Fire scene rescue tactical acquisition method and device and electronic equipment

Abstract

The invention provides a fire scene rescue tactical acquisition method and device and electronic equipment, and the method comprises the steps: obtaining the live line classification feature of any first grid unit in a grid unit set corresponding to a live line and the terrain classification feature of any first grid unit; acquiring a feature vector corresponding to any first grid unit according to a first feature vector corresponding to the live wire classification feature, a second feature vector corresponding to the terrain classification feature and a third feature vector corresponding to the environment feature of any first grid unit; and analyzing the feature vector corresponding to any first grid unit by adopting a target map convolutional neural network model and a target time sequence Transform model to obtain a fire scene rescue tactical corresponding to any first grid unit, thereby solving the problem that in the prior art, a fire remedy method cannot perform adaptive change, and the fire scene rescue tactics cannot be changed. The technical problems that the fire remedy method is not matched with the current fire situation and the fire extinguishing effect is poor are solved.

Description

Technical Field

[0001] The present application relates to the field of fire fighting technology, and in particular to a method, device and electronic equipment for acquiring fire rescue tactics. Background Art

[0002] With the frequent occurrence of forest fires and urban fires, for example, fire remediation methods corresponding to forest fires or urban fires can be used to extinguish fires. Fire remediation methods can be determined based on experience or preset rules, but fire remediation methods cannot be adaptively changed, so that the fire remediation methods do not match the current fire situation and the fire extinguishing effect is poor. Summary of the invention

[0003] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first purpose of this application is to propose a method for obtaining fire rescue tactics to improve the accuracy of fire rescue tactics determination, optimize fire rescue resource allocation, reduce fire fighting time, and improve fire extinguishing efficiency.

[0005] The second objective of the present application is to provide a fire rescue tactics acquisition device.

[0006] The third objective of the present application is to provide an electronic device.

[0007] A fourth objective of the present application is to provide a computer-readable storage medium.

[0008] A fifth object of the present application is to provide a computer program product.

[0009] To achieve the above-mentioned purpose, the first embodiment of the present application proposes a method for acquiring fire rescue tactics, comprising the following steps:

[0010] Obtaining a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell;

[0011] Acquire a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell;

[0012] A target graph convolutional neural network (GCN) model and a target temporal transformer model are used to analyze the feature vector corresponding to any of the first grid cells to obtain the fire rescue tactics corresponding to any of the first grid cells.

[0013] To achieve the above-mentioned purpose, the second embodiment of the present application proposes a fire rescue tactics acquisition device, comprising:

[0014] A feature acquisition unit, used to acquire a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell;

[0015] a vector acquisition unit, configured to acquire a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell;

[0016] A tactical acquisition unit is used to analyze the feature vector corresponding to any of the first grid units using a target graph convolutional neural network model and a target time series Transformer model to obtain the fire rescue tactics corresponding to any of the first grid units.

[0017] To achieve the above-mentioned purpose, the third aspect of the present application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0018] The memory stores computer-executable instructions;

[0019] The processor executes the computer-executable instructions stored in the memory to implement any method described in the first aspect above.

[0020] To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer execution instructions, and when the computer execution instructions are executed by a processor, they are used to implement any one of the methods described in the first aspect above.

[0021] To achieve the above-mentioned purpose, the fifth aspect of the present application proposes a computer program product, including a computer program, which, when executed by a processor, implements any method described in the first aspect above.

[0022] The present application provides a method, device and electronic device for acquiring fire rescue tactics. The method obtains the fire line classification characteristics of any first grid unit in the set of grid units corresponding to the fire line and the terrain classification characteristics of any first grid unit; obtains the feature vector corresponding to any first grid unit according to the first feature vector corresponding to the fire line classification characteristics, the second feature vector corresponding to the terrain classification characteristics and the third feature vector corresponding to the environmental characteristics of any first grid unit; uses a target graph convolutional neural network model and a target time series Transformer model to analyze the feature vector corresponding to any first grid unit to obtain the fire rescue tactics corresponding to any first grid unit, thereby solving the problem that the fire remediation method cannot be adaptively changed. , which makes the fire remediation method mismatch with the current fire situation and the fire extinguishing effect is poor. Since the fire rescue tactics can be determined by the fire line classification characteristics, terrain classification characteristics and current environmental characteristics, the interaction between terrain changes and environmental factors during the fire spread process can be considered, and the real-time changes of the fire process can be adaptively adjusted. The fire rescue tactics can be determined by combining the graph convolutional neural network model and the time series Transformer model. The fire rescue tactics can be determined by combining spatial and time series information, which can improve the matching of fire rescue tactics and the current fire situation without relying on experience or artificial rules. The accuracy of the determination of fire rescue tactics can be improved, the allocation of fire rescue resources can be optimized, the fire fighting time can be reduced, and the fire extinguishing efficiency and fire extinguishing effect can be improved.

[0023] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A flowchart of a method for acquiring fire rescue tactics provided in an embodiment of the present application;

[0026] Figure 2 A flowchart of a method for acquiring fire rescue tactics provided in an embodiment of the present application;

[0027] Figure 3 An exemplary schematic diagram of a method for acquiring fire line classification features provided in an embodiment of the present application;

[0028] Figure 4 An example schematic diagram of a neighborhood voting method provided in an embodiment of the present application;

[0029] Figure 5An exemplary schematic diagram of a method for determining fire rescue tactics provided in an embodiment of the present application;

[0030] as well as Figure 6 A schematic diagram of the structure of a fire scene rescue tactics acquisition device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0032] The following describes the fire scene rescue tactics acquisition method and device according to an embodiment of the present application with reference to the accompanying drawings.

[0033] Figure 1 A flowchart of a method for acquiring fire rescue tactics provided in an embodiment of the present application.

[0034] To address this problem, the present application provides a method for acquiring fire rescue tactics, which can improve the accuracy of acquiring fire rescue tactics in the cooking stage while reducing the cost of acquiring fire rescue tactics, thereby improving the cooking effect of ingredients. Figure 1 As shown, the fire rescue tactics acquisition method includes the following steps:

[0035] Step 101, obtaining a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell;

[0036] According to some embodiments, the execution subject of the embodiments of the present application may be, for example, an electronic device. The electronic device does not specifically refer to a fixed device, and the name of the electronic device is not limited. The electronic device may also be called, for example, a terminal, a mobile device, etc. For example, when the device identification of the electronic device changes, the electronic device may also change accordingly.

[0037] In some embodiments, the fire line may refer to, for example, the boundary line of the fire spread, that is, the place where the open fire is. Different fire locations or fire methods may correspond to different fire lines. For example, different fire time points may also correspond to different fire lines. That is, the fire line of the embodiment of the present application does not specifically refer to a fixed fire line. The fire line of the embodiment of the present application may be used to indicate the fire line corresponding to the currently determined fire rescue tactics.

[0038] According to some embodiments, the grid unit may be, for example, a fire line area obtained by dividing the fire line. The name of the grid unit is not limited. For example, the grid unit may also be called a study area, a fire line area, or a grid unit.

[0039] In some embodiments, the grid unit set may be, for example, a collection of at least one grid unit. The grid unit set does not specifically refer to a fixed set. For example, when the grid units included in the grid unit set change, the grid unit set may also change accordingly. For example, when the division method corresponding to the grid unit changes, the grid unit set may also change accordingly.

[0040] According to some embodiments, the first grid cells may be used, for example, to indicate grid cells that meet density requirements, wherein the first of the first grid cells is used to distinguish from the remaining grid cells.

[0041] In some embodiments, the fire line classification feature can be used to indicate which area of ​​the fire line a certain grid unit belongs to. The fire line classification feature can include, for example, a fire head area, a fire wing area, and a fire tail area. One grid unit can correspond to one fire line classification feature, that is, the fire line classification feature of one grid unit is a single fire line classification feature. For example, when the fire line classification feature of grid unit A is the fire head area, the fire line classification feature of grid unit A is not the fire wing area or the fire tail area.

[0042] In some embodiments, the terrain classification feature may be used to indicate the terrain of a certain grid unit, for example, wherein a grid unit may correspond to at least one type of terrain, which is not limited in the embodiments of the present application.

[0043] According to some embodiments, a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of the any first grid cell may be obtained.

[0044] There is no distinction in the order of obtaining the fire line classification features and the terrain classification features. For example, the fire line classification features and the terrain classification features can be obtained at the same time, or the fire line classification features can be obtained first and then the terrain classification features, or the terrain classification features can be obtained first and then the fire line classification features.

[0045] Step 102, obtaining a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell;

[0046] According to some embodiments, the feature vector may be, for example, a mathematical representation of model input data, and the feature vector in the embodiments of the present application may be used, for example, to indicate a mathematical identifier corresponding to the current feature.

[0047] In some embodiments, the first feature vector may be, for example, a feature vector corresponding to a line-of-fire classification feature. The first feature vector in the first feature vector is used to distinguish from other feature vectors. The first feature vector does not specifically refer to a fixed vector. For example, when the line-of-fire classification feature changes, the first feature vector may also change accordingly.

[0048] According to some embodiments, the second feature vector may be, for example, a feature vector corresponding to a terrain classification feature. The second in the second feature vector is used to distinguish from the remaining feature vectors. The second feature vector does not specifically refer to a fixed vector. For example, when the terrain classification feature changes, the second feature vector may also change accordingly.

[0049] According to some embodiments, the environmental characteristics may be used to indicate the environmental characteristics corresponding to the fire line, for example, the environmental characteristics may include wind direction, wind force, etc.

[0050] In some embodiments, the third feature vector may be, for example, a feature vector corresponding to an environmental feature. The third in the third feature vector is used to distinguish it from other feature vectors. The third feature vector does not specifically refer to a fixed vector. For example, when the environmental feature changes, the third feature vector may also change accordingly.

[0051] In some embodiments, the feature vector corresponding to any first grid unit may be used to refer to a comprehensive feature vector corresponding to any first grid unit, for example.

[0052] According to some embodiments, a feature vector corresponding to any first grid cell may be acquired based on a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell.

[0053] There is no limitation on the order of obtaining the first eigenvector, the second eigenvector and the third eigenvector.

[0054] Step 103: Use the target graph convolutional neural network model and the target time series Transformer model to analyze the feature vector corresponding to any of the first grid units to obtain the fire rescue tactics corresponding to any of the first grid units.

[0055] In some embodiments, the target graph convolutional neural network model may refer to, for example, a graph convolutional neural network model that has been trained. The target graph convolutional neural network model does not specifically refer to a fixed model. For example, when the parameters corresponding to the target graph convolutional neural network model change, the target graph convolutional neural network model may also change accordingly. For example, when the training data set corresponding to the target graph convolutional neural network model changes, the target graph convolutional neural network model may also change accordingly.

[0056] According to some embodiments, the target time series Transformer model may, for example, refer to a time series Transformer model that has been trained. The target time series Transformer model does not specifically refer to a fixed model. For example, when the parameters corresponding to the target time series Transformer model change, the target time series Transformer model may also change accordingly. For example, when the training data set corresponding to the target time series Transformer model changes, the target time series Transformer model may also change accordingly.

[0057] According to some embodiments, the fire rescue tactics may be, for example, the rescue tactics corresponding to the current first grid unit. Different first grid units may correspond to different fire rescue tactics. Alternatively, when the mapping relationship between the fire rescue tactics and the feature vector changes, the fire rescue tactics corresponding to the first grid unit may also change accordingly.

[0058] According to some embodiments, a target graph convolutional neural network model and a target temporal Transformer model are used to analyze the feature vector corresponding to any of the first grid units to obtain the fire rescue tactics corresponding to any of the first grid units.

[0059] The present application provides a method, device and electronic device for acquiring fire rescue tactics. The method acquires the fire line classification characteristics of any first grid unit in the set of grid units corresponding to the fire line and the terrain classification characteristics of any first grid unit; acquires the feature vector corresponding to any first grid unit according to the first feature vector corresponding to the fire line classification characteristics, the second feature vector corresponding to the terrain classification characteristics and the third feature vector corresponding to the environmental characteristics of any first grid unit; uses a target graph convolutional neural network model and a target time series Transformer model to analyze the feature vector corresponding to any first grid unit to acquire the fire rescue tactics corresponding to any first grid unit, thereby solving the problem that fire remediation methods cannot be implemented. The problem of adaptive changes in fire prevention methods, which makes the fire remediation methods mismatch with the current fire situation and the fire extinguishing effect is poor, can be solved by determining the fire rescue tactics through the fire line classification characteristics, terrain classification characteristics and current environmental characteristics, and the interactive relationship between terrain changes and environmental factors during the fire spread process can be considered. The real-time changes of the fire process can be adaptively adjusted, and the fire rescue tactics can be determined by combining the graph convolutional neural network model and the time series Transformer model. Not only the static characteristics of the fire scene are considered, but also the changes in the fire spread process can be dynamically responded to, which can improve the matching of fire rescue tactics with the current fire situation, improve the accuracy of the determination of fire rescue tactics, optimize the allocation of fire rescue resources, reduce the time of fire fighting, and improve the efficiency of fire extinguishing.

[0060] This embodiment provides another method for obtaining fire rescue tactics. Figure 2 A flowchart of a method for acquiring fire rescue tactics provided in an embodiment of the present application.

[0061] like Figure 2 As shown, the fire scene rescue tactics acquisition method may include the following steps:

[0062] Step 201, obtaining a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line;

[0063] The specific process is as described above and will not be repeated here.

[0064] In some embodiments, the technical solutions of the embodiments of the present application can be applied to the field of firefighting, especially in the field of fire rescue tactics matching and fire line area classification technology.

[0065] According to some embodiments, Figure 3 The following is an example schematic diagram of a method for acquiring a fire line classification feature provided in an embodiment of the present application. The method of acquiring the fire line classification feature of any first grid unit in a set of grid units corresponding to the fire line includes:

[0066] Using a cellular automaton model to identify the fire line, obtaining fire line spread data of any first grid cell in a set of grid cells corresponding to the fire line, wherein the fire line spread data includes a spread speed and a spread direction;

[0067] Obtaining the angle between the spreading direction and the current environmental wind direction, and obtaining the angle weight of the angle;

[0068] According to the spreading speed and the angle weight, obtaining a feature scoring matrix corresponding to any one of the first grid units;

[0069] According to the feature scoring matrix of each grid unit in the grid unit set and the adaptive density clustering algorithm, the fire line classification feature of any first grid unit is obtained. Therefore, the fire line classification feature can be obtained according to the environmental information, and the adaptive density clustering algorithm can be used to obtain the fire line classification feature, which can reduce the rigid classification based on rules, can flexibly identify the dynamic changes of the fire line, and improve the classification accuracy and adaptability.

[0070] According to some embodiments, the fire line area classification is performed using an adaptive density clustering algorithm, which can automatically cope with the diversity and complexity of the fire area and enhance the adaptability to different fire situations.

[0071] According to some embodiments, the cellular automaton model does not specifically refer to a fixed model. The cellular automaton model may also be called a forest fire spread model or a fire line identification model, for example.

[0072] In some embodiments, the propagation direction may be, for example, a propagation direction corresponding to the next time step. The first grid unit may be, for example, a grid unit that is being ignited.

[0073] In some embodiments, the spreading direction of each first grid unit in the ignition state may be matched with the current environmental wind direction, and the angle θ between the spreading direction and the wind direction may be calculated using formula (1):

[0074]

[0075] Where d is the spreading direction vector, w is the wind direction vector, · represents the dot product of the vectors, and |d| and |w| represent the magnitudes of the vectors, respectively.

[0076] According to some embodiments, the angle weight factor can be set and the angle weight can be calculated using formula (2). Specifically, for example, the angle relationship can be converted into a smooth weight to reduce the sensitivity of the model in different fire environments and enhance adaptability. θ :

[0077]

[0078] Among them, θ is the angle between the spreading direction and the wind direction, θ0 is the angle with the highest weight, and k is a constant that controls the speed of weight change.

[0079] Among them, the Sigmoid function can avoid the drastic fluctuation of the weight factor when the angle changes, and convert the angle relationship into a smooth weight value, so that it changes slowly at small angles and changes quickly at large angles. By adjusting the constants θ0 and k, it can adapt to different fire environments and improve the flexibility of the model.

[0080] According to some embodiments, the spreading speed V and the angle weight factor W may be combined θ , the comprehensive score S of each grid cell is calculated using formula (3). The smoothed feature score matrix is ​​obtained to determine the fire line classification features.

[0081] S=V·W θ (3)

[0082] According to some embodiments, an adaptive density clustering method based on a feature score matrix can be used to automatically merge adjacent grid cells of the same type into fire head, fire wing, and fire tail regions. By applying the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, density clustering can be performed on the smoothed feature scores to automatically identify different types of connected regions.

[0083] According to some embodiments, where a grid unit may be, for example, a data point, the following two key parameters may be set:

[0084] Neighborhood radius∈: defines the neighborhood range of a data point, that is, the points within this radius are considered to be neighbors.

[0085] Minimum number of samples MinPts: specifies the minimum number of neighboring points required to form a density cluster.

[0086] The selection of these two key parameters can be optimized by cross-validation method based on the actual data distribution of the feature scoring matrix.

[0087] According to some embodiments, isolated points may be processed, specifically including: DBSCAN marks points that do not meet density requirements as noise points (isolated points), and analyzes the noise points.

[0088] According to some embodiments, the output result of the adaptive density clustering model is: the DBSCAN algorithm generates a set of connected regions, each region corresponds to a cluster and belongs to a single category (fire head, fire wing, fire tail). That is, a grid unit belongs to a single category, and a grid unit can correspond to a cluster, for example.

[0089] According to some embodiments, the method further comprises:

[0090] When it is determined that there is a grid cell subset that does not meet the density requirement in the grid cell set, obtaining a neighborhood range corresponding to any second grid cell in the grid cell subset;

[0091] Obtain the number of grid cells corresponding to any fire line classification feature within the neighborhood;

[0092] When the amount of data of the grid unit is less than a first quantity threshold, ignoring at least one grid unit within the neighborhood;

[0093] In the case where the amount of data in the grid unit is greater than the second quantity threshold, the fire line classification features corresponding to each grid unit in at least one grid unit within the neighborhood are obtained, and based on the fire line classification features corresponding to each grid unit, the fire line classification features corresponding to any second grid unit are obtained by majority voting, and the second quantity threshold is greater than or equal to the first quantity threshold. Therefore, the few misclassified or isolated noise points marked by the DBSCAN algorithm can be checked and adjusted to ensure the consistency of classification within the region, improve the reliability of the overall classification results, obtain clear category labels corresponding to each grid unit, characterize different parts of the fire line, and improve the accuracy of fire rescue tactics.

[0094] According to some embodiments, the second grid unit may be, for example, a grid unit in a subset of grid units that does not meet the density requirement. Wherein, when the grid unit is referred to as a data point, the second grid unit may be, for example, referred to as an isolated point.

[0095] In some embodiments, Figure 4 An example schematic diagram of a neighborhood voting method provided in an embodiment of the present application. Specifically, for example, it may include: determining a neighborhood range, wherein the neighborhood range may be, for example, the same neighborhood radius ∈ as the DBSCAN algorithm. The neighborhood range does not specifically refer to a fixed range. For example, when a modification instruction for the neighborhood range is received, the neighborhood range may also change accordingly.

[0096] According to some embodiments, the number of grid units corresponding to any fire line classification feature within the neighborhood range is obtained. For example, the number of data points of each category in the neighborhood of the isolated point can be counted. Specifically, for example, the number of data points of the fire head in the neighborhood of the isolated point can be counted, the number of data points of the fire wing in the neighborhood of the isolated point can be counted, and the number of data points of the fire tail in the neighborhood of the isolated point can be counted.

[0097] According to some embodiments, the majority voting principle may be adopted according to the statistical results to classify the isolated points into the category with the largest number in the neighborhood, and the category labels of the isolated points may be updated to the labels corresponding to the voting results.

[0098] According to some embodiments, for example, a processing strategy may be determined based on the number of grid cells corresponding to any of the hotline classification features. For example, when the number of grid cells is small and has little impact on the overall classification result, these points may be ignored and their noise labels may be retained. If the number of grid cells is large and may affect the continuity and accuracy of the classification result, the neighborhood voting method should be used to reclassify them.

[0099] Step 202, using a digital elevation model (DEM) to obtain terrain parameters of any grid cell;

[0100] According to some embodiments, the digital elevation model DEM may, for example, obtain DEM data of a study area to calculate terrain parameters.

[0101] In some embodiments, terrain parameters may include, for example, slope, aspect, curvature, flow direction, flow accumulation, and wind direction.

[0102] Slope and aspect: The slope and aspect of each grid cell can be calculated using DEM data.

[0103] Curvature: Calculate the curvature of the terrain surface, including profile curvature and plane curvature, reflecting the concave and convex characteristics of the terrain.

[0104] Flow direction and flow accumulation: Calculate flow direction and flow accumulation based on DEM data to identify valley areas.

[0105] Wind direction data: Obtain wind direction data at different times as auxiliary information for terrain classification.

[0106] According to some embodiments, a digital elevation model (DEM) may be used to acquire the terrain parameters of any grid cell.

[0107] Step 203, determining an order according to the terrain parameters and terrain features, and obtaining the terrain classification features of any first grid unit;

[0108] According to some embodiments, the order of determining terrain features can be used to indicate a preset order of determining terrain features. The order of determining terrain features does not specifically refer to a fixed order. For example, when the application scenario of the embodiment of the present application changes, the order of determining terrain features can also change accordingly.

[0109] In some embodiments, the order of determining terrain features may be, for example, the order of determining ridges, valleys, windward slopes, leeward slopes, and flat terrain.

[0110] 1) Ridge identification:

[0111] Ridges are usually characterized by areas with large elevation values ​​and positive curvature values. In order to ensure the accuracy of classification, a dynamic threshold model is established by calculating the local maximum value and standard deviation. Define the local maximum elevation z max and the local elevation standard deviation σ z , calculate the threshold Δz of the elevation difference threshold for:

[0112] Δz threshold = k × σ z (4)

[0113] Where k is an empirical coefficient used to adjust the sensitivity of elevation differences. When the elevation value meets the following conditions, it is determined to be a ridge:

[0114] z ≥ z max -Δz threshold And curvature>C positive

[0115] Among them, C positive is the threshold for positive curvature.

[0116] 2) Valley Identification:

[0117] Valleys usually have lower elevation values ​​and negative curvature, and higher flow accumulation values. In order to eliminate the influence of noise, the flow accumulation threshold A is used. threshold , and combined with elevation and curvature to make a judgment. The judgment conditions for valleys are:

[0118] And the curvature <C negative And A>A threshold

[0119] Among them, C negative is the threshold for negative curvature.

[0120] 3) Windward slope and leeward slope:

[0121] The determination of windward slope, leeward slope and crosswind slope is based on the angle difference between the slope direction and the wind direction. The specific process is as follows:

[0122] Calculation of slope aspect and wind direction difference:

[0123] Formula (5) is used to calculate the angle difference Δθ between the slope direction and the wind direction:

[0124] Δθ=|θ aspect -θ wind |(5)

[0125] Among them, θ aspect is the slope direction, θ wind is the wind direction. By setting the angle threshold, the windward slope and leeward slope can be determined separately.

[0126] 4) Flat terrain:

[0127] Flat terrain is limited by slope and flow accumulation:

[0128] S threshold且 A small

[0129] It is judged as flat terrain.

[0130] Step 204, obtaining a feature vector corresponding to any first grid cell according to the first feature vector corresponding to the fire line classification feature, the second feature vector corresponding to the terrain classification feature, and the third feature vector corresponding to the environmental feature of any first grid cell;

[0131] The specific process is as described above and will not be repeated here.

[0132] According to some embodiments, the grid unit can be called a node, for example, and the first feature vector is the fire line classification feature vector xifire: each fire line area is defined as a node vi, and the fire line feature is the fire head, fire wing, and fire tail obtained by the above extraction and classification, which are used as feature input after the classification label is generated.

[0133] According to some embodiments, the second feature vector is the terrain classification feature vector xiterrain: the feature vector includes the basic data of slope and aspect and the terrain classification features including ridges, valleys, windward slopes, leeward slopes, and flat terrain. Since the model research unit is the fire line area, the feature vector xiterrain contains the probability distribution of the terrain classification contained in the fire line area:

[0134]

[0135] According to some embodiments, the third feature vector, namely the environmental feature vector xienv, includes external environmental factors such as wind speed, wind direction, and temperature, to further enrich the classified feature information.

[0136] ​​The eigenvector corresponding to any first grid unit, that is, the comprehensive eigenvector of each node i, is expressed as:

[0137] Step 205: Use the target graph convolutional neural network model and the target time series Transformer model to analyze the feature vector corresponding to any of the first grid units to obtain the fire rescue tactics corresponding to any of the first grid units.

[0138] The specific process is as described above and will not be repeated here.

[0139] According to some embodiments, Figure 5 This is an example schematic diagram of a method for determining fire rescue tactics provided in an embodiment of the present application. The target graph convolutional neural network model and the target time series Transformer model are used to analyze the feature vector corresponding to any first grid unit to obtain the fire rescue tactics corresponding to any first grid unit, including:

[0140] Using a target graph convolutional neural network model, propagating the feature vector corresponding to any of the first grid cells forward to obtain the spatial features of the fire scene at each time step;

[0141] The target time series Transformer model is used to process the fire scene spatial features of each time step to obtain the position features of each time step;

[0142] According to the fire scene spatial characteristics of each time step and the position characteristics of each time step, obtaining the fire scene representation corresponding to any first grid unit;

[0143] According to the mapping relationship between the fire scene representation and the fire scene rescue tactics, the fire scene rescue tactics corresponding to the fire scene representation are obtained, and the fire scene rescue tactics are used as the fire scene rescue tactics corresponding to any of the first grid units. Therefore, the fire scene rescue tactics can be determined by combining the graph convolutional neural network model and the temporal Transformer model to improve the accuracy of the fire scene rescue display determination. Among them, GCN can effectively extract the spatial characteristics of the fire line area and capture the complex relationship between the fire line and the terrain, while Transformer models the temporal changes of fire spread through the self-attention mechanism, thereby realizing accurate modeling of dynamic changes during the fire spread process. This deep learning framework not only improves the accuracy and flexibility of the selection of tactics, but also automatically selects the most appropriate rescue strategy in a complex fire environment, significantly improving the efficiency of fire extinguishing.

[0144] In some embodiments, based on the mapping relationship between the fire scene representation and the fire scene rescue tactics, the fire scene rescue tactics corresponding to the fire scene representation are obtained. For example, the probability of each fire scene rescue tactic among at least one fire scene rescue tactic corresponding to the fire scene representation may be obtained, and the fire scene rescue tactic with the greatest probability may be used as the fire scene rescue tactic corresponding to any first grid unit.

[0145] According to some embodiments, the target graph convolutional neural network model is used to propagate forward the feature vector corresponding to any first grid unit to obtain the fire scene spatial features at each time step, including:

[0146] According to the spatial proximity relationship between the grid units, acquiring an adjacency matrix corresponding to any first grid unit;

[0147] According to the weight matrix of the target graph convolutional neural network model, the activation function of the target graph convolutional neural network model and the adjacency matrix, the fire scene spatial features at each time step are obtained. Therefore, the fire scene spatial features can be determined according to the weight matrix, activation function and adjacency matrix, the accuracy of determining the fire scene spatial features is improved, and the accuracy of determining the fire scene rescue tactics is improved.

[0148] According to some embodiments, the spatial proximity relationship between nodes is used to construct an adjacency matrix A of the graph, where:

[0149]

[0150] According to some embodiments, the method further comprises:

[0151] Acquire a fire scene training data set, wherein the fire scene training data set includes a fire line classification of each historical grid unit in a historical grid unit set corresponding to the historical fire line;

[0152] The fire line training data set is used to train the initial graph convolutional neural network model and the initial time series Transformer model to obtain the historical fire scene rescue tactics corresponding to each historical grid unit;

[0153] When the historical fire rescue tactics corresponding to each historical grid unit meets the tactical requirements, it is determined to obtain the target graph convolutional neural network model and the target target time series Transformer model.

[0154] According to some embodiments, the optimal tactics label yi corresponding to each node has a label set of {1, 2, ..., C}, where C is the number of tactics. The optimal tactics are obtained through rescue and disposal of historical fire cases.

[0155] In some embodiments, the training process for the model may include, for example:

[0156] 1. Input feature vector construction:

[0157] Historical fire line classification feature vector xifire: Each fire line area is defined as a node vi. The fire line features are the fire head, fire wing, and fire tail obtained by the above extraction and classification. After the classification label is generated, it is used as the feature input.

[0158] Historical terrain classification feature vector xiterrain: The feature vector includes the basic data of slope and aspect as well as the classification results of ridges, valleys, windward slopes, leeward slopes, and flat terrain generated by terrain classification labels. Since the model research unit is the fire line area, the feature vector xiterrain contains the probability distribution of the terrain classification contained in the fire line area:

[0159] Historical environmental feature vector xienv: includes external environmental factors such as wind speed, wind direction, and temperature, which further enriches the classified feature information.

[0160] 2. Feature vector representation:

[0161] The comprehensive historical feature vector of each node i is expressed as:

[0162] 3. Label data:

[0163] The optimal tactics label yi corresponding to each node, the label set is {1,2,…,C}, where C is the number of tactics. The optimal tactics are obtained through the rescue and disposal of historical fire cases.

[0164] 4. Adjacency matrix construction:

[0165] Using the spatial proximity relationship between nodes, the adjacency matrix A of the graph is constructed, where:

[0166]

[0167] 5.GCN layer:

[0168] For the node feature H(l) of the lth layer, the forward propagation formula of GCN is (6):

[0169]

[0170] in: is the adjacency matrix with self-loops added. yes The degree matrix of W (l) is the weight matrix of layer l. σ(·) is the activation function.

[0171] Among them, the initial input is H (0)=[x1,x2,…,x N ] T .

[0172] 2) Temporal dynamic modeling: Transformer model, wherein the Transformer model may also be called a Transformer encoder, and the Transformer model may include a multi-head attention mechanism, residual connection and normalization processing, and a feedforward neural network, etc.

[0173] During the fire rescue process, the changes in the fire scene are time-series, that is, the fire intensity changes over time and is affected by multiple factors (such as wind direction, terrain, etc.). In order to effectively capture these time-series dynamics, the model introduces the Transformer architecture, especially the self-attention mechanism, to process the time-series features in the fire scene data.

[0174] 1. Input feature sequence: The output of the GCN layer provides the spatial features of the fire scene at each time step, and these spatial feature sequences become the input of the Transformer model.

[0175] 2. Position encoding: In order to enable the model to process the time information in the time series data, we add position encoding to the input spatial features to help the model distinguish the features of different time steps. The position encoding formula (7) is as follows:

[0176]

[0177] Where t is the time step, d model is the dimension of the model and i is the dimension index.

[0178] 3. Attention mechanism: The self-attention mechanism captures long-range dependencies in fire scene data by calculating the relationship between the features of each position and the features of all other positions.

[0179] For the input sequence X = [x1, x2, ..., x N ] T , the self-attention is calculated as follows:

[0180] Linear change: Q = XW Q ,K=XW K ,V=XW V .

[0181] Among them, W Q ,W K ,W V is the learnable parameter matrix.

[0182] Attention weight calculation:

[0183] Among them, dk is the dimension of the key vector.

[0184] Multi-head attention:

[0185] By computing multiple attention heads in parallel, the results are then concatenated and linearly transformed:

[0186] MultiHead(Q,K,V)=Concat(head1,…,head h )W O

[0187] Each attention head i yes:

[0188] 3) Model fusion and output:

[0189] After extracting spatial features through GCN, the Transformer model will further process the temporal information, combining spatial features and temporal dynamics to generate a complete fire scene representation. This representation will be mapped to the final tactics category through a fully connected layer.

[0190] Final output: The output of the model is a probability distribution, which indicates the category of rescue tactics suitable for the current fire scene. The output layer uses the softmax activation function for classification, and the output categories include "direct attack", "indirect attack", etc.

[0191] y=softmax(H final W O )

[0192] Among them, H final is the final feature representation after Transformer processing, W O is the weight matrix of the output layer.

[0193] According to some embodiments, during the model training process, for example, a cross entropy loss function such as formula (8) may be used to optimize the prediction result:

[0194]

[0195] Among them, y i is the true label, p i is the probability of tactics predicted by the model.

[0196] Evaluation indicators: Use accuracy, recall, F1 score and other indicators to evaluate model performance, and optimize the generalization ability of the model through cross-validation and hyperparameter tuning.

[0197] In some embodiments, the historical fire rescue tactics corresponding to each historical grid unit meet the tactical requirements. For example, based on the optimal tactics label yi and the label set corresponding to each node, when the consistency between the fire rescue tactics output by the model and the optimal tactics label is greater than the consistency threshold, it is determined that the trained model can be applied to the determination of actual fire rescue tactics.

[0198] In some embodiments,

[0199] Among them, fire rescue tactics may include at least one of the following:

[0200] 1) Direct attack: Extinguish the flames directly at the front of the fire line.

[0201] 2) Indirect attack: Prevent the spread of fire by setting up control lines and fire isolation zones.

[0202] 3) Fire isolation zone: Establish a fire isolation zone in the direction of the fire to control the spread of the fire.

[0203] 4) Breakthrough at one point, advance on both wings: Concentrate strength to break through one point on the fire line, and then advance on both wings. It is suitable for the fire head area where the fire line is narrow.

[0204] 5) Two wings advancing in a pincer attack: Advance from both wings of the fire line toward the fire head at the same time to form a pincer attack. This is suitable for a wide fire line and a large fire head area.

[0205] 6) Breakthrough at multiple points and annihilate the fire in sections: Firefighting is carried out at multiple points on the fire line at the same time to extinguish the fire in sections. This is suitable for fires with scattered heads and long fire lines.

[0206] 7) Interweaving, detouring, and progressive overtaking: bypassing the weak points of the fire line and fighting from the rear or flank. Applicable to situations where the fire is intense in front of the fire line, the wind direction is unfavorable, the terrain is complex, and the weak points of the fire line are obvious.

[0207] 8) Use support to fight fire with fire: Use the burned area to control the direction of new fire heads and form a situation of fighting fire with fire. It is suitable for areas where a large area has been burned, forming obvious open space or low fuel area, terrain is conducive to fire fighting, and wind direction is favorable for fire fighting.

[0208] 9) Preset isolation to prevent forest fire: Preset isolation belts in the direction of fire spread to prevent the fire from spreading further. Applicable to the protected target on the fire path, there is enough time to set up isolation belts before the fire arrives.

[0209] 10) Ground-air coordination and three-dimensional firefighting: Ground and air forces work together to achieve all-round firefighting. Applicable to areas where ground forces cannot quickly reach the core of the fire.

[0210] 11) Do not fight fires on the mountain (corresponding to the slope), fight from the back or flank: For areas with large terrain slopes, avoid frontal firefighting and fight from the flank or back. This is applicable to areas with large terrain slopes and relatively gentle terrain on the flanks or back.

[0211] In one or related embodiments, a digital elevation model DEM is used to obtain the terrain parameters of any one of the grid cells; an order is determined according to the terrain parameters and terrain features to obtain the terrain classification features of any one of the first grid cells. Therefore, the terrain classification features can be determined in an order according to the terrain parameters and terrain features, which can reduce the time for determining the terrain classification features, improve the accuracy and efficiency of determining the terrain classification features, and improve the accuracy of determining the fire rescue tactics.

[0212] In order to implement the above-mentioned embodiment, the present application also proposes a fire rescue tactics acquisition device.

[0213] Figure 6 A schematic diagram of the structure of a fire scene rescue tactics acquisition device provided in an embodiment of the present application.

[0214] like Figure 6 As shown, the fire rescue tactics acquisition device includes:

[0215] A feature acquisition unit 601 is used to acquire a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell;

[0216] A vector acquisition unit 602 is used to acquire a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell;

[0217] The tactical acquisition unit 603 is used to analyze the feature vector corresponding to any of the first grid units using a target graph convolutional neural network model and a target time series Transformer model to obtain the fire rescue tactics corresponding to any of the first grid units.

[0218] Further, in a possible implementation of the embodiment of the present application, the feature acquisition unit 601 is used to acquire the fire line classification feature of any first grid unit in the set of grid units corresponding to the fire line, and is specifically used to:

[0219] Using a cellular automaton model to identify the fire line, obtaining fire line spread data of any first grid cell in a set of grid cells corresponding to the fire line, wherein the fire line spread data includes a spread speed and a spread direction;

[0220] Obtaining the angle between the spreading direction and the current environmental wind direction, and obtaining the angle weight of the angle;

[0221] According to the spreading speed and the angle weight, obtaining a feature scoring matrix corresponding to any one of the first grid units;

[0222] According to the feature scoring matrix of each grid unit in the grid unit set and the adaptive density clustering algorithm, the fire line classification feature of any first grid unit is obtained.

[0223] Furthermore, in a possible implementation of the embodiment of the present application, the feature acquisition unit 601 is further configured to:

[0224] When it is determined that there is a grid cell subset that does not meet the density requirement in the grid cell set, obtaining a neighborhood range corresponding to any second grid cell in the grid cell subset;

[0225] Obtain the number of grid cells corresponding to any fire line classification feature within the neighborhood;

[0226] When the amount of data of the grid unit is less than a first quantity threshold, ignoring at least one grid unit within the neighborhood;

[0227] When the amount of data in the grid unit is greater than a second quantity threshold, the fire line classification features corresponding to each grid unit in at least one grid unit within the neighborhood are obtained, and based on the fire line classification features corresponding to each grid unit, the majority voting principle is adopted to obtain the fire line classification features corresponding to any second grid unit, and the second quantity threshold is greater than or equal to the first quantity threshold.

[0228] Further, in a possible implementation of the embodiment of the present application, the feature acquisition unit 601, when used to acquire the terrain classification feature of any first grid unit, is specifically used to:

[0229] Using a digital elevation model (DEM) to obtain terrain parameters of any grid cell;

[0230] An order is determined according to the terrain parameters and terrain features, and a terrain classification feature of any first grid unit is obtained.

[0231] Further, in a possible implementation of the embodiment of the present application, the tactic acquisition unit 603 is used to analyze the feature vector corresponding to any first grid unit using the target graph convolutional neural network model and the target time series Transformer model to obtain the fire rescue tactics corresponding to any first grid unit, specifically for:

[0232] Using a target graph convolutional neural network model, propagating the feature vector corresponding to any of the first grid cells forward to obtain the spatial features of the fire scene at each time step;

[0233] The target time series Transformer model is used to process the fire scene spatial features of each time step to obtain the position features of each time step;

[0234] According to the fire scene spatial characteristics of each time step and the position characteristics of each time step, obtaining the fire scene representation corresponding to any first grid unit;

[0235] According to the mapping relationship between the fire scene representation and the fire scene rescue tactics, the fire scene rescue tactics corresponding to the fire scene representation are obtained, and the fire scene rescue tactics are used as the fire scene rescue tactics corresponding to any one of the first grid units.

[0236] Further, in a possible implementation of the embodiment of the present application, the tactical acquisition unit 603 is used to adopt the target graph convolutional neural network model to propagate the feature vector corresponding to any first grid unit forward to obtain the fire scene spatial features of each time step, specifically for:

[0237] According to the spatial proximity relationship between the grid units, acquiring an adjacency matrix corresponding to any first grid unit;

[0238] According to the weight matrix of the target graph convolutional neural network model, the activation function of the target graph convolutional neural network model and the adjacency matrix, the spatial characteristics of the fire scene at each time step are obtained.

[0239] Furthermore, in a possible implementation of the embodiment of the present application, the tactic acquisition unit 603 is further configured to:

[0240] Acquire a fire scene training data set, wherein the fire scene training data set includes a fire line classification of each historical grid unit in a historical grid unit set corresponding to the historical fire line;

[0241] The fire line training data set is used to train the initial graph convolutional neural network model and the initial time series Transformer model to obtain the historical fire scene rescue tactics corresponding to each historical grid unit;

[0242] When the historical fire rescue tactics corresponding to each historical grid unit meets the tactical requirements, it is determined to obtain the target graph convolutional neural network model and the target target time series Transformer model.

[0243] It should be noted that the aforementioned explanation of the fire scene rescue tactics acquisition method embodiment is also applicable to the fire scene rescue tactics acquisition device of this embodiment, and will not be repeated here.

[0244] The fire rescue tactics acquisition device provided by the present application is used to acquire the fire line classification feature of any first grid unit in the set of grid units corresponding to the fire line and the terrain classification feature of any first grid unit through a feature acquisition unit; a vector acquisition unit is used to acquire the feature vector corresponding to any first grid unit according to the first feature vector corresponding to the fire line classification feature, the second feature vector corresponding to the terrain classification feature and the third feature vector corresponding to the environmental feature of any first grid unit; a tactics acquisition unit is used to analyze the feature vector corresponding to any first grid unit by using a target graph convolutional neural network model and a target time series Transformer model to acquire the fire rescue tactics corresponding to any first grid unit. The problem that fire remediation methods cannot be adaptively changed, resulting in a mismatch between fire remediation methods and the current fire situation and poor fire extinguishing effect is solved. Since the fire rescue tactics can be determined by fire line classification characteristics, terrain classification characteristics and current environmental characteristics, the interaction between terrain changes and environmental factors during the spread of fire can be considered, and the real-time changes of the fire process can be adaptively adjusted. The fire rescue tactics can be determined by combining the graph convolutional neural network model and the time series Transformer model. The fire rescue tactics can be determined by combining spatial and time series information, which can improve the matching of fire rescue tactics with the current fire situation, improve the accuracy of the determination of fire rescue tactics, optimize the allocation of fire rescue resources, reduce fire fighting time, and improve fire extinguishing efficiency.

[0245] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0246] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0247] In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0248] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0249] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0250] This application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, this application is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, risks can be minimized by limiting data collection and deleting data. In addition, when applicable, personal identifiers are removed from such personal information to protect the privacy of users.

[0251] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0252] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0253] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0254] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0255] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0256] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0257] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0258] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for acquiring fire rescue tactics, characterized in that: include: Obtaining a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell; Acquire a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell; A target graph convolutional neural network model and a target time series Transformer model are used to analyze the feature vector corresponding to any first grid unit to obtain the fire rescue tactics corresponding to any first grid unit.

2. The method according to claim 1, characterized in that The obtaining of the fire line classification feature of any first grid cell in the set of grid cells corresponding to the fire line includes: Using a cellular automaton model to identify the fire line, obtaining fire line spread data of any first grid cell in a set of grid cells corresponding to the fire line, wherein the fire line spread data includes a spread speed and a spread direction; Obtaining the angle between the spreading direction and the current environmental wind direction, and obtaining the angle weight of the angle; According to the spreading speed and the angle weight, obtaining a feature scoring matrix corresponding to any one of the first grid units; According to the feature scoring matrix of each grid unit in the grid unit set and the adaptive density clustering algorithm, the fire line classification feature of any first grid unit is obtained.

3. The method according to claim 2, characterized in that The method further comprises: When it is determined that there is a grid cell subset that does not meet the density requirement in the grid cell set, obtaining a neighborhood range corresponding to any second grid cell in the grid cell subset; Obtain the number of grid cells corresponding to any fire line classification feature within the neighborhood; When the amount of data of the grid unit is less than a first quantity threshold, ignoring at least one grid unit within the neighborhood; When the amount of data in the grid unit is greater than a second quantity threshold, the fire line classification features corresponding to each grid unit in at least one grid unit within the neighborhood are obtained, and based on the fire line classification features corresponding to each grid unit, the majority voting principle is adopted to obtain the fire line classification features corresponding to any second grid unit, and the second quantity threshold is greater than or equal to the first quantity threshold.

4. The method according to claim 1, characterized in that: The obtaining of the terrain classification feature of any first grid cell includes: Using a digital elevation model (DEM) to obtain terrain parameters of any grid cell; An order is determined according to the terrain parameters and terrain features, and a terrain classification feature of any first grid unit is obtained.

5. The method according to claim 1, characterized in that The target graph convolutional neural network model and the target time series Transformer model are used to analyze the feature vector corresponding to any first grid unit to obtain the fire rescue tactics corresponding to any first grid unit, including: Using a target graph convolutional neural network model, propagating the feature vector corresponding to any of the first grid cells forward to obtain the spatial features of the fire scene at each time step; The target time series Transformer model is used to process the fire scene spatial features of each time step to obtain the position features of each time step; According to the fire scene spatial characteristics of each time step and the position characteristics of each time step, obtaining the fire scene representation corresponding to any first grid unit; According to the mapping relationship between the fire scene representation and the fire scene rescue tactics, the fire scene rescue tactics corresponding to the fire scene representation are obtained, and the fire scene rescue tactics are used as the fire scene rescue tactics corresponding to any one of the first grid units.

6. The method according to claim 5, characterized in that The target graph convolutional neural network model is used to propagate the feature vector corresponding to any first grid unit forward to obtain the fire scene spatial features at each time step, including: According to the spatial proximity relationship between the grid units, acquiring an adjacency matrix corresponding to any first grid unit; According to the weight matrix of the target graph convolutional neural network model, the activation function of the target graph convolutional neural network model and the adjacency matrix, the spatial characteristics of the fire scene at each time step are obtained.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: Acquire a fire scene training data set, wherein the fire scene training data set includes a fire line classification of each historical grid unit in a historical grid unit set corresponding to the historical fire line; The fire line training data set is used to train the initial graph convolutional neural network model and the initial time series Transformer model to obtain the historical fire scene rescue tactics corresponding to each historical grid unit; When the historical fire rescue tactics corresponding to each historical grid unit meets the tactical requirements, it is determined to obtain the target graph convolutional neural network model and the target target time series Transformer model.

8. A fire rescue tactics acquisition device, characterized in that: include: A feature acquisition unit, used to acquire a fire line classification feature of any first grid cell in a set of grid cells corresponding to the fire line and a terrain classification feature of any first grid cell; a vector acquisition unit, configured to acquire a feature vector corresponding to any first grid cell according to a first feature vector corresponding to the fire line classification feature, a second feature vector corresponding to the terrain classification feature, and a third feature vector corresponding to the environmental feature of any first grid cell; The tactical acquisition unit is used to analyze the feature vector corresponding to any of the first grid units by using a target graph convolutional neural network model and a target time series Transformer model to obtain the fire rescue tactics corresponding to any of the first grid units.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images