Finally, after adaptive weighted fusion, the weights are dynamically adjusted according to the contribution of each feature, and then standardized and normalized to generate accurate first state information, providing a decision-making basis for fire analysis and enhancing the effectiveness of forest fire monitoring and early warning.

[0020] In a possible implementation, the first state information extraction module further includes: an acoustic, optical and thermal joint sensing device component unit, and the acoustic, optical and thermal joint sensing device component unit is used for the acoustic, optical and thermal joint sensing device to include an acoustic sensor, an optical sensor and a temperature sensor. Specifically, the acoustic, optical and thermal joint sensing device component unit integrates acoustic, optical and temperature sensors to construct a multimodal monitoring architecture. The acoustic sensor captures sound wave vibrations with a high-sensitivity microphone array or piezoelectric element based on the acoustic principle. Its design parameters are adapted to the complex acoustic environment of the forest and accurately perceive the frequency, amplitude and phase information of the sound wave; the optical sensor integrates infrared thermal imaging and visible light imaging technology, focuses on optical lenses, filters light in specific bands and senses changes in light intensity with detectors to achieve multi-spectral image acquisition, and accurately images based on the thermal radiation and optical reflection characteristics of the object; the temperature sensor uses thermistors, thermocouples or infrared temperature measuring probes to sense temperature based on the thermoelectric effect and thermistor characteristics, improves the measurement accuracy through calibration and compensation, and forms a three-dimensional thermal field monitoring network at multiple points around the tower. The three work together to provide all-round and multi-dimensional data support for forest status monitoring.

[0021] The first sound information monitoring unit is used to monitor and obtain the first sound information of the first tower point through the acoustic sensor. Specifically, the first sound information monitoring unit drives the acoustic sensor to carry out sound monitoring at the first tower point. The sensor converts the sound wave pressure signal into a digital audio sequence according to the preset sampling frequency and quantization accuracy standard to build a digital basis for the sound signal. The fast Fourier transform (FFT) technology is used to analyze the audio spectrum structure, accurately extract the frequency components and amplitude distribution characteristics, and use the time-frequency analysis method to capture the dynamic change characteristics of the signal over time, and obtain the characteristic pattern of potential abnormal acoustic events. The complex signal is decomposed by using signal processing methods such as wavelet transform and Hilbert-Huang transform, and transient and non-stationary characteristic components are extracted to effectively capture key information such as high-frequency bursts caused by burning trees and low-frequency interference sounds caused by airflow surging. According to the physical model of sound propagation, the signal loss caused by factors such as distance attenuation, environmental absorption and scattering is compensated, the acoustic parameters of the sound source are restored, and the first sound information is formed to provide key acoustic dimension data support for subsequent fire identification.

[0022] An image recognition and analysis unit is used to perform image recognition and analysis on the first infrared image of the first tower point through the optical sensor to obtain the first optical information. Specifically, the image recognition and analysis unit controls the optical sensor to collect the first infrared image based on the first tower point and performs deep image analysis. The image contrast and detail recognition are enhanced by using preprocessing techniques such as grayscale stretching and histogram equalization to optimize the image quality. A target detection model is constructed based on the deep learning convolutional neural network (CNN) architecture, and deep training and optimization are performed using a labeled fire image sample data set. The model automatically learns and accurately identifies the hazy texture features of smoke, the irregular shape features of flames, and the characteristic patterns of vegetation heat discoloration, accurately locates, classifies and frames the target area in the image, and outputs structured information such as target category probability, position coordinates and size parameters. The image segmentation algorithm is used to analyze the distribution gradient of smoke concentration, the structural details of the flame combustion layer, and the quantitative information of the proportion of damaged vegetation area, so as to obtain the fire information contained in the image in all directions. The multi-dimensional analysis results are integrated and quantified into the first optical information to provide image data support for accurate judgment of the fire situation.

[0023] The first temperature information acquisition unit is used to obtain the first temperature information of the first tower point through the temperature sensor monitoring. Specifically, the first temperature information acquisition unit monitors the dynamic changes of the temperature field in real time at the first tower point based on the temperature sensor network. The sensor measures the surface temperature of air, soil and vegetation according to the principle of thermal balance, and constructs a continuous temperature time series data set after analog-to-digital conversion and data filtering. Statistical analysis technology is used to obtain the characteristics of the change rules such as the mean, extreme value, and standard deviation of the temperature in the data, and the spatial distribution characteristics of the temperature field are analyzed according to the thermodynamic gradient theory. With the help of spatial interpolation methods such as Kriging interpolation and spline interpolation, the data gaps in the monitoring blind area are filled, and continuous temperature contour maps are drawn to accurately show the distribution pattern of heat flow direction and intensity. Combined with real-time meteorological data, coupled with heat exchange physical models, the relationship between environmental heat balance and heat conduction flux changes is quantified, the root cause of abnormal temperature fluctuations is found, and the first temperature information is extracted to form a key thermal data foundation for deep insight into fire thermal hazards.

[0024] A variation weighted processing unit is used to perform variation weighted processing on the first sound information, the first optical information and the first temperature information in sequence, and obtain a first sound coefficient, a first optical coefficient and a first temperature coefficient respectively. Specifically, the variation weighted processing unit performs weighted processing on the first sound information, the first optical information and the first temperature information. Based on the dynamic change of the importance of different information sources to fire characterization with time, space and environmental factors, a multi-dimensional comprehensive evaluation index system covering information recognition, reliability and fire relevance is constructed. In the dimension of information recognition, the signal-to-noise ratio and feature significance quantitative indicators are analyzed; in the reliability level, the sensor stability and data consistency evaluation parameters are analyzed; for the dimension of fire relevance, the degree of correlation between each information feature and key factors such as the fire spread rate and intensity change is deeply analyzed. Subsequently, the weight determination methods such as the hierarchical analysis method and the entropy weight method are used to calculate the weight coefficient of each information source according to the evaluation index system, and the weight configuration is dynamically adjusted and optimized according to the real-time monitoring data. For example, if the acoustic information sensitivity is high in the early stage of the fire, the weight of the sound information is appropriately increased, and if the optical features are critical in the fire development stage, the optical information coefficient is weighted. After weighted calculation and normalization processing, the first sound coefficient, the first optical coefficient and the first temperature coefficient are accurately generated, providing an information basis for accurate fire judgment.

[0025] The first state information composition unit is used for the first sound coefficient, the first optical coefficient and the first temperature coefficient to form the first state information. Specifically, the first state information composition unit fuses the first sound coefficient, the first optical coefficient and the first temperature coefficient. According to the predefined coefficient weight allocation combination rule, a multi-dimensional state vector is constructed, and each component of the vector accurately quantifies the contribution of acoustic, optical and temperature dimensional information to the fire situation. Standardization processing technology is used to unify the dimensions of information in each dimension, effectively eliminate the impact caused by magnitude differences, and systematically organize and integrate multivariate information to finally form accurate and normalized first state information. Provide data basis for accurate decision-making, improve the intelligent decision-making ability of forest fire monitoring system, and optimize the formulation and implementation of fire prevention and control strategies.

[0026] In a possible implementation, the image recognition and analysis unit further includes: a first infrared pixel acquisition subunit, which is used to acquire the first infrared pixel in the first infrared image through the optical sensor. Specifically, first interact with the hardware driver, technical manual and previous calibration data of the optical sensor to acquire its acquisition parameters for the first tower point, such as lens focal length, aperture size, detector sensitivity and infrared band filter characteristics. In this way, the sensor is driven to collect infrared images, the lens focuses on the forest area, and the detector photosensitive element is excited by infrared radiation to generate an electrical signal. The signal is low-noise amplified and filtered by the readout circuit to remove noise and distortion, and then converted into a digital signal by the analog-to-digital converter according to the preset sampling frequency and quantization accuracy, and a two-dimensional pixel matrix is ​​formed according to the pixel array rules to obtain the first infrared pixel. The whole process strictly follows the sensor's pixel resolution, response band and sensitivity indicators to ensure that the pixel accurately reflects the difference in infrared radiation intensity, effectively captures the fire-related bands, provides high-quality data for subsequent optical analysis, and builds a precise conversion channel from infrared radiation to digital images to meet the front-end data quality requirements of fire thermal radiation monitoring, and lays an accurate image information foundation for fire monitoring.

[0027] The first light color parameter acquisition subunit is used to analyze the first RGB corresponding to the first infrared pixel to obtain the first light color parameter. Specifically, after receiving the first infrared pixel data, the analysis is performed according to the radiation intensity-RGB conversion algorithm of the optical sensor. This algorithm comprehensively analyzes the spectral response characteristics of the detector, the optical constants of the material and the interference factors of the ambient light to ensure the accuracy of the conversion to the RGB color space. The high-precision spectral analysis technology is used to determine the proportion of the radiation intensity of each band of the pixel, and the three-color component values ​​are determined and the proportional relationship is calculated in combination with the calibrated color correction matrix. The color tendency characteristics are determined by algorithms such as the CIEDE2000 color difference formula, and the absolute values ​​of each component are calculated based on the absolute value of the pixel radiation intensity and the detector calibration parameters to construct the first light color parameter. This parameter contains the law of the interaction between the thermal radiation of the object, the optical characteristics of the material and the ambient light. For example, the pixels in the burning area of ​​the forest fire have a unique color combination in the RGB space, which provides a key color basis for identifying the optical characteristics of the fire, improves the ability to extract fire-related characteristics, and provides color dimension information support for judging the fire situation.

[0028] The first light intensity parameter acquisition subunit is used to analyze the first infrared pixel corresponding to the first HSB to obtain the first light intensity parameter. Specifically, the light intensity analysis of the first infrared pixel is performed based on the HSB color model. First, according to the energy response curve of the detector calibrated by the blackbody radiation standard source and the Planck blackbody radiation law formula, the pixel radiation energy is accurately converted into the HSB brightness value. Based on the fact that the contrast of the local area affects the visual perception of light intensity, the contrast is measured using algorithms such as the grayscale co-occurrence matrix and the edge detection operator, and a contrast-brightness dynamic correction model is constructed. According to the contrast change, a brightness perception weight adjustment function is set to accurately correct the brightness value. At the same time, the saturation-brightness weighting function is determined by experimental data fitting and intelligent model optimization, and the brightness value is further corrected to obtain an accurate first light intensity parameter. The parameters can accurately capture the absolute value and dynamic changes of light intensity. In forest fire monitoring, they can monitor the fluctuations in flame radiation intensity and the attenuation gradient of smoke diffusion light intensity in real time, provide key light intensity quantitative parameters for fire monitoring, and help to judge the fire and smoke spread trends from the perspective of light intensity, improve the fire optical monitoring and evaluation system, enhance the data's ability to characterize fire evolution, provide a reliable basis for fire decision-making, and enhance the scientificity and effectiveness of monitoring.

[0029] The first optical information composition subunit is used for the first light color parameter and the first light intensity parameter to form the first optical information. Specifically, the first optical information composition subunit organically integrates the first light color parameter and the first light intensity parameter to construct a complete first optical information. The parameters are organized in the form of structured vectors, and the vector elements are light color parameters (covering RGB ratio values ​​and absolute values ​​of each component) and light intensity parameters (including HSB brightness values ​​and correction values ​​obtained through a complex correction process). In order to ensure that the optical information corresponding to different pixels has high consistency and comparability, the Z-score normalization method is used to unify the dimensions of the parameters based on the general theory and method of data standardization. The Z-score normalization method is based on statistical principles. By calculating the deviation of each parameter value from the mean and dividing it by the standard deviation, the parameter value is converted into a value under the standard normal distribution. After large-scale sample test verification and in-depth theoretical derivation and optimization, it can effectively eliminate the influence of factors such as pixel acquisition device performance differences, complex environmental factors interference, and signal transmission noise on the accuracy and comparability of optical information, ensuring that all optical information can be accurately measured and compared under the same standard scale. The vector fully encapsulates the full-dimensional optical properties of the pixel, reflecting the fire optical characteristics of the first infrared pixel from the dual core perspectives of color quality vividness and light intensity level accuracy.

[0030] The output data acquisition module 30 is used to use the first state information as the input data of the fire analysis device, and obtain output data through the fire analysis device, wherein the output data includes the first fire analysis result of the first tower point. Specifically, the output data acquisition module 30 receives the first state information from the first state information extraction module 20, checks and standardizes its format, and inputs it into the fire analysis device. The fire analysis device loads the fire discrimination model trained based on historical data, and the input first state information is subjected to feature extraction, including acoustic spectrum and time-frequency features, optical image features, temperature field features, etc., and is normalized and weighted to form a comprehensive feature vector. According to the model's judgment logic decision, the probability of fire occurrence is calculated and compared with the threshold, and the result of "fire" or "no fire" is determined. At the same time, the corresponding auxiliary information is generated. Finally, the output data acquisition module 30 packages and transmits the data including the first fire analysis result and auxiliary information to the subsequent module according to the protocol, so as to improve the coordination and response efficiency of the forest fire prevention and control system and ensure forest safety.

[0031] In a possible implementation, the output data acquisition module 30 further includes: a forest fire log acquisition unit, the forest fire log acquisition unit is used to acquire a forest fire log, and the forest fire log includes a first record and a second record. Specifically, the forest fire log acquisition unit interacts with data sources such as the forestry management department database, the fire monitoring station historical record library, and field survey archives to collect fire logs in specific forest areas, covering long-term fire-related information, including a first record and a second record. The first record focuses on detailed status information of the forest during past fire periods, such as meteorology, vegetation, topography, and multi-source data collected by various monitoring equipment, forming the first historical status information, and annotating the first historical fire label; the second record focuses on recording the state of the forest during periods when no fire occurred, including meteorology, vegetation, topography, and monitoring data, etc., constituting the second historical status information and the second historical no-fire label.

[0032] The first data group forming unit is used to form a first data group according to the first historical state information and the first historical fire label in the first record. Specifically, after the first data group forming unit obtains the first historical state information and the first historical fire label in the first record, it performs the forming operation. First, the first historical state information is preprocessed: the meteorological data is cleaned to remove outliers, interpolated to supplement missing values, and smoothed to reduce noise; the vegetation coverage data is corrected by cross-validation of satellite remote sensing and field surveys; the topographic data is digitized and modeled by GIS technology; and the multi-source monitoring data is subjected to signal enhancement, feature extraction and normalization. After the preprocessing is completed, the processed information is paired with the fire label to form the first data group, and each sample contains the comprehensive forest status information and label during the fire period.

[0033] The second data group forming unit is used to form a second data group according to the second historical state information and the second historical no-fire label in the second record. Specifically, the second data group forming unit forms a data group according to the second historical state information and the second historical no-fire label of the second record. Its preprocessing steps are similar to those of the first data group: meteorological data is ensured in the same way; vegetation coverage data is corrected with the help of satellite remote sensing and field surveys; topographic data is improved with GIS technology; multi-source monitoring data is processed to be comparable. After preprocessing, the information is paired with the label to form a second data group. The samples represent the state of the forest when no fire occurs and the no-fire mark, providing high-quality negative samples for the fire analysis device. It helps the model learn non-fire feature patterns, accurately distinguish whether there is a fire or not, reduce the false alarm rate, provide comprehensive data support for the precise prevention and control of forest fires, and ensure the safety of forest resources and ecology, so that it can accurately judge fires in complex forest environments and prevent and control them in a timely manner.

[0034] A fire analysis device acquisition unit is used to perform supervised learning on the first data group and the second data group after sample enhancement based on the principle of support vector machine to obtain the fire analysis device. Specifically, the fire analysis device acquisition unit uses the pre-processed first and second data groups to construct a fire analysis device based on the principle of support vector machine. Sample features are mined using feature engineering technology: meteorological data calculates interaction and time series features; vegetation coverage data extracts multiple features; terrain data calculates complexity and coupling features; multi-source monitoring data extracts time-frequency domain features to enhance separability. Then, sample enhancement technology is used to expand the data set through random perturbation and data synthesis to prevent overfitting. Subsequently, the enhanced positive and negative sample data groups are input into the support vector machine model for supervised learning, and the hyperplane is optimized by adjusting the kernel function type, penalty parameters, etc. After multiple rounds of iterations, the training is stopped when the performance of the verification set reaches the optimal level to obtain the fire analysis device. The analysis device can accurately judge the possibility of fire based on forest status information, provide support for prevention and control decisions, improve the scientificity and effectiveness of forest fire prevention and control, and play a key role in forest ecosystems.

[0035] The predetermined support verification mechanism reading module 40 is used to read the predetermined support verification mechanism when the first fire analysis result meets the predetermined verification constraint. Specifically, after the first fire analysis result is generated, the predetermined support verification mechanism reading module 40 performs a verification process to monitor whether the result meets the preset "fire" verification constraint. If it meets the requirements, it will interact with a database or storage system that specifically stores the predetermined support verification mechanism to read the mechanism content completely and accurately according to the established procedures and protocols. The mechanism covers multiple levels of verification, including re-analysis of the original data of acoustic, optical, and temperature sensors, review of the algorithm parameters and model structure of the fire analysis device, review of the logical reasoning of the analysis, and introduction of external auxiliary information (such as data from surrounding meteorological stations, records of human activities and lightning activities in the forest, etc.) for cross-comparison. By implementing this mechanism, the risk of false alarms can be effectively reduced, the accuracy and reliability of fire judgment can be improved, a reliable decision-making basis can be provided for subsequent emergency responses, fire losses can be reduced, and the safety and stability of forest ecology can be guaranteed.

[0036] The first comprehensive support acquisition module 50 is used to obtain the first comprehensive support of the first fire assessment result according to the predetermined support verification mechanism. Specifically, after the predetermined support verification mechanism is read, the first comprehensive support acquisition module 50 first organizes the verification dimensions of the mechanism. In the data integrity and accuracy verification, the acoustic, optical, and temperature sensor data are processed and analyzed separately, compared with the corresponding feature library, and the support score S1 is calculated according to the weight; the model algorithm reliability verification dimension is to evaluate the accuracy and stability indicators by running the model through a simulated data set, and the support score S2 is given in combination with the historical performance; the logical reasoning rationality verification dimension reconstructs the logical deduction steps of the assessment, and summarizes the rationality scores of each link to obtain S3; the external auxiliary information verification dimension integrates information such as meteorology, human activities, and lightning activities, and weighted calculations to obtain S4. Finally, the first comprehensive support is obtained by summing up S1, S2, S3, and S4 according to their weights (w1, w2, w3, w4 and the sum is 1, determined according to the importance and reliability of each dimension), which provides a quantitative decision-making basis for subsequent fire judgment and measures, and improves the reliability and effectiveness of prevention and control decisions.

[0037] In a possible implementation, the first comprehensive support acquisition module 50 further includes: a first support verification mechanism extraction unit, which is used to extract the first support verification mechanism in the predetermined support verification mechanism. Specifically, after receiving the predetermined support verification mechanism, the first support verification mechanism extraction unit performs an extraction process. The mechanism is usually a complex, multi-dimensional set of rules and processes, which is intended to verify the fire assessment results. According to the preset classification standard and index system, the first support verification mechanism is identified and extracted from the predetermined support verification mechanism. The mechanism is based on a specific verification direction, such as a verification dimension based on geographic spatial relationships. Its core is to use the distribution characteristics of tower points in the forest and the geographic spatial laws of fire spread to construct a set of targeted verification rules, so that the information of the surrounding tower points can be used to further verify or correct the fire assessment results of the first tower point, so as to provide a key rule basis for improving the accuracy of fire determination, ensure that the verification process has a clear direction and focus, and avoid invalid verification operations without purpose, so as to efficiently use limited resources and data and improve the reliability and stability of the entire fire monitoring system.

[0038] The first neighborhood determination unit is used to determine the first neighborhood of the first tower point according to the first support verification mechanism, and the first neighborhood includes multiple tower points. Specifically, the first neighborhood determination unit determines the first neighborhood of the first tower point after obtaining the first support verification mechanism. The determination method is based on factors such as geographical distance, topography, similarity of vegetation cover types or range of fire propagation risk: when geographical distance is the main factor, a radius is set to define the neighborhood; considering topography, points in the same valley, hillside or similar altitude slope are given priority; based on vegetation similarity, points with similar vegetation community distribution are selected. The neighborhood set constructed in this way is closely related to the first tower point, providing a targeted sample space for subsequent information collection and analysis, so that the surrounding point information can more effectively verify the fire assessment results of the first tower point, and enhance the persuasiveness and credibility of the verification.

[0039] An arbitrary tower point extraction unit, the arbitrary tower point extraction unit is used to randomly extract any tower point from the multiple tower points, and the arbitrary tower point corresponds to arbitrary status information. Specifically, after the arbitrary tower point extraction unit clarifies the multiple tower points in the first neighborhood, it uses a sampling method based on a random number generation algorithm to extract an arbitrary tower point. This algorithm generates a random index value within the neighborhood point set, and selects points accordingly to ensure that each point has the same probability of being drawn, avoids human interference with fixed patterns, and ensures that the sampling is random and representative. During multiple verifications, different points can be drawn in different rounds, and diverse status information can be collected to fully reflect the overall situation of the neighborhood, providing rich comparative data for the analysis of fire analysis results, which is helpful to discover abnormal problems and improve the accuracy and reliability of analysis.

[0040] Any fire assessment result analysis unit, the said any fire assessment result analysis unit is used to analyze the said any state information through the said fire assessment device to obtain any fire assessment result. Specifically, after extracting any tower point, the any fire assessment result analysis unit obtains its any state information and inputs it into the fire assessment device. The assessment device analyzes the spectrum of acoustic data and extracts time-frequency domain features based on the pre-trained model algorithm, and determines the possibility of fire based on the acoustic model; uses image recognition and visual algorithms for optical data, combined with the feature library, to judge the signs of fire based on smoke and flame characteristics; and uses heat propagation theory for temperature data to infer the possibility of fire based on its value, change trend, and gradient. After comprehensive analysis of multi-source data and model calculation, the assessment result of "fire" or "no fire" is output, which provides a key basis for the calculation of the first support degree, verifies the assessment result of the first tower point from the perspective of surrounding points, and improves the intelligent analysis and decision-making capabilities of the monitoring system.

[0041] The first support acquisition unit is used to obtain the first support of the first fire assessment result according to the arbitrary fire assessment result. Specifically, the first support acquisition unit calculates the first support of the first fire assessment result after obtaining the arbitrary fire assessment result. First, a support evaluation model is constructed to consider multiple factors for quantification. If the two results are consistent (both are "there is a fire"), points are added dynamically according to the pre-set scoring rules, combined with the correlation between the surrounding points and the first tower point (such as close geographical distance, similar environmental characteristics, good data acquisition equipment) and data reliability factors; if they are inconsistent (one is "there is a fire" and the other is "no fire"), points are deducted according to the difference deduction rules, also based on the above factors. The specific value is calculated through this dynamic scoring mechanism, which intuitively reflects the degree of support of the judgment result of any tower point for the first fire assessment result, and provides a quantitative basis for evaluating its reliability.

[0042] The first comprehensive support acquisition unit is used to use the first support as the first comprehensive support. Specifically, the first comprehensive support acquisition unit outputs the first support as the first comprehensive support, and will first perform preprocessing and standardization operations on it. If the calculation scale and unit of the first support are different, it will be converted into a unified standard scale to facilitate comparison and analysis with other support indicators; at the same time, its numerical range is checked and corrected to make it consistent with the expectations and logic of the evaluation system. After processing, the first comprehensive support can accurately reflect the comprehensive support level for the first fire assessment result based on the first support verification mechanism and the surrounding tower point information, and become an intuitive and reliable quantitative indicator for subsequent decision-making. Decision makers can judge the credibility of the first fire assessment result based on its numerical value and decide whether to take further action, such as launching a higher-level monitoring program, deploying fire-fighting resources, etc., so as to improve the efficiency and pertinence of forest fire prevention and control, minimize fire losses, and maintain forest ecological stability and safety.

[0043] In a possible implementation, the first comprehensive support acquisition unit further includes: a first mapping relationship forming subunit, which is used to form a first mapping relationship between the first tower point and the first fire assessment result. Specifically, the first mapping relationship forming unit constructs a first mapping relationship between the first tower point and the first fire assessment result. First, accurately locate and identify the first tower point, and clarify its key information such as geographical location, altitude, surrounding topography, covered forest area and vegetation type; then analyze the details of the first fire assessment result, such as the possibility of fire, fire level, spread direction and confidence index. Then, using the database management system, the geographic space of the point is correspondingly bound to the environmental feature information and fire-related information to form a two-way mapping relationship. Through this relationship, information can be queried in both directions, providing a clear and efficient information association basis for fire monitoring, analysis and decision-making, making information flow smoother and more accurate, which is conducive to understanding the fire situation, helping to take prevention and control measures in a timely manner, and enhancing the pertinence and timeliness of forest fire prevention and control.

[0044] The second mapping relationship forming subunit is used to form a second mapping relationship between the arbitrary tower point and the arbitrary fire assessment result. Specifically, the second mapping relationship forming unit constructs the second mapping relationship, and the process is similar to the formation of the first mapping relationship. First, the detailed information of any tower point is collected, covering the geographical coordinates, surrounding environment characteristics, data acquisition equipment performance parameters, working status, etc., to fully grasp its basic information and monitoring capabilities. Then, the arbitrary fire assessment results are deeply analyzed, including the assessment results, risk factors and corresponding quantitative evaluation indicators. The data association tool is used to closely associate the detailed information of any tower point with any fire assessment result to establish a second mapping relationship. This relationship records the fire assessment situation of each point, provides rich materials for comprehensive analysis, and compares the mapping relationships of different points to discover the spatial propagation trend of fire, regional risk differences and monitoring data, which helps to have a more comprehensive and in-depth understanding of the target forest fire situation.

[0045] The visual map construction subunit is used to construct the target forest fire judgment visual map of the target forest according to the first mapping relationship and the second mapping relationship. Specifically, the visual map construction unit constructs the target forest fire judgment visual map according to the first mapping relationship and the second mapping relationship. First, the GIS data of the target forest is imported into the visualization software platform, and a geographic space framework containing layer information such as terrain, rivers, roads, and vegetation is constructed as the background support for the display of fire judgment information. Then, according to the first mapping relationship, the first tower point is marked at the corresponding position with a specific icon, and the relevant information of its fire judgment result is displayed by means of color, size, flashing, etc., such as red represents "fire" and the depth indicates the size of the fire. According to the second mapping relationship, multiple arbitrary tower points and their judgment results are also marked on the map in a similar visual way to form a comprehensive display layer. In addition, an interactive function is added to the visual map. Clicking the icon can pop up a detailed information window, zooming in and out with the mouse wheel can see the details of different areas, and dragging the map can browse the overall overview. By constructing this visual graph, complex fire analysis data can be presented intuitively, making it convenient for relevant personnel to quickly understand the fire distribution, development trends and tower point monitoring status, providing efficient and accurate information for fire decision-making, and improving the visual management level and decision-making efficiency of forest fire prevention and control.

[0046] In a possible implementation, the visual map construction unit further includes: an initial forest fire analysis and judgment visual map construction subunit, and the initial forest fire analysis and judgment visual map construction subunit is used to construct an initial forest fire analysis and judgment visual map according to the first mapping relationship and the second mapping relationship. Specifically, the initial forest fire analysis and judgment visual map construction subunit constructs an initial forest fire analysis and judgment visual map based on the first mapping relationship and the second mapping relationship. First, the GIS data of the target forest (basic geographic information such as terrain, rivers, roads, and forest coverage) is obtained and imported into the visualization software platform to build a geographic space framework. Then, based on the first mapping relationship, the first tower point is presented on the map with a specific icon, and the corresponding first fire analysis and judgment result information is associated, and the relevant content is displayed with the help of color depth, icon size, flashing effect, etc. Then, according to the second mapping relationship, any tower point and its analysis and judgment results are marked in a similar manner, and the information layer is superimposed to construct a visual map. The visual graph integrates the correlation information of the tower points and the fire analysis results, providing a visual platform for blank point analysis and fire situation assessment, which helps managers to intuitively understand the situation, identify risk areas, and assist in formulating prevention and control strategies, thereby improving the efficiency and accuracy of fire monitoring and response.

[0047] The blank point acquisition subunit is used to acquire the blank points in the initial forest fire assessment visual map and to form an interpolation reference set for the blank points. Specifically, the blank point acquisition subunit scans and analyzes the constructed initial forest fire assessment visual map, and accurately identifies the areas in the map that are not covered by the tower, i.e., the blank points, through image recognition and data analysis technology. Blank points lack direct fire monitoring data due to geographical restrictions, tower construction planning or other factors. For each blank point, the surrounding geographical environment information is collected, such as the relative position with the adjacent tower, terrain features (valleys, ridges, plains, etc.), vegetation types (coniferous forests, broad-leaved forests, bushes, etc.) and meteorological conditions (wind direction, wind speed, temperature, humidity, etc.), and the blank points with similar characteristics are classified and grouped, and combined with the tower point information with monitoring data in the surrounding area, an interpolation reference set for each blank point is formed. The construction of the interpolation reference set aims to provide multi-source and relevant reference data for the estimation of subsequent blank point fire assessment results, make full use of the known fire monitoring information in the surrounding area to infer the possibility of fire in the blank area, and improve the accuracy and reliability of fire situation judgment at blank points by analyzing multiple factors, fill the blank areas of fire monitoring, make the fire monitoring of the entire forest more comprehensive and complete, reduce the risk of fire spread due to monitoring blind spots and not being discovered in time, and enhance the comprehensiveness and effectiveness of forest fire prevention and control.

[0048] The first reference fire judgment result extraction subunit is used to extract the first reference fire judgment result of the first reference point in the interpolation reference set. Specifically, the first reference fire judgment result extraction subunit operates for the interpolation reference set of each blank point. The first reference point is selected from the interpolation reference set. The point is usually a tower point that is relatively close to the blank point in geographical space and whose fire monitoring data has high reliability and representativeness. By interacting with the database that stores the fire judgment results, the first reference fire judgment result of the first reference point is extracted. The result includes the fire judgment made by the point based on its monitoring data (such as fire, no fire) and related auxiliary information, such as the estimated value of the probability of fire occurrence, the assessment of the fire development trend, the analysis of possible fire causes (such as man-made fire source, natural fire source, etc.) and other details. After the information is extracted, it is used as the key basic data for estimating the subsequent fire assessment results of blank points. By analyzing the fire situation at the first reference point and combining its spatial relationship and environmental similarity with the blank point, the fire possibility of the blank point is inferred, providing a direct reference basis for the fire assessment of the blank point. In the absence of direct monitoring data, it is possible to use the effective information in the surrounding area to make reasonable speculations on the fire situation, improve the ability to grasp the entire forest fire situation, ensure the continuity and integrity of fire monitoring and prevention and control work, and reduce fire monitoring loopholes and misjudgment risks due to missing data.

[0049] The first spatial distance acquisition subunit is used to obtain the first spatial distance between the first reference point and the blank point. Specifically, after determining the first reference point and the blank point, the first spatial distance acquisition subunit uses a geographic spatial analysis algorithm to calculate the first spatial distance between the two. The algorithm is based on the principles of geodesy and the geographic coordinate system, accurately measures the straight-line distance between the two points, and makes corrections considering the influence of terrain undulations on the actual distance. By obtaining high-precision geographic information data, including a terrain elevation model (DEM), the actual surface distance between the two points is calculated, rather than a simple plane straight-line distance, to ensure the accuracy and reliability of the distance calculation. The first spatial distance value is one of the important bases for the subsequent sorting and weight allocation of reference points, because the closer the spatial distance, the greater the influence of the fire assessment results of the reference point on the blank point, and the higher its relevance and reference value. Accurate calculation of spatial distance can reasonably reflect the closeness between different reference points and blank points, thereby providing precise spatial relationship data support for building a scientific and reasonable blank point fire assessment model. When using surrounding reference point information to infer the fire situation at blank points, the influence of spatial factors can be considered more accurately, thereby improving the accuracy and credibility of fire assessment results and providing strong technical support for the precise prevention and control of forest fires.

[0050] A reference descending list acquisition subunit is used to obtain a reference descending list of the first reference fire in descending order based on the first spatial distance. Specifically, the reference descending list acquisition subunit uses the first spatial distance as the key sorting index to perform a sorting operation on all reference points in the interpolation reference set. First, the spatial distance between each reference point and the blank point is obtained, and these reference points are arranged in order from small to large. The sorted reference point list is reversed to obtain a descending list based on the first spatial distance, that is, a reference descending list. In this list, the reference point closest to the blank point is ranked first, and the one with the farthest distance is ranked last. This sorting method enables the fire assessment results of reference points with closer distances and higher relevance to be given priority when determining the fire assessment results of the blank point in the subsequent determination, giving them greater weight and influence, while reference points with farther distances are relatively secondary. Through descending arrangement based on spatial distance, a reasonable reference point weight distribution system was constructed, so that the fire assessment results of blank points can more scientifically and accurately integrate the information of multiple reference points in the surrounding area, make full use of spatial correlation to improve the reliability of fire situation judgment, avoid assessment errors caused by unreasonable weight distribution, and provide an optimized data analysis framework for blank point fire assessment, thereby enhancing the ability to comprehensively monitor and accurately prevent and control forest fires.

[0051] The blank fire analysis result determination subunit is used to determine the blank fire analysis result of the blank point according to the reference fire analysis result of the predetermined ranking in the reference descending list, and the blank point and the blank fire analysis result form a third mapping relationship. Specifically, the blank fire analysis result determination subunit determines the blank fire determination result of the blank point based on the reference descending list. First, the reference points of the predetermined ranking are determined according to the preset rules. The predetermined ranking is usually determined based on practical experience and data analysis. Several reference points with high rankings are selected, such as the top 3 or top 5, because they are close to the blank points in space and have a higher correlation with the fire analysis results. The fire analysis results of the selected reference points are comprehensively analyzed and processed. The weighted average method can be used. According to the inverse of the spatial distance between the reference point and the blank point as the weight (the closer the distance, the greater the weight), the fire probability values ​​of the reference point are weighted and summed to obtain the estimated fire probability of the blank point; for the judgment of whether the fire occurs or not, the majority voting method can be used, that is, if the majority of the selected reference points are judged to have a fire, the blank point is preliminarily judged to have a fire, otherwise it is judged to have no fire. At the same time, combined with the auxiliary information such as the fire development trend and the cause of the fire at the reference point, the fire situation of the blank point is described and analyzed to form a complete blank fire judgment result. A third mapping relationship is established between the blank point and the blank fire judgment result, which is recorded in the database, so that the fire situation of the blank point can be reasonably inferred and evaluated, filling the gap in fire monitoring data, improving the comprehensive grasp of the entire target forest fire situation, improving the scientificity and accuracy of forest fire prevention and control work, and reducing fire risks and losses.

[0052] The target forest fire assessment visual map acquisition subunit is used to render the third mapping relationship to the initial forest fire assessment visual map to obtain the target forest fire assessment visual map. Specifically, the target forest fire assessment visual map acquisition subunit integrates the third mapping relationship into the initial forest fire assessment visual map and performs rendering and updating. First, the layer and graphic element information of the initial visual map are read on the visualization software platform, and then the blank points are marked with specific icons (such as gray blocks, dynamically adjusting the color and flashing effects according to the blank fire assessment results) according to the third mapping relationship, and the corresponding detailed information is associated. By using graphic rendering technology, the blank point information and the original information are integrated to form a more complete and comprehensive target forest fire assessment visual map. The visual map can present the fire situation of the tower monitoring area and the blank points, and use reasonable methods to show the fire risk situation, provide an intuitive and accurate display platform for relevant personnel, help them fully grasp the fire distribution and development trend, more efficiently formulate prevention and control strategies and allocate resources, improve the efficiency and effect of forest fire prevention and control work, and ensure the safety and stability of forest resources and ecological environment.

[0053] In a possible implementation, the first comprehensive support acquisition unit further includes: a second support verification mechanism extraction subunit, the second support verification mechanism extraction subunit is used to extract the second support verification mechanism in the predetermined support verification mechanism. Specifically, after receiving the predetermined support verification mechanism, the second support verification mechanism extraction subunit performs in-depth analysis and screening on it to extract the second support verification mechanism. This mechanism is usually a set of rules and processes based on time series analysis and historical data comparison, which provides further verification support for the first fire assessment result from a historical perspective. It contains key elements such as specific time window settings, data screening conditions, and comparative analysis methods. For example, it stipulates the time range for comparing historical data (such as the past year, a quarter, etc.), the standard for selecting historical periods similar to current environmental conditions for comparison, and the comparison weight setting for different monitoring parameters (acoustic, optical, temperature, etc.), etc., to provide precise direction and rule guidance for subsequent historical monitoring record acquisition and analysis, ensure that the information extracted from the historical data is targeted and effective, avoid blindly using a large amount of irrelevant historical data, improve the efficiency and accuracy of the verification work, and lay the foundation for accurately evaluating the reliability of the first fire assessment result.

[0054] A historical monitoring record acquisition subunit, the historical monitoring record acquisition subunit is used to acquire the historical monitoring records of the first tower point according to the second support verification mechanism. Specifically, the historical monitoring record acquisition subunit interacts with the database storing historical monitoring data according to the time range, data screening conditions and other requirements set in the second support verification mechanism, and acquires the historical monitoring records of the first tower point. The records cover the detailed data collected by the tower point through various sensors (acoustic, optical, temperature, etc.) in the past specific time period, including the sensor readings at each time point, the timestamp of data collection, the environmental conditions at that time (such as temperature, humidity, wind speed and other meteorological data) and possible manual records (such as whether there are abnormal observations, maintenance records, etc.). By screening and collating data according to the requirements of the second support verification mechanism, it is ensured that the acquired historical monitoring records are comparable and relevant to the current fire assessment situation, providing a high-quality data source for subsequent year-on-year and month-on-month analysis, so that useful information can be mined from historical data, potential fire occurrence laws and trends can be discovered, and strong historical data support can be provided for judging the rationality of the current fire assessment results, thereby enhancing the reliability and stability of the fire monitoring and early warning system.

[0055] The first historical year-on-year slice extraction subunit is used to perform year-on-year segmentation on the historical monitoring records to obtain a first segmentation result, and extract the first historical year-on-year slice in the first segmentation result. Specifically, the first historical year-on-year slice extraction subunit performs year-on-year segmentation on the historical monitoring records to extract the first historical year-on-year slice. Year-on-year segmentation is performed according to the same time period comparison and analysis, and the historical monitoring records are divided into multiple time periods according to the time range and granularity (such as quarters) determined by the second support verification mechanism, and the historical data corresponding to the current fire judgment time period are screened out to obtain the first segmentation result, and then the monitoring parameter data related to the current fire judgment is extracted from it to form the first historical year-on-year slice. For example, if the summer fire situation at the first tower point is currently being judged, the acoustic, optical, temperature and other data of the point in the past few summers are extracted from the historical records for slicing. Through year-on-year analysis, the trend and characteristics of historical monitoring data changes under similar conditions can be observed, which provides a reference for judging the possibility of the current fire, helps to identify anomalies, improve the accuracy and scientificity of fire judgment, timely discover potential risks, and enhance the foresight and effectiveness of forest fire prevention and control.

[0056] The first historical year-on-year slice extraction subunit is used to perform year-on-year segmentation on the historical monitoring record to obtain a second segmentation result, and extract the first historical year-on-year slice from the second segmentation result. Specifically, the first historical month-on-month slice extraction subunit uses the month-on-month segmentation method to process the historical monitoring records to obtain the first historical month-on-month slice. The month-on-month segmentation is a method of comparing data from adjacent time periods to observe short-term change trends. The historical monitoring records are divided in turn according to the month-on-month time step (such as one week, one month, etc.) specified by the second support verification mechanism to obtain a combination of data from multiple adjacent time periods to form a second segmentation result. The data from the previous time period closest to the current fire assessment time is then selected to extract the monitoring parameter data related to the fire assessment to form the first historical month-on-month slice. For example, when the fire situation in a certain week in July is currently being assessed, the historical monitoring data from the last week of June is extracted as a slice. The month-on-month analysis can help understand recent changes in historical data, especially whether there are any abnormal fluctuations in monitoring data in the short term before the fire occurs, such as a sudden increase in temperature, an increase in abnormal acoustic signals, the appearance of optical smoke features, etc., thereby providing timely and sensitive short-term data support for judging the possibility of the current fire, helping to capture early signs of fire, improve the timeliness and accuracy of fire warnings, and gain precious time and opportunities for early prevention and control of forest fires.

[0057] The first historical year-on-year fire label extraction subunit is used to extract the first historical year-on-year fire label in the first historical year-on-year slice and the first historical year-on-year fire label in the first historical year-on-year slice in sequence. Specifically, after obtaining the first historical year-on-year slice and the first historical year-on-year slice, the first historical year-on-year fire label extraction subunit deeply analyzes the data therein to extract the first historical year-on-year fire label and the first historical year-on-year fire label. The label is marked according to whether the fire occurred and the severity in the historical monitoring data. For the first historical year-on-year slice, by querying historical fire record archives, fire fighting reports, manual observation records and other materials, it is determined whether there is a fire in the area where the first tower point in the same period of history is located. If it occurs, the corresponding labels such as small fire, medium fire, and large fire are assigned as the first historical year-on-year fire label according to the scale, duration, and scope of influence of the fire; for the first historical year-on-year slice, a similar method is used to extract the fire occurrence and label from the historical data of the adjacent time period as the first historical year-on-year fire label. These fire labels provide key qualitative information for subsequent support calculations. They can evaluate the rationality of current assessment results based on actual historical fire conditions, judge current fire risks by comparing historical regularity characteristics, improve the accuracy and reliability of assessments, and provide a scientific and objective basis for forest fire prevention and control decisions.

[0058] The second support acquisition subunit is used to obtain the second support according to the first historical year-on-year fire label and the first historical quarter-on-quarter fire label. Specifically, the second support acquisition subunit calculates the second support according to the first historical year-on-year fire label and the first historical quarter-on-quarter fire label, constructs a support evaluation model, and quantifies its value by comprehensively considering multiple factors. When the historical fire label is consistent with the current first fire assessment result, such as fires occurred in the same period and adjacent time periods in history and the current assessment result is also "there is a fire", points are added to the basic score according to pre-set rules, and the amount of added points is dynamically adjusted according to factors such as the severity of the historical fire, the frequency of occurrence, and the similarity with the current environment; on the contrary, if no fire occurred in history and the current assessment is "there is a fire", it is handled according to the difference deduction rules, and the deduction also refers to the above factors. The second support value is calculated through a dynamic scoring mechanism based on historical labels. This value intuitively reflects the degree of support of historical fire conditions for the current analysis and judgment results. As an important quantitative basis for evaluating reliability, it verifies and improves the current analysis and judgment results from a historical perspective, reduces the risk of misjudgment, improves the accuracy and scientific nature of forest fire monitoring and early warning, and ensures the safety and stability of forest resources and the ecological environment.

[0059] The first comprehensive support acquisition subunit is used to replace the first support with the average of the second support and the first support as the first comprehensive support. Specifically, the first comprehensive support acquisition subunit uses the average of the second support and the first support as the new first comprehensive support. First, the first support based on the fire assessment results of the surrounding tower points and the second support just calculated based on the historical monitoring data are obtained, and then the arithmetic mean of the two is calculated, that is, (first support + second support) / 2, and this new value is used as the first comprehensive support. This is to integrate the support of the current surrounding environment information and historical experience data to the first fire assessment results, so that the first comprehensive support can more comprehensively and objectively reflect its reliability and credibility, avoid the limitations of single factor judgment, make full use of multi-source and multi-angle information, and provide more scientific and accurate quantitative indicators for fire decision-making. Decision makers can judge the credibility of the first fire assessment results more accurately based on this value and decide whether to take further action.

[0060] The fire warning module 60 is used to activate the warning signal to warn the first tower point of the target forest when the first comprehensive support reaches the predetermined support threshold. Specifically, the fire warning module 60 monitors the first comprehensive support, and the predetermined support threshold is a key value preset based on historical data, fire characteristics and risk assessment model. After obtaining the new first comprehensive support, the module uses an accurate algorithm to compare with the threshold. If the threshold is reached or exceeded, it means that the multi-party data shows that the possibility of fire at the first tower point is high. The module selects an adaptive warning signal from the library (based on regional characteristics and risk levels, such as multi-channel signals such as high loudness in densely populated areas and remote persistent signals in remote areas), and sends it through the connection communication and alarm equipment, and at the same time attaches information such as tower location, fire possibility quantification value and fire prediction, which helps personnel and departments make quick decisions, reduce losses, ensure forest safety, and improve the timeliness, effectiveness, system reliability and adaptability of prevention and control.

[0061] In the above, refer to Figure 1 The forest fire real-time monitoring and early warning device based on the combined perception of sound, light and heat according to the embodiment of the present invention is described in detail. Figure 2 A forest fire real-time monitoring and early warning method using acoustic, optical and thermal combined sensing according to an embodiment of the present invention is described.

[0062] Real-time forest fire monitoring and early warning methods based on combined acoustic, optical and thermal sensing, such as Figure 2 As shown, the method includes:

[0063] The target real-time status information of the target forest is obtained by dynamic monitoring using an acoustic, optical and thermal combined sensing device, wherein the acoustic, optical and thermal combined sensing device is mounted on a distributed tower of the target forest; the first status information corresponding to the first tower point in the target real-time status information is extracted, and the first tower point refers to any one of the distributed towers; the first status information is used as input data of a fire analysis device, and output data is obtained through the fire analysis device, wherein the output data includes a first fire analysis result of the first tower point; when the first fire analysis result meets a predetermined verification constraint, a predetermined support verification mechanism is read; according to the predetermined support verification mechanism, a first comprehensive support degree of the first fire analysis result is obtained; when the first comprehensive support degree reaches a predetermined support degree threshold, an early warning signal is activated to issue a fire warning to the first tower point of the target forest.

[0064] In a possible implementation, the real-time monitoring and early warning method for forest fires using the combined acoustic, optical and thermal perceptions also includes: obtaining first sound information of the first tower point through the acoustic sensor; performing image recognition analysis on the first infrared image of the first tower point through the optical sensor to obtain first optical information; obtaining first temperature information of the first tower point through the temperature sensor; performing variation weighted processing on the first sound information, the first optical information and the first temperature information in turn to obtain a first sound coefficient, a first optical coefficient and a first temperature coefficient, respectively; the first sound coefficient, the first optical coefficient and the first temperature coefficient constitute the first state information.

[0065] In a possible implementation, the real-time monitoring and early warning method for forest fires using the combined acoustic, optical and thermal perception also includes: acquiring a first infrared pixel in the first infrared image through the optical sensor; analyzing the first infrared pixel corresponding to the first RGB to obtain a first light color parameter; analyzing the first infrared pixel corresponding to the first HSB to obtain a first light intensity parameter; the first light color parameter and the first light intensity parameter constitute the first optical information.

[0066] In a possible implementation, the real-time monitoring and early warning method for forest fires using combined acoustic, optical and thermal perception also includes: obtaining a forest fire log, the forest fire log including a first record and a second record; forming a first data group based on first historical status information and a first historical fire label in the first record; forming a second data group based on second historical status information and a second historical no-fire label in the second record; and performing supervised learning on the first data group and the second data group after sample enhancement processing based on the principle of support vector machines to obtain the fire analysis device.

[0067] In a possible implementation, the real-time monitoring and early warning method for forest fires using combined acoustic, optical and thermal perception also includes: extracting a first support verification mechanism from the predetermined support verification mechanism; determining a first neighborhood of the first tower point based on the first support verification mechanism, the first neighborhood including multiple tower points; randomly extracting any tower point from the multiple tower points, the arbitrary tower point corresponding to arbitrary status information; obtaining an arbitrary fire analysis result by analyzing the arbitrary status information through the fire analysis device; obtaining a first support degree of the first fire analysis result based on the arbitrary fire analysis result; and using the first support degree as the first comprehensive support degree.

[0068] In a possible implementation, the real-time monitoring and early warning method for forest fires using combined acoustic, optical and thermal perception also includes: forming a first mapping relationship between the first tower point and the first fire assessment result; forming a second mapping relationship between the arbitrary tower point and the arbitrary fire assessment result; and constructing a target forest fire assessment visual graph of the target forest based on the first mapping relationship and the second mapping relationship.

[0069] In a possible implementation, the real-time monitoring and early warning method for forest fires using combined acoustic, optical and thermal perception also includes: constructing an initial forest fire assessment visual graph based on the first mapping relationship and the second mapping relationship; obtaining blank points in the initial forest fire assessment visual graph, and forming an interpolation reference set of the blank points; extracting a first reference fire assessment result of a first reference point in the interpolation reference set; obtaining a first spatial distance between the first reference point and the blank point; arranging the first reference fire in descending order based on the first spatial distance to obtain a reference descending list of the first reference fire; determining a blank fire assessment result of the blank point based on a predetermined ranking of reference fire assessment results in the reference descending list, and forming a third mapping relationship between the blank point and the blank fire assessment result; rendering the third mapping relationship to the initial forest fire assessment visual graph to obtain the target forest fire assessment visual graph.

[0070] In a possible implementation, the real-time monitoring and early warning method for forest fires based on combined acoustic, optical and thermal perception also includes: extracting a second support verification mechanism from the predetermined support verification mechanism; obtaining historical monitoring records of the first tower point according to the second support verification mechanism; performing year-on-year segmentation on the historical monitoring records to obtain a first segmentation result, and extracting a first historical year-on-year slice from the first segmentation result; performing month-on-month segmentation on the historical monitoring records to obtain a second segmentation result, and extracting a first historical month-on-month slice from the second segmentation result; sequentially extracting a first historical year-on-year fire label from the first historical year-on-year slice and a first historical month-on-month fire label from the first historical month-on-month slice; obtaining a second support degree based on the first historical year-on-year fire label and the first historical month-on-month fire label; replacing the first support degree with the average of the second support degree and the first support degree as the first comprehensive support degree.

[0071] The real-time monitoring and early warning device for forest fires with combined sound, light and heat perception provided in the embodiment of the present invention can execute the real-time monitoring and early warning method for forest fires with combined sound, light and heat perception provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0072] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0073] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A real-time forest fire monitoring and early warning device with combined acoustic, optical and thermal sensing, characterized in that: include: A target real-time status information monitoring module, wherein the target real-time status information monitoring module is used to dynamically monitor and obtain the target real-time status information of the target forest through an acoustic, optical and thermal joint sensing device, wherein the acoustic, optical and thermal joint sensing device is mounted on a distributed tower of the target forest; A first state information extraction module, the first state information extraction module is used to extract first state information corresponding to a first tower point in the target real-time state information, the first tower point refers to any one of the distributed towers; an output data acquisition module, the output data acquisition module is used to use the first state information as input data of a fire analysis device, and obtain output data through the fire analysis device, wherein the output data includes a first fire analysis result of the first tower point; A predetermined support verification mechanism reading module, wherein the predetermined support verification mechanism reading module is used to read the predetermined support verification mechanism when the first fire analysis result meets the predetermined verification constraint; A first comprehensive support acquisition module, the first comprehensive support acquisition module is used to obtain a first comprehensive support of the first fire assessment result according to the predetermined support verification mechanism; A fire warning module is used to activate a warning signal to issue a fire warning to the first tower point of the target forest when the first comprehensive support reaches a predetermined support threshold.

2. The real-time monitoring and early warning device for forest fires with combined acoustic, optical and thermal sensing according to claim 1 is characterized in that: The acoustic, optical and thermal joint sensing device includes an acoustic sensor, an optical sensor and a temperature sensor, and the first state information extraction module includes: A first sound information monitoring unit, the first sound information monitoring unit is used to monitor and obtain first sound information of the first tower point through the acoustic sensor; an image recognition and analysis unit, the image recognition and analysis unit being used to perform image recognition and analysis on the first infrared image of the first tower point through the optical sensor to obtain first optical information; A first temperature information acquisition unit, the first temperature information acquisition unit is used to obtain first temperature information of the first tower point through monitoring by the temperature sensor; a variation weighted processing unit, the variation weighted processing unit being used to sequentially perform variation weighted processing on the first sound information, the first optical information and the first temperature information to obtain a first sound coefficient, a first optical coefficient and a first temperature coefficient respectively; The first state information composition unit is used for the first sound coefficient, the first optical coefficient and the first temperature coefficient to compose the first state information.

3. The forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing according to claim 2 is characterized in that: The image recognition and analysis unit comprises: a first infrared pixel acquisition subunit, the first infrared pixel acquisition subunit being configured to acquire a first infrared pixel in the first infrared image through the optical sensor; A first light color parameter acquisition subunit, the first light color parameter acquisition subunit is used to analyze the first infrared pixel corresponding to the first RGB to obtain a first light color parameter; A first light intensity parameter acquisition subunit, the first light intensity parameter acquisition subunit is used to analyze the first infrared pixel corresponding to the first HSB to obtain a first light intensity parameter; The first optical information composition subunit is used for the first light color parameter and the first light intensity parameter to compose the first optical information.

4. The real-time monitoring and early warning device for forest fires with combined acoustic, optical and thermal sensing according to claim 1 is characterized in that: The output data acquisition module comprises: A forest fire log acquisition unit, the forest fire log acquisition unit is used to acquire a forest fire log, the forest fire log includes a first record and a second record; a first data group forming unit, the first data group forming unit being used to form a first data group according to the first historical state information and the first historical fire tag in the first record; a second data group forming unit, the second data group forming unit being used to form a second data group according to the second historical state information and the second historical no-fire tag in the second record; A fire analysis device acquisition unit is used to perform supervised learning on the first data group and the second data group after sample enhancement processing based on the support vector machine principle to obtain the fire analysis device.

5. The forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing according to claim 1 is characterized in that: The first comprehensive support acquisition module includes: a first supported verification mechanism extraction unit, the first supported verification mechanism extraction unit being used to extract a first supported verification mechanism from the predetermined supported verification mechanisms; A first neighborhood determining unit, the first neighborhood determining unit being configured to determine a first neighborhood of the first tower point according to the first supported verification mechanism, the first neighborhood including a plurality of tower points; An arbitrary tower point extraction unit, the arbitrary tower point extraction unit is used to randomly extract any tower point from the multiple tower points, the arbitrary tower point corresponds to any state information; An arbitrary fire assessment result analysis unit, the arbitrary fire assessment result analysis unit is used to analyze the arbitrary state information through the fire assessment device to obtain an arbitrary fire assessment result; A first support degree acquisition unit, the first support degree acquisition unit is used to obtain a first support degree of the first fire assessment result according to the arbitrary fire assessment result; A first comprehensive support acquisition unit is configured to use the first support as the first comprehensive support.

6. The forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing according to claim 5 is characterized in that: The first comprehensive support acquisition unit further includes: a first mapping relationship forming subunit, the first mapping relationship forming subunit being used to form a first mapping relationship between the first tower point and the first fire analysis result; A second mapping relationship forming subunit, the second mapping relationship forming subunit is used to form a second mapping relationship between the arbitrary tower point and the arbitrary fire assessment result; The judgment visible graph construction subunit is used to construct a target forest fire judgment visible graph of the target forest according to the first mapping relationship and the second mapping relationship.

7. The forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing according to claim 6 is characterized in that: The visual graph construction unit comprises: an initial forest fire assessment visible graph construction subunit, the initial forest fire assessment visible graph construction subunit being used to construct an initial forest fire assessment visible graph according to the first mapping relationship and the second mapping relationship; A blank point acquisition subunit, the blank point acquisition subunit is used to acquire blank points in the initial forest fire analysis visual map and to form an interpolation reference set of the blank points; A first reference fire assessment result extraction subunit, the first reference fire assessment result extraction subunit is used to extract a first reference fire assessment result of a first reference point in the interpolation reference set; A first spatial distance acquisition subunit, the first spatial distance acquisition subunit is used to acquire a first spatial distance between the first reference point and the blank point; a reference descending list obtaining subunit, the reference descending list obtaining subunit being used to obtain a reference descending list of the first reference fires in descending order based on the first spatial distance; A blank fire assessment result determination subunit, the blank fire assessment result determination subunit is used to determine the blank fire assessment result of the blank point according to the reference fire assessment result of the predetermined ranking in the reference descending list, and the blank point and the blank fire assessment result form a third mapping relationship; The target forest fire assessment visible graph acquisition subunit is used to render the third mapping relationship to the initial forest fire assessment visible graph to obtain the target forest fire assessment visible graph.

8. The forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal sensing according to claim 5 is characterized in that: The first comprehensive support acquisition unit further includes: a second supported verification mechanism extraction subunit, the second supported verification mechanism extraction subunit being used to extract a second supported verification mechanism from the predetermined supported verification mechanism; A historical monitoring record acquisition subunit, the historical monitoring record acquisition subunit is used to acquire the historical monitoring record of the first tower point according to the second support verification mechanism; a first historical year-on-year slice extraction subunit, the first historical year-on-year slice extraction subunit being used to perform year-on-year segmentation on the historical monitoring record to obtain a first segmentation result, and extract a first historical year-on-year slice from the first segmentation result; a first historical year-on-year slice extraction subunit, the first historical year-on-year slice extraction subunit being used to perform year-on-year segmentation on the historical monitoring record to obtain a second segmentation result, and extract the first historical year-on-year slice from the second segmentation result; A first historical year-on-year fire label extraction subunit, the first historical year-on-year fire label extraction subunit is used to sequentially extract the first historical year-on-year fire label in the first historical year-on-year slice and the first historical year-on-year fire label in the first historical year-on-year slice; A second support acquisition subunit, the second support acquisition subunit is used to obtain a second support according to the first historical year-on-year fire label and the first historical month-on-month fire label; The first comprehensive support acquisition subunit is used to replace the first support with the average of the second support and the first support as the first comprehensive support.

9. A forest fire real-time monitoring and early warning method based on combined acoustic, optical and thermal sensing, characterized in that: The method is applied to the forest fire real-time monitoring and early warning device with combined acoustic, optical and thermal perception according to any one of claims 1 to 8, and the method comprises: The target real-time status information of the target forest is obtained by dynamic monitoring through the acoustic, optical and thermal combined sensing device, wherein the acoustic, optical and thermal combined sensing device is mounted on a distributed tower of the target forest; Extracting first status information corresponding to a first tower point in the target real-time status information, where the first tower point refers to any one of the distributed towers; Using the first state information as input data of a fire analysis device, and obtaining output data through the fire analysis device, wherein the output data includes a first fire analysis result of the first tower point; When the first fire analysis result meets the predetermined verification constraint, reading the predetermined support verification mechanism; Obtaining a first comprehensive support degree of the first fire assessment result according to the predetermined support verification mechanism; When the first comprehensive support reaches a predetermined support threshold, an early warning signal is activated to issue a fire warning to the first tower point of the target forest.

Images