A machine learning-based forest fire risk prediction method and system

Abstract

The present invention discloses a method and system for predicting forest fire risk based on machine learning, including using the data of the monitoring data source of a preset forest within a specified time period as the data to be analyzed; extracting the spatio-temporal distribution characteristics of the data to be analyzed and the reference data source, dividing the spatio-temporal distribution characteristics into regions according to similarity, taking the region where the data to be analyzed with similarity higher than the similarity threshold is located as the candidate region, and vice versa as the fuzzy region; performing autocorrelation analysis on the fuzzy region based on the to-be-determined data source, adding the fuzzy region with autocorrelation higher than the autocorrelation threshold to the candidate region, and performing time-varying impact analysis on the environmental data and the human activity data to obtain the fire driving coefficient; constructing a forest fire risk prediction model according to the fire driving coefficient and the candidate region, optimizing the forest fire risk prediction model according to the fire spread law, and outputting the prediction result.

Description

Technical Field

[0001] The present invention relates to the field of risk prediction, and particularly to a method and system for predicting forest fire risk based on machine learning. Background Art

[0002] As a global natural disaster, forest fires are characterized by strong suddenness, great destructiveness, rapid spread, etc., posing a serious threat to the ecological environment, social economy and human safety. In recent years, affected by factors such as climate change, vegetation drying and increased human activities, the occurrence frequency and harm degree of forest fires have increased significantly. There is an urgent need to develop efficient and accurate risk prediction technologies to support early warning and prevention and control decisions.

[0003] Traditional forest fire risk prediction methods mostly rely on a single data source and use statistical models or empirical indexes for evaluation. However, such methods have significant limitations: First, they fail to fully integrate the spatio-temporal correlation characteristics of multi-dimensional data, resulting in an incomplete description of fire driving factors; Second, they ignore the spatio-temporal heterogeneity of fire risks in large-scale regions and are difficult to distinguish the dynamic differences between high-risk and low-risk sub-regions; Third, the time-varying impacts of human activities and the physical laws of fire spread are not effectively modeled, limiting the adaptability and accuracy of the prediction model.

[0004] With the development of remote sensing technology, Internet of Things monitoring and machine learning, the fusion of multi-source data and the mining of spatio-temporal characteristics provide new ideas for fire prediction. However, existing research still faces challenges: the heterogeneity of multi-source data leads to difficulties in feature alignment; the reliability differences of different monitoring data sources may introduce noise; at the same time, the non-linear relationship between fire risk and driving factors is difficult to analyze through traditional models under complex terrain and vegetation conditions. In addition, existing methods handle the "fuzzy area" rather roughly, do not fully utilize the spatial autocorrelation between data for risk correction, and are prone to prediction blind spots.

[0005] Therefore, it is necessary to propose a new method for predicting forest fire risk based on machine learning. By integrating multi-source monitoring data, combining spatio-temporal distribution feature analysis, autocorrelation correction and time-varying driving modeling, a dynamic risk prediction model is constructed to improve the spatial rationality and time sensitivity of the prediction results and provide technical support for the refined risk management of forest fires. Summary of the Invention

[0006] The object of the present invention is to provide a method for predicting forest fire risk based on machine learning.

[0007] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0008] The present invention includes the following steps:

[0009] Use the data from the monitoring data sources of a preset forest within a specified time period as the data to be analyzed; the monitoring data sources include a reference data source and a pending data source; the fire risk pattern of the reference data source is higher than that of the pending data source; the data includes environmental data, image data, historical fire data, human activity data, and remote sensing data; the environmental data includes meteorological data, vegetation data, and terrain data;

[0010] Extract the spatio-temporal distribution characteristics of the data to be analyzed and the reference data source, perform regional division on the spatio-temporal distribution characteristics according to similarity, and take the area where the data to be analyzed with similarity higher than the similarity threshold as the candidate area, and vice versa as the fuzzy area;

[0011] Based on the pending data source, perform autocorrelation analysis on the fuzzy area, add the fuzzy areas with autocorrelation higher than the autocorrelation threshold to the candidate area, and perform time-varying impact analysis on the environmental data and the human activity data to obtain the fire driving coefficient;

[0012] Construct a forest fire risk prediction model based on the fire driving coefficient and the candidate area, optimize the forest fire risk prediction model according to the fire spread law, input the data to be predicted into the forest fire risk prediction model, and output the prediction result.

[0013] Further, the method for performing regional division on the spatio-temporal distribution characteristics according to similarity includes:

[0014] Calculate the similarity between the spatio-temporal distribution characteristics in the data to be analyzed and the spatio-temporal distribution characteristics in the reference data source, and mark the spatio-temporal distribution characteristics in the data to be analyzed with similarity higher than 0.473 in the preset regional longitude and latitude map;

[0015] Perform primary regional division on the marked preset regional longitude and latitude map according to the aggregation density of the spatio-temporal distribution characteristics to obtain primary regions; calculate the ignition probability of the spatio-temporal distribution characteristics:

[0016]

[0017] where the ignition probability of the c-th spatio-temporal distribution characteristic is the wind intensity in the a-th direction is the neighbor ignition temperature in the a-th direction is the ignition point of the c-th spatio-temporal distribution characteristic is the wind direction vector in the a-th direction is the neighbor coordinate vector in the a-th direction is the number of directions is ;

[0018] The density clustering algorithm is used to perform primary regional adjustment on the spatio-temporal distribution characteristics with a fire probability higher than 0.316 to obtain an adjusted area. The adjusted area containing the spatio-temporal distribution characteristics with a fire probability higher than 0.316 is used as the first area, and vice versa as the fuzzy area.

[0019] Further, a method for performing autocorrelation analysis on the fuzzy area based on the to-be-determined data source includes:

[0020] Extract the spatial distribution characteristics of the to-be-determined data source to obtain a to-be-spatial feature set. Use the spatial distribution characteristics of the reference data source as the reference features; use the spatial distribution characteristics of the fuzzy area as the fuzzy features;

[0021] Use the different spatial distribution characteristics between the to-be-spatial feature set and the reference features as the to-be-determined spatial features, and calculate the autocorrelation between the to-be-determined spatial features and the fuzzy features.

[0022] Further, a method for performing time-varying impact analysis on the environmental data and the human activity data to obtain a fire driving coefficient includes:

[0023] Use a long short-term memory network to perform impact analysis on meteorological data, vegetation data, and terrain data in the environmental data respectively to obtain a meteorological coefficient, a combustible coefficient, and a terrain coefficient; perform binning and discretization processing on elevation, slope, and aspect to obtain the terrain coefficient, and use the human footprint index method to perform impact analysis on the human activity data to obtain a human factor coefficient; where the meteorological coefficient includes the impacts of wind speed, temperature, and precipitation; the terrain coefficient includes the impacts of elevation, slope, and aspect;

[0024] Input the meteorological data, vegetation data, and terrain data into a coefficient fusion model, and perform coefficient standardization processing on the meteorological coefficient, combustible coefficient, terrain coefficient, and human factor coefficient. Among them, perform standard score standardization and weighting on wind speed, temperature, and precipitation in the meteorological coefficient, and use the reinforcement learning parameter tuning method to obtain the dynamic coefficients of wind speed, temperature, and precipitation respectively. The expression of the meteorological coefficient after coefficient standardization is:

[0025]

[0026] Among them, the meteorological coefficient after coefficient standardization is , the wind speed is , the temperature is T, the precipitation is P, the dynamic coefficient of wind speed is , the dynamic coefficient of temperature is , the dynamic coefficient of precipitation is ;

[0027] Obtain the fire driving coefficient according to the meteorological coefficient, combustible coefficient, terrain coefficient, and human factor coefficient after coefficient standardization. The expression is:

[0028]

[0029] where the fire driving coefficient of the x-th area is , the meteorological data after normalization of the x-th area coefficient is , the combustible coefficient after normalization of the x-th area coefficient , the terrain coefficient after normalization of the x-th area coefficient is , the human factor coefficient after normalization of the x-th area coefficient is , and the trainable parameters are respectively 、 ;

[0030] The sliding window mechanism is used to update the trainable parameters, and the expression is:

[0031]

[0032]

[0033] where the learning rate is , the loss function is Loss, and the updated trainable parameter is , and the updated trainable parameter is ; Output the adjusted fire driving coefficient.

[0034] Furthermore, the method for constructing a forest fire risk prediction model according to the fire driving coefficient and the candidate area includes:

[0035] Label the candidate area through the fire driving coefficient, regard the candidate area with the fire driving coefficient greater than 0.615 as the key monitoring area, take the fire driving coefficient as the risk probability of the key monitoring area, and construct an objective function based on the actual fire situation and the key monitoring area. The expression is:

[0036]

[0037] where the objective function is , the fire classification loss function is , the fire intensity regression loss is , the fire spread direction consistency loss is , the first parameter is , the second parameter is , and the third parameter is ;

[0038] The fire classification loss function is the consistency between the risk probability and whether there is a fire. When there is a fire, the fire classification loss function is 0, otherwise it is 1;

[0039] The forest fire risk prediction model includes an autoencoder, a comparison algorithm, and a machine learning algorithm;

[0040] The autoencoder compresses the input data into a low-dimensional feature representation through the encoder, reconstructs the input data through the decoder, and is trained according to minimizing the reconstruction error to extract a low-dimensional feature vector that effectively represents the core features of the input data;

[0041] The comparison algorithm performs a difference analysis by comparing the low-dimensional feature vector with a preset standard low-dimensional feature vector, and takes the low-dimensional feature vector with a difference lower than 0.371 as the key feature;

[0042] The machine learning algorithm optimizes the objective function, trains the model based on the key features using historical data to learn the fire occurrence pattern, and adjusts the parameters through the validation set to achieve the prediction of the forest fire risk.

[0043] Furthermore, the method for optimizing the forest fire risk prediction model according to the fire spread law includes:

[0044] Taking the spatio-temporal distribution characteristics as nodes, screening the spatio-temporal distribution characteristics according to the ignition probability, taking the nodes with a value greater than the ignition threshold as wave source nodes, and calculating the spread radius and energy value of the wave source nodes as:

[0045]

[0046]

[0047] where the fire spread speed at the t-th moment is , the energy attenuation coefficient is , the spread radius of the wave source node at the t-th moment is , the spread radius of the wave source node at the (t - 1)-th moment is , the energy value of the wave source node at the t-th moment is , the energy value of the initial wave source node is , the constant of pi is , the energy of the b-th wave source node at the t-th moment is ;

[0048] If the distance from the wave source node to the j-th node is greater than , then the initial spread fails to reach the j-th node and needs to arrive at the t = t + 2 or further moment; if the distance from the wave source node to the j-th node is less than or equal to , and is greater than or equal to the connection threshold of the j-th node, then the initial spread spreads to the j-th node and the j-th node is activated, and a directed edge from the wave source node to the j-th node is established;

[0049] Calculate the fire spread probability of the activated node:

[0050]

[0051] where the probability decay coefficient is , the connection threshold of the j-th activated node is , the energy value of the j-th wave source node at the t-th moment is , the fire spread probability of the j-th activated node is ;

[0052] Calculate the initial energy value of the activated node:

[0053]

[0054] where the response coefficient of the j-th activated node is , the initial energy value of the j-th activated node is ;

[0055] Calculate the updated node spread radius and energy value:

[0056]

[0057]

[0058] where the spread radius of the b-th activated node at the t-th updated moment is , the spread radius of the b-th activated node at the (t - 1)-th updated moment is , the energy value of the b-th activated node at the t-th updated moment is , the initial energy value of the b-th activated node is ;

[0059] If the distance from the activated node to the i-th neighboring node is greater than , then the fire spread of the b-th activated node fails to spread to the i-th neighboring node; the distance from the activated node to the i-th neighboring node is less than or equal to , and is greater than or equal to the connection threshold of the b-th node, then the fire spread of the b-th activated node spreads to the i-th neighboring node, and calculate the energy value generated by the activation of the i-th neighboring node:

[0060]

[0061] where the energy value of the i-th neighboring node is , the response coefficient of the i-th neighboring node is , the energy value of the i-th neighboring node at the t-th updated moment is , the fire spread probability of the i-th activated node is ;

[0062] Iterate continuously until all nodes are traversed, present the fire spread situation, and adjust the fire driving coefficient of the forest fire risk prediction model according to the variance between the fire spread situation and the actual spread situation.

[0063] In a second aspect, a forest fire risk prediction system based on machine learning includes:

[0064] Data acquisition module: used to take the data of the monitoring data sources of a preset forest within a specified time period as the data to be analyzed; the monitoring data sources include reference data sources and undetermined data sources; the fire risk law of the reference data sources is higher than that of the undetermined data sources; the data includes environmental data, image data, historical fire data, human activity data, and remote sensing data; the environmental data includes meteorological data, vegetation data, and terrain data;

[0065] Comparison and division module: used to extract the spatio-temporal distribution characteristics of the data to be analyzed and the reference data sources, compare and divide the spatio-temporal distribution characteristics according to the similarity, take the area where the data to be analyzed with a similarity higher than the similarity threshold as the candidate area, and vice versa as the fuzzy area;

[0066] Addition impact analysis module: used to perform autocorrelation analysis on the fuzzy area based on the undetermined data sources, add the fuzzy areas with an autocorrelation higher than the autocorrelation threshold to the candidate area, and perform time-varying impact analysis on the environmental data and the human activity data to obtain the fire driving coefficient;

[0067] Model construction and optimization module: used to construct a forest fire risk prediction model according to the fire driving coefficient and the candidate area, optimize the forest fire risk prediction model according to the fire spread law, input the data to be predicted into the forest fire risk prediction model, and output the prediction result.

[0068] The beneficial effects of the present invention are:

[0069] The present invention is a forest fire risk prediction method and system based on machine learning. Compared with the prior art, the present invention has the following technical effects:

[0070] By extracting spatio-temporal distribution characteristics, regional division, autocorrelation analysis, time-varying impact analysis, obtaining model construction, and optimizing the model steps, the present invention can improve the accuracy of forest fire risk prediction, thereby improving the precision of forest fire risk prediction, optimizing the forest fire risk prediction, greatly saving resources, improving work efficiency, realizing intelligent prediction of forest fire risk, performing regional division and fire spread optimization on forest fire risk prediction in real time, which is of great significance for forest fire risk prediction, and can adapt to different standards of forest fire risk prediction and different forest fire risk prediction requirements, having a certain universality. Description of the Drawings

[0071] Figure 1 It is a flowchart of the steps of a method for predicting forest fire risk based on machine learning according to the present invention. Detailed Embodiments

[0072] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention, but do not limit the present invention.

[0073] A method and system for predicting forest fire risk based on machine learning according to the present invention include the following steps:

[0074] As Figure 1 shown, in this embodiment, it includes the following steps:

[0075] Taking the data of the monitoring data sources of a preset forest within a specified time period as the data to be analyzed; the monitoring data sources include reference data sources and undetermined data sources; the fire risk pattern of the reference data source is higher than that of the undetermined data source; the data includes environmental data, image data, historical fire data, human activity data, and remote sensing data; the environmental data includes meteorological data, vegetation data, and terrain data;

[0076] In actual evaluation, taking a certain forest as the research object, collecting the data from October 2019 to October 2024 as historical data, and studying the forest fire risk from 17:34 to 19:00 on December 3, 2024. The meteorological data is temperature 25°C, humidity 40%, wind speed 5 m / s, and wind direction southeast; the vegetation data is vegetation coverage rate 70%, mainly coniferous forest as the vegetation type, and vegetation dryness 0.6; the terrain data is altitude 513 meters, slope 17°, and slope aspect south slope; the historical fire data is that in the past 5 years, there have been 3 small fires in this area, all occurring during the high-temperature and dry period in summer; the human activity data is that there is a small road around the forest, with about 100 vehicles passing by every day, mainly local residents' vehicles and a small number of tourist vehicles; the remote sensing data is to obtain the overall distribution and vegetation health status information of the forest through satellite remote sensing, showing that there are no obvious abnormal hot spots in the current forest;

[0077] Extracting the spatio-temporal distribution characteristics of the data to be analyzed and the reference data source, and dividing the spatio-temporal distribution characteristics into regions according to similarity. The region where the data to be analyzed with similarity higher than the similarity threshold is the candidate region, and vice versa is the fuzzy region;

[0078] The spatio-temporal distribution characteristics of the data to be analyzed are vegetation distribution, terrain distribution, human activity distribution, meteorological conditions, and the time pattern of human activities; the vegetation distribution includes type and coverage, dryness, and health status; the terrain distribution includes altitude and slope, and the impact of terrain on vegetation; the human activity distribution includes along roads and camping areas; the meteorological conditions include seasonal changes and short-term meteorological changes; the time pattern of human activities includes camping activities and vehicle passage;

[0079] The similarity threshold is 0.748; the candidate area is D, and the fuzzy areas are A, B, and C;

[0080] Based on the to-be-determined data source, perform autocorrelation analysis on the fuzzy area, add the fuzzy areas with autocorrelation higher than the autocorrelation threshold to the candidate area, and perform time-varying impact analysis on the environmental data and the human activity data to obtain the fire driving coefficient;

[0081] In the actual evaluation, the autocorrelation threshold is 0.263, and area A is added to the candidate area; the fire driving coefficients of candidate areas A and D are 0.532 and 0.624 respectively;

[0082] Construct a forest fire risk prediction model based on the fire driving coefficient and the candidate area, optimize the forest fire risk prediction model according to the fire spread law, input the data to be predicted into the forest fire risk prediction model, and output the prediction result;

[0083] In the actual evaluation, the risk probabilities of candidate areas A and D are 0 and 0.624 respectively.

[0084] In this embodiment, the method for dividing regions based on the similarity of the spatio-temporal distribution characteristics includes:

[0085] Calculate the similarity between the spatio-temporal distribution characteristics in the data to be analyzed and the spatio-temporal distribution characteristics in the reference data source, and mark the spatio-temporal distribution characteristics in the data to be analyzed with similarity higher than 0.473 in the preset regional longitude and latitude map;

[0086] Perform primary regional division on the marked preset regional longitude and latitude map according to the aggregation density of the spatio-temporal distribution characteristics to obtain primary regions; calculate the ignition probability of the spatio-temporal distribution characteristics:

[0087]

[0088] Among them, the ignition probability of the c-th spatio-temporal distribution characteristic is , the wind intensity in the a-th direction is , the ignition temperature of the neighbor in the a-th direction is , the ignition point of the c-th spatio-temporal distribution characteristic is , the wind direction vector in the a-th direction is , the neighbor coordinate vector in the a-th direction is , the number of directions is ;

[0089] The density clustering algorithm is used to perform primary region adjustment on the spatio-temporal distribution characteristics with a fire probability higher than 0.316 to obtain an adjusted region. The adjusted region containing the spatio-temporal distribution characteristics with a fire probability higher than 0.316 is used as the first region, and vice versa as the fuzzy region.

[0090] In this embodiment, the method for performing autocorrelation analysis on the fuzzy region based on the to-be-determined data source includes:

[0091] Extract the spatial distribution characteristics of the to-be-determined data source to obtain a to-be-spatial feature set. Use the spatial distribution characteristics of the reference data source as the reference features; use the spatial distribution characteristics of the fuzzy region as the fuzzy features;

[0092] Use the different spatial distribution characteristics between the to-be-spatial feature set and the reference features as the to-be-determined spatial features, and calculate the autocorrelation between the to-be-determined spatial features and the fuzzy features.

[0093] In this embodiment, the method for performing time-varying impact analysis on the environmental data and the human activity data to obtain a fire driving coefficient includes:

[0094] Use a long short-term memory network to perform impact analysis on meteorological data, vegetation data, and terrain data in the environmental data respectively to obtain a meteorological coefficient, a combustible coefficient, and a terrain coefficient; perform binning and discretization processing on elevation, slope, and aspect to obtain the terrain coefficient, and use the human footprint index method to perform impact analysis on the human activity data to obtain a human factor coefficient; among them, the meteorological coefficient includes the impacts of wind speed, temperature, and precipitation; the terrain coefficient includes the impacts of elevation, slope, and aspect;

[0095] Input the meteorological data, vegetation data, and terrain data into a coefficient fusion model, and perform coefficient standardization processing on the meteorological coefficient, the combustible coefficient, the terrain coefficient, and the human factor coefficient. Among them, after standard score standardization and weighting of wind speed, temperature, and precipitation in the meteorological coefficient, use the reinforcement learning parameter tuning method to obtain the dynamic coefficients of wind speed, temperature, and precipitation respectively. The expression of the meteorological coefficient after coefficient standardization is:

[0096]

[0097] Among them, the meteorological coefficient after coefficient standardization is , the wind speed is , the temperature is T, the rainfall is P, the dynamic coefficient of wind speed is , the dynamic coefficient of temperature is , the dynamic coefficient of rainfall is ;

[0098] Obtain the fire driving coefficient based on the meteorological coefficient, combustible coefficient, terrain coefficient, and human factor coefficient after coefficient standardization. The expression is as follows:

[0099]

[0100] Among them, the fire driving coefficient of the x-th area is , the meteorological data after coefficient standardization of the x-th area is , the combustible coefficient after coefficient standardization of the x-th area , the terrain coefficient after coefficient standardization of the x-th area is , the human factor coefficient after coefficient standardization of the x-th area is , and the trainable parameters are respectively 、 ;

[0101] Adopt a sliding window mechanism to update the trainable parameters. The expression is as follows:

[0102]

[0103]

[0104] Among them, the learning rate is , the loss function is Loss, and the updated trainable parameter is , and the updated trainable parameter is ; Output the adjusted fire driving coefficient.

[0105] In this embodiment, the method for constructing a forest fire risk prediction model based on the fire driving coefficient and the candidate area includes:

[0106] Label the candidate area through the fire driving coefficient. The candidate area with a fire driving coefficient greater than 0.615 is used as the key monitoring area, and the fire driving coefficient is used as the risk probability of the key monitoring area. Based on the actual fire situation and the key monitoring area, construct an objective function. The expression is as follows:

[0107]

[0108] Among them, the objective function is , the fire classification loss function is , the fire intensity regression loss is , the fire spread direction consistency loss is , the first parameter is , the second parameter is , the third parameter is ;

[0109] The fire classification loss function is the consistency between the risk probability and whether there is a fire. When there is a fire, the fire classification loss function is 0, otherwise it is 1;

[0110] The forest fire risk prediction model includes an autoencoder, a comparison algorithm, and a machine learning algorithm;

[0111] The autoencoder compresses the input data into a low-dimensional feature representation through the encoder, reconstructs the input data through the decoder, and is trained according to the minimization of the reconstruction error to extract a low-dimensional feature vector that effectively represents the core features of the input data;

[0112] The comparison algorithm performs a difference analysis by comparing the low-dimensional feature vector with a preset standard low-dimensional feature vector, and takes the low-dimensional feature vector with a difference lower than 0.371 as the key feature;

[0113] The machine learning algorithm optimizes the objective function, trains the model based on the key features using historical data to learn the fire occurrence pattern, and adjusts the parameters through the validation set to achieve the prediction of the forest fire risk.

[0114] In this embodiment, the method for optimizing the forest fire risk prediction model according to the fire spread law includes:

[0115] Taking the spatio-temporal distribution characteristics as nodes, screening the spatio-temporal distribution characteristics according to the fire probability, taking the nodes greater than the fire threshold as the wave source nodes, and calculating the spread radius and energy value of the wave source nodes as:

[0116]

[0117]

[0118] where the fire spread speed at the t-th moment is , the energy attenuation coefficient is , the spread radius of the wave source node at the t-th moment is , the spread radius of the wave source node at the (t - 1)-th moment is , the energy value of the wave source node at the t-th moment is , the energy value of the initial wave source node is , the constant of pi is , the energy of the b-th wave source node at the t-th moment is ;

[0119] If the distance from the wave source node to the j-th node is greater than , then the initial spread fails to reach the j-th node and needs to arrive at the t = t + 2 or later moment; if the distance from the wave source node to the j-th node is less than or equal to , and If it is greater than or equal to the connection threshold of the j-th node, it initially spreads to the j-th node and the j-th node is activated, and a directed edge from the wave source node to the j-th node is established.

[0120] Calculate the fire spread probability of the activated node:

[0121]

[0122] Where the probability decay coefficient is , the connection threshold of the j-th activated node is , the energy value of the j-th wave source node at the t-th moment is , and the fire spread probability of the j-th activated node is ;

[0123] Calculate the initial energy value of the activated node:

[0124]

[0125] Where the response coefficient of the j-th activated node is , and the initial energy value of the j-th activated node is ;

[0126] Calculate the updated node spread radius and energy value:

[0127]

[0128]

[0129] Where the spread radius of the b-th activated node at the t-th updated moment is , the spread radius of the b-th activated node at the (t - 1)-th updated moment is , the energy value of the b-th activated node at the t-th updated moment is , and the initial energy value of the b-th activated node is ;

[0130] If the distance from the activated node to the i-th neighboring node is greater than , then the fire spread of the b-th activated node fails to spread to the i-th neighboring node; the distance from the activated node to the i-th neighboring node is less than or equal to , and is greater than or equal to the connection threshold of the b-th node, then the fire spread of the b-th activated node spreads to the i-th neighboring node, and calculate the energy value generated by the activation of the i-th neighboring node:

[0131]

[0132] Where the energy value of the i-th neighboring node is , the response coefficient of the i-th neighboring node is , the energy value of the i-th neighboring node at the t-th moment after update is , the fire spread probability of the i-th active node is ;

[0133] Iterate continuously until all nodes are traversed, give the fire spread situation, and adjust the fire driving coefficient of the forest fire risk prediction model according to the variance between the fire spread situation and the actual spread situation.

[0134] In a second aspect, a forest fire risk prediction system based on machine learning includes:

[0135] Data acquisition module: used to take the data of the monitoring data sources of a preset forest within a specified time period as the data to be analyzed; the monitoring data sources include reference data sources and pending data sources; the fire risk law of the reference data source is higher than that of the pending data source; the data includes environmental data, image data, historical fire data, human activity data, remote sensing data; the environmental data includes meteorological data, vegetation data, terrain data;

[0136] Comparison and division module: used to extract the spatio-temporal distribution characteristics of the data to be analyzed and the reference data source, compare and divide the spatio-temporal distribution characteristics according to the similarity, take the area where the data to be analyzed with a similarity higher than the similarity threshold as the candidate area, and vice versa as the fuzzy area;

[0137] Addition impact analysis module: used to perform autocorrelation analysis on the fuzzy area based on the pending data source, add the fuzzy area with an autocorrelation higher than the autocorrelation threshold to the candidate area, and perform time-varying impact analysis on the environmental data and the human activity data to obtain the fire driving coefficient;

[0138] Model construction and optimization module: used to construct a forest fire risk prediction model according to the fire driving coefficient and the candidate area, optimize the forest fire risk prediction model according to the fire spread law, input the data to be predicted into the forest fire risk prediction model, and output the prediction result.

[0139] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A forest fire risk prediction method based on machine learning, characterized in that: The following steps are involved: The data of the monitoring data source of the preset forest within the specified time period is used as the data to be analyzed; the monitoring data source includes a reference data source and a pending data source; The fire risk pattern of the reference data source is higher than that of the pending data source; the data includes environmental data, image data, historical fire data, human activity data, and remote sensing data; the environmental data includes meteorological data, vegetation data, and terrain data; Extracting the spatiotemporal distribution features of the data to be analyzed and the reference data source, dividing the spatiotemporal distribution features into regions according to similarity, and defining the regions where the data to be analyzed are located with similarity higher than a similarity threshold as candidate regions, and vice versa as fuzzy regions; Performing autocorrelation analysis on the fuzzy area based on the pending data source, adding the fuzzy area with autocorrelation higher than the autocorrelation threshold to the candidate area, and performing time-varying impact analysis on the environmental data and the human activity data to obtain a fire driving coefficient; A forest fire risk prediction model is constructed according to the fire driving coefficient and the candidate area, the forest fire risk prediction model is optimized according to the fire spread law, the data to be predicted is input into the forest fire risk prediction model, and the prediction result is output.

2. A forest fire risk prediction method based on machine learning according to claim 1, characterized in that: The method for dividing the spatiotemporal distribution features into regions according to similarity includes: Calculate the similarity between the spatiotemporal distribution features in the data to be analyzed and the spatiotemporal distribution features in the reference data source, and mark the spatiotemporal distribution features in the data to be analyzed with a similarity higher than 0.473 in the latitude and longitude map of the preset area; The spatiotemporal distribution characteristics are divided into primary regions according to the clustering density of the pre-annotated latitude and longitude map of the preset area to obtain the primary region; the fire probability of the spatiotemporal distribution characteristics is calculated: The ignition probability of the cth spatiotemporal distribution feature is , the wind intensity in the ath direction is , the ignition temperature of the neighbor in the ath direction is , the ignition point of the cth spatiotemporal distribution feature is , the wind direction vector in the ath direction is , the neighbor coordinate vector in the ath direction is , the number of directions is ; The density clustering algorithm is used to perform primary regional adjustment on the spatiotemporal distribution features with a fire probability higher than 0.316 to obtain the adjusted area. The adjusted area containing the spatiotemporal distribution features with a fire probability higher than 0.316 is taken as the first area, otherwise it is taken as the fuzzy area.

3. The method for predicting forest fire risk based on machine learning according to claim 1, characterized in that: The method for performing autocorrelation analysis on the fuzzy area based on the pending data source includes: Extract spatial distribution features of the pending data source to obtain a pending spatial feature set, take the spatial distribution features of the reference data source as reference features; take the spatial distribution features of the fuzzy area as fuzzy features; The different spatial distribution characteristics of the to-be-determined spatial feature set and the reference features are taken as the to-be-determined spatial features, and the autocorrelation between the to-be-determined spatial features and the fuzzy features is calculated.

4. The method for predicting forest fire risk based on machine learning according to claim 1, characterized in that: The method for performing time-varying impact analysis on the environmental data and the human activity data to obtain a fire driving coefficient includes: The long short-term memory network is used to analyze the meteorological data, vegetation data, and terrain data in the environmental data to obtain the meteorological coefficient, combustible coefficient, and terrain coefficient. The elevation, slope, and slope direction are processed by binning to obtain the terrain coefficient. The human footprint index method is used to analyze the human activity data to obtain the human factor coefficient. The meteorological coefficient includes the influence of wind speed, temperature, and precipitation; the terrain coefficient includes the influence of elevation, slope, and slope direction. The meteorological data, vegetation data, and terrain data are input into the coefficient fusion model, and the meteorological coefficient, combustible coefficient, terrain coefficient, and human factor coefficient are standardized. The wind speed, temperature, and precipitation in the meteorological coefficient are standardized by standard scores and weighted. The dynamic coefficients of wind speed, temperature, and precipitation are obtained by reinforcement learning parameter adjustment method. The expression of the meteorological coefficient after coefficient standardization is: The meteorological coefficient after coefficient standardization is , wind speed is , temperature is T, rainfall is P, and the dynamic coefficient of wind speed is , the dynamic coefficient of temperature is , the dynamic coefficient of rainfall is ; The fire driving coefficient is obtained based on the standardized meteorological coefficient, combustible coefficient, terrain coefficient and human factor coefficient. The expression is: The fire driving coefficient of the xth area is , the meteorological data after the x-th regional coefficient is standardized: , the combustible coefficient after the x-th region coefficient is standardized , the terrain coefficient after normalization of the x-th regional coefficient is , the human factor coefficient after standardization of the x-th regional coefficient is , the trainable parameters are , ; The sliding window mechanism is used to update the trainable parameters, and the expression is: The learning rate is , the loss function is Loss, and the updated trainable parameters for , the updated trainable parameters for ; Output the adjusted fire driving coefficient.

5. The method for predicting forest fire risk based on machine learning according to claim 1, characterized in that: The method for constructing a forest fire risk prediction model according to the fire driving coefficient and the candidate area comprises: The candidate areas are marked by the fire driving coefficient, and the candidate areas with a fire driving coefficient greater than 0.615 are taken as key monitoring areas. The fire driving coefficient is taken as the risk probability of the key monitoring area. The objective function is constructed based on the actual fire situation and the key monitoring area. The expression is: The objective function is , the fire classification loss function is , the fire intensity regression loss is , the consistency loss of fire spread direction is , the first parameter is , the second parameter is , the third parameter is ; The fire classification loss function is the consistency between the risk probability and whether there is a fire. When there is a fire, the fire classification loss function dimension is 0, otherwise it is 1; The forest fire risk prediction model includes autoencoders, comparison algorithms, and machine learning algorithms; The autoencoder compresses the input data into a low-dimensional feature representation through the encoder, reconstructs the input data through the decoder, and is trained based on minimizing the reconstruction error to extract a low-dimensional feature vector that effectively represents the core features of the input data; The comparison algorithm analyzes the difference between the low-dimensional feature vector and the preset standard low-dimensional feature vector, and uses the low-dimensional feature vector with a difference less than 0.371 as the key feature; The machine learning algorithm optimizes the objective function, uses historical data to train the model based on key features to learn the fire occurrence pattern, and adjusts the parameters through the validation set to predict the risk of forest fires.

6. The method for predicting forest fire risk based on machine learning according to claim 1, characterized in that: The method for optimizing the forest fire risk prediction model according to the fire spread law includes: The spatiotemporal distribution features are taken as nodes, and the spatiotemporal distribution features are screened according to the probability of ignition. The nodes with a value greater than the ignition threshold are taken as wave source nodes, and the spreading radius and energy value of the wave source nodes are calculated as follows: The fire spread speed at time t is , the energy attenuation coefficient is , the spreading radius of the wave source node at time t is , the spreading radius of the wave source node at time t-1 is , the energy value of the wave source node at time t is , the energy value of the initial wave source node is , the circumference constant is , the spreading radius of the bth wave source node at the tth time is ; If the distance from the wave source node to the jth node is greater than , then the initial spread fails to spread to the jth node and needs to arrive at t=t+2 or later; if the distance from the source node to the jth node is less than or equal to ,and If the value is greater than or equal to the connection threshold of the jth node, the initial spread spreads to the jth node and the jth node is activated, and a directed edge is established from the source node to the jth node; Calculate the fire spread probability of the activated node: The probability attenuation coefficient is , the connection threshold of the jth activated node is , the energy value of the j-th wave source node at the t-th time is , the fire spread probability of the jth activated node is ; Calculate the initial energy value of the activated node: The response coefficient of the jth activated node is , the initial energy value of the jth activated node is ; Calculate the updated node spreading radius and energy value: The spreading radius of the bth activated node at the tth moment after the update is , the spreading radius of the bth activated node at the t-1th moment after the update is , the energy value of the bth activated node at the tth moment after the update is , the initial energy value of the bth activated node is ; If the distance from the activated node to the i-th neighboring node is greater than , the fire spread of the bth activated node fails to spread to the i-th neighboring node; the distance from the activated node to the i-th neighboring node is less than or equal to ,and If the connection threshold of the bth node is greater than or equal to that of the bth node, the fire of the bth activated node spreads to the i-th neighboring node, and the energy value generated by the activation of the i-th neighboring node is calculated: The energy value of the i-th neighboring node is , the response coefficient of the i-th neighboring node is , the energy value of the i-th neighboring node at the t-th time after the update is , the fire spread probability of the i-th activated node is ; Continue to iterate until all nodes are traversed and the fire spread situation is given. The fire driving coefficient of the forest fire risk prediction model is adjusted according to the variance between the fire spread situation and the actual spread situation.

7. A forest fire risk prediction system based on machine learning, used to execute the method according to any one of claims 1 to 6, characterized in that: include: Data collection module: used to collect data from the monitoring data source of the preset forest within a specified time period as data to be analyzed; The monitoring data source includes a reference data source and a pending data source; The fire risk pattern of the reference data source is higher than that of the pending data source; the data includes environmental data, image data, historical fire data, human activity data, and remote sensing data; the environmental data includes meteorological data, vegetation data, and terrain data; Comparison and division module: used to extract the spatiotemporal distribution features of the data to be analyzed and the reference data source, divide the spatiotemporal distribution features into regions according to the similarity, and take the region where the data to be analyzed with a similarity higher than a similarity threshold as a candidate region, and vice versa as a fuzzy region; Adding an impact analysis module: used to perform autocorrelation analysis on the fuzzy area based on the pending data source, add the fuzzy area with autocorrelation higher than the autocorrelation threshold to the candidate area, and perform time-varying impact analysis on the environmental data and the human activity data to obtain a fire driving coefficient; Model building and optimization module: used to build a forest fire risk prediction model according to the fire driving coefficient and the candidate area, optimize the forest fire risk prediction model according to the fire spread law, input the data to be predicted into the forest fire risk prediction model, and output the prediction result.

Images