Lightweight traditional village fire spreading prediction method, system, equipment and medium

Abstract

The invention discloses a lightweight traditional village fire spreading prediction method, system, equipment and medium, and the method comprises the following steps: making a traditional village fire spreading data set and verification set, and endowing a building with combustible characteristic attributes and combustion environment characteristic attributes; by using the data set, based on a dichotomy logistic regression analysis model, screening out main influence factors of fire spreading with significant correlation, and determining a fire spreading probability calculation formula; building a neighborhood-based fire spreading simulation mode, formulating a fire spreading judgment standard, and further forming a lightweight fire spreading prediction model; and gradually calculating the building fire spreading probability in the neighborhood of the on-fire building, judging the path direction of fire spreading, and calculating and visualizing the fire spreading sequence and the key building. The method can be used for simulating fire spreading among traditional village building groups, predicting a fire spreading sequence and detecting key buildings influencing fire spreading, and a scientific basis is provided for deployment of traditional village fire-fighting facilities.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional village protection based on geographic information data, and in particular to a lightweight traditional village fire spread prediction method, system, electronic equipment and storable medium. Background Art

[0002] Due to the dense layout of traditional village buildings, a large amount of wood in the building structure, and the storage of flammable items indoors, there is a high risk of fire. Once a fire breaks out, the fire will spread rapidly, especially when there is wind outside, which is easy to cause a disaster of "burning the camp", causing irreparable and huge losses to the historical and cultural value of the village, the safety of residents' lives and economic development. Fires between buildings in traditional villages occur in a complex combustion environment, interacting with external factors such as people, village characteristics, geographical environment and other intervening factors, resulting in the coexistence of certainty and randomness in the combustion and spread process. Predicting the spread of fire between buildings is of great significance to fire prevention and control in traditional villages.

[0003] At present, domestic and foreign scholars have conducted a lot of scientific research on the prediction of fire spread between buildings, with two main research directions. One direction is to measure the changes of various relevant values ​​during fire spread based on actual experiments, and to restore the physical characteristics of the research object to the greatest extent; the other direction is to calculate and simulate based on the principles of fire dynamics. The current prediction of fire spread in traditional villages still has the following problems: (1) No overall fire spread prediction method for traditional village building complexes has been proposed. The existing technology can only predict whether a fire spread path will occur between two buildings. (2) No fire spread simulation method that comprehensively considers the fire spread conditions in traditional villages has been proposed. The existing technology mainly simulates the changes in a single influencing factor, simplifies the characteristics of the traditional village pattern and terrain environmental factors, and cannot comprehensively consider the complex combustion properties of traditional villages; (3) No lightweight fire spread probability calculation and spread simulation method has been proposed. The existing technology considers redundant factors, the calculation method is complex, and the time and manpower required are high; Therefore, there is an urgent need to propose a lightweight fire spread prediction method that comprehensively considers the complex combustion properties of traditional villages, grasps the key influencing factors and constructs a probability calculation formula, so as to efficiently evaluate the fire spread path in traditional villages, determine key building nodes, and guide the deployment of subsequent fire protection strategies. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned defects in the prior art and provide a lightweight traditional village fire spread prediction method, system, electronic device and storable medium. The prediction method collects the fire spread related data of traditional villages, uses a binary logistic regression analysis model to screen the main influencing factors of fire spread with significant correlation, and determines the fire spread probability calculation formula, constructs a neighborhood-based fire spread simulation model, forms a lightweight fire spread prediction model, and after simulating and predicting the target traditional village, determines the path direction of the fire spread, and calculates and visualizes the fire spread sequence and key buildings. This method can be used for fire spread simulation of traditional villages in various regions, simplifies the calculation steps, and improves the accuracy and automation of the overall fire spread assessment of traditional villages.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a lightweight traditional village fire spread prediction method, the positioning method comprising the following steps: S1. Create a traditional village fire spread data set and verification set: import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, where the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability step by step in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, and then the simulation stops; S5. Establish fire spread judgment criteria: In the same calculation stage, the fire order of buildings in the neighborhood of the same burning building that are not on fire and have a probability P ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths of traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

[0006] Furthermore, the index layer factors included in the traditional village fire spread data set T and the validation set Y in step S1 are based on the principle of fire dynamics as the logical basis, and are finally determined and formed by referring to the existing studies on fire spread in traditional villages and the influencing factors of a large number of traditional villages where fire spread disasters have occurred. Fire dynamics is a science that studies the basic principles and laws of fire occurrence, development and control. Fire dynamics shows that the occurrence of fire requires three elements: combustibles, oxidants and ignition sources, which exist at the same time, that is, combustion will occur. Since in traditional villages, the spread of fire between buildings occurs in open spaces, and oxygen as a oxidant is always sufficient, the characteristics of building combustibles and the characteristics of the combustion environment are the main reasons affecting the spread of fire in traditional villages. Fire dynamics explains that the spread of fire is divided into two processes: the spread of fire within the building and the spread of fire between buildings. Among them, the spread of fire between buildings is a complex process dominated by heat radiation, and multiple pathways such as heat conduction, heat convection and flying fire coexist.

[0007] The present invention aims to predict the sequence relationship of fire spread in traditional villages and explore the probability of fire spreading between buildings. Therefore, the constructed traditional village fire spread data set T and validation set Y contain two types of values: building combustible characteristic attributes and combustion environment characteristic attributes. The data set T mostly contains one type of value: actual disaster situation. The actual disaster situation is a binary variable. If the statistical target building is ignited by the burning building, it is marked as 1, and if the statistical target building is not ignited by the burning building, it is marked as 0; the building combustible characteristic attributes are composed of four indicator layer factors: fixed fire load density Qi , mobile fire load density Q , building surface area S f , Building opening area S 0 The combustion environment characteristic attributes are composed of two index layer factors: building proximity D and building relative wind direction position R; among them, building relative wind direction position R is a categorical variable, and the other five index layer factors are continuous variables, and the values ​​are obtained from actual observation and measurement. The specific statistical standards are as follows: ①Fixed fire load density QiIt is defined as: the fire load density of building structures and components, which is determined by the materials and structure of the building, where the reference type values ​​are: reinforced concrete structure: 0MJ / ㎡, masonry structure: 200MJ / ㎡, rammed earth structure: 300MJ / ㎡, brick-wood structure / stone-wood structure: 1000MJ / ㎡, all-wood structure: 1500MJ / ㎡, grass-wood structure: 2000MJ / ㎡; ②Mobile fire load density Q It is defined as: the total fire load density of indoor items in a building, which is determined by the function of the building, with reference types as follows: office buildings: 400 MJ / ㎡, public exhibition buildings: 500 MJ / ㎡, residential buildings: 780 MJ / ㎡, leisure and entertainment buildings: 1000 MJ / ㎡, commercial buildings: 1500 MJ / ㎡, educational buildings: 2500 MJ / ㎡; ③Building surface area S f It is defined as: the sum of the building's facade and roof area, obtained by actual measurement; ④Building opening area S 0 It is defined as: the sum of the areas of doors, windows and openings on the building facade, obtained by actual measurement; ⑤ The building proximity D is defined as: the closest spatial distance between the target building facade and the burning building facade, which is obtained by actual measurement and is the square root of the sum of the square of the horizontal spacing value and the square of the vertical spacing value of the facade; ⑥ The relative wind direction position R of the building is defined as: under a certain wind direction condition, the position relationship between the target building and the burning building, when it is in the upwind direction it is recorded as 1, and when it is in the downwind direction it is recorded as 0; in both the data set T and the validation set Y, the wind direction on the day when the fire occurred is used as the benchmark for statistics. When simulating and predicting traditional villages that are not affected by the fire, the annual average wind direction of the target village is used as the benchmark for statistical calculations.

[0008] Furthermore, in step S1, in order to avoid accidental results when using the binary logistic regression analysis model to screen the main influencing factors of fire spread and determine the fire spread probability calculation formula, samples with more than 20 times the number of independent variables should be selected for regression analysis. The present invention has a total of 6 independent variables, so it is required that the collected data set T contains ≥2 traditional villages and ≥120 traditional village buildings.

[0009] Furthermore, in step S2, the binary logistic regression analysis model selected is a multivariate statistical analysis model, which is applicable to binary response variables, that is, the dependent variable can only be represented by the numbers 0 and 1. The binary logistic regression analysis model can construct a scientific mathematical analysis approach between the independent variable and the dependent variable, and determine the correlation between the independent variable and the dependent variable. The calculation formula is as follows: Formula (1) In the formula, y =1 indicates that an event has occurred, P ( y =1) represents the probability value of an event occurring; x i is the ith independent variable, i=1,2,3,…n; n is the total number of independent variables, Based on regression calculation x i The corresponding logistic regression coefficients; Therefore, the present invention screens the main influencing factors of fire spread, and the process includes: The actual damage situation of traditional village buildings in the data set T formed in step S1 is used as the dependent variable, and the fire load density of the six index layer factors with continuous values ​​is used as the dependent variable. Qi , mobile fire load density Q , building surface area S f , Building opening area S 0 The building proximity D is taken as a continuous variable, and the building relative wind direction position R, which takes a categorical value, is taken as a virtual variable. They are included in the calculation formula for analysis to calculate the probability of fire spread when the target building is ignited by the burning building. It is stipulated that in the comprehensive test of the coefficients of the binary logistic regression model, when the significance value of the model is <0.005, the binary logistic regression model is statistically significant. When the significance value of the independent variable is ≤0.05, it has a significant correlation, which means that this independent variable can be used as the main influencing factor of fire spread and included in the construction of the subsequent fire spread probability calculation formula. Otherwise, the independent variable needs to be eliminated.

[0010] Furthermore, in step S3, the binary logistic regression analysis model can calculate and evaluate the contribution and significant correlation of each independent variable to the change of the dependent variable, and integrate the relationship strength between the predicted variable and the predicted result into an interval expression of 0 to 1 through the maximum likelihood method, thereby constructing a corresponding probability formula. The probability formula threshold is usually set to 0.5, that is, samples with P≥0.5 are classified as "1" corresponding to the occurrence of the event, and samples with P<0.5 are classified as "0" corresponding to the non-occurrence of the event. The present invention calculates the correlation coefficient corresponding to the influencing factor and constructs a fire spread probability calculation formula, and the process is as follows: The main influencing factors of fire spread screened in S2 are put into the binary logistic regression model as independent variables, and the actual disaster situation is used as the dependent variable to calculate the probability of fire spread when the target building is ignited by the burning building. The calculation formula is the same as the above formula (1), where the independent variable x iCorresponding to the main influencing factors of fire spread obtained by screening, the fire spread probability calculation formula is constructed according to formula (1), and the fire spread probability formula threshold is defined as 0.5. The constructed formula is as follows: Formula (2) In the formula, y =1 means a building is ignited by a burning building, P ( y =1) indicates the probability value of a building being ignited by a burning building; is the logistic regression coefficient corresponding to the main influencing factors of fire spread obtained based on regression calculation, i=1,2,3,…n, n is the total number of independent variables, n=6.

[0011] Furthermore, in step S4, the spread of fire between buildings is a process of heat transfer in multiple ways, among which thermal radiation is the main heat transfer way. The characteristic of thermal radiation heat transfer is that the heat energy generated by the fire is transmitted from the burning building to the outside in the form of electromagnetic waves, and weakens with the increase of distance, so thermal radiation heat transfer has a certain range of influence. According to a large number of research results and actual disaster analysis, when the distance between buildings exceeds 10m, it is basically difficult for the fire to spread between buildings. In order to ensure the rigor of the research, the present invention introduces the concept of neighborhood, and defines the neighborhood as a buffer zone formed by the outer contour of the burning building expanding outward by a distance of 10m. Using the concept of neighborhood to divide the spread process into multiple stages of development can ensure the extraction of the time series of the spread, while avoiding the increase in calculation amount due to considering the difference in the time for complete combustion of the building. The specific process is as follows: The spread of fire between buildings is simplified as gradually expanding the fire-affected area in stages. The building neighborhood range is defined as the farthest range of each burning building when the fire spreads directly outward in each stage. The buffer zone formed by expanding 10m outward from the outer contour of the burning building is the building neighborhood. When a building is on fire, the fire spread neighborhood caused by the burning building is formed, and it is further judged whether the buildings located in the neighborhood of the burning building meet the conditions of not being on fire, such as Figure 2 As shown; if there is no fire yet, calculate the probability of fire spreading from the burning building to the unburned building according to formula (2); If only part of a building is in the neighborhood of the burning building, the building will also be regarded as a potential target for the fire to spread to the burning building. It is necessary to determine whether the building is on fire. If the conditions are met, the probability of fire spread of the building in this neighborhood will be calculated. If there are multiple unburned buildings with a spread probability P ≥ 0.5 in a neighborhood, they are defined as being ignited in the same calculation phase, that is, when a building in the neighborhood of a burning building is ignited, the fire spread neighborhood calculation for the next phase is already generated at the same time, forming the neighborhood of the burning building in the new calculation phase; If a certain unburned building is in the neighborhood of two burning buildings at a certain calculation stage, the spread probability of the unburned building in the neighborhood of the two burning buildings is calculated respectively, and it is determined whether the fire of the unburned building will be spread by the two burning buildings at the same time; In the building neighborhood of the new calculation phase, if a building has been spread in the previous calculation phase, the spread probability calculation is not repeated, that is, it has been determined that it does not meet the condition of not yet on fire; If there is no burning building in the building neighborhood in the new calculation stage or the fire spread probability of the building P is less than 0.5, the simulation stops.

[0012] Furthermore, since heat radiation is the main heat transfer pathway in the process of fire spreading outward from a burning building, for multiple buildings that will be ignited in the same building neighborhood at the same calculation stage, they have a sequence due to different probabilities of being spread by fire, and the one with a larger probability has a higher spread sequence. Therefore, in order to distinguish the order of ignition of buildings ignited at the same calculation stage, in step S5, in the same calculation stage, the buildings that simultaneously meet the two conditions of not ignited and fire spread probability P≥0.5 are compared in size, and the one with a larger probability has a spread sequence at the front, and the one with a smaller probability has a spread sequence at the back; If there are multiple neighborhoods of burning buildings in the same calculation stage, the spreading sequence of the buildings that meet the two conditions of no fire and fire spread probability P ≥ 0.5 in the neighborhood of the burning building at the front of the spreading sequence in the previous stage will be ranked first.

[0013] Furthermore, the spread of fire between buildings is not a process in which a single building affects another building. A burning building can cause multiple buildings that meet the conditions to catch fire. An unburned building can receive heat energy from multiple burning buildings and be ignited by multiple buildings at the same time. Buildings that generate more fire spread paths play a more critical role in the spread of fire in traditional villages. In step S6, when the probability of fire spread of an unburned building in the neighborhood of a certain burning building is P ≥ 0.5, a spread path is marked between the unburned building and the burning building. After the fire spread prediction simulation stops, the number of fire spread paths generated by traditional village buildings is counted, and a fire spread sequence diagram of the target traditional village and a distribution diagram of key buildings with multiple spread paths are drawn; If there are n unburned buildings with probability P ≥ 0.5 in the neighborhood of the same burning building, n outward spreading paths are formed; If a non-burning building is in the neighborhood of n burning buildings at a certain calculation stage, and the corresponding spread probability P ≥ 0.5, then there are n spread paths pointing to the building.

[0014] In a second aspect, the present invention provides a lightweight traditional village fire spread prediction system, which is used to execute the lightweight traditional village fire spread prediction method, and the high-speed communication system includes: The data set making module is used to import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, wherein the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; The influencing factor screening module is used to put the data set T into a binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; The fire spread probability calculation module is used to put the main fire spread influencing factors with significant correlation into a binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct a fire spread probability calculation formula; The fire spread simulation model construction module is used to determine the scope of the building neighborhood, calculate the fire spread probability in stages, and define the buildings that meet the probability P ≥ 0.5 as the burning buildings in the next calculation stage, forming the building neighborhood of the new calculation stage. The probability calculation of the unburned buildings in the building neighborhood of the new calculation stage is performed until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building P < 0.5, and the simulation stops; The fire spread judgment standard formulation module is used for the buildings that are not on fire and have a probability P ≥ 0.5 in the same calculation stage and in the neighborhood of the same burning building. The fire order is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. The data visualization module is used to visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths generated by traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

[0015] In a third aspect, the present invention provides an electronic device, comprising a processor and a memory for storing programs executable by the processor, wherein when the processor executes the program stored in the memory, the above-mentioned lightweight traditional village fire spread prediction method is implemented.

[0016] In a fourth aspect, the present invention provides a storage medium storing a program, which, when executed by a processor, implements the above-mentioned lightweight traditional village fire spread prediction method.

[0017] Compared with the prior art, the present invention has the following advantages and beneficial effects: (1) The method of the present invention proposes an overall fire spread prediction method for traditional village building complexes, and comprehensively predicts the space and time of fire occurrence. By calculating the fire spread probability of the target unburned building and the burned building in stages, it can be determined whether a fire spread path from the burned building to the target unburned building will be generated. By calculating the fire spread probability at different calculation stages in turn, the fire spread sequence of the traditional village as a whole can be predicted. On the one hand, this method can clarify the spatial direction of fire spread at different calculation stages, and on the other hand, by counting the number of spread paths generated by each building, it can determine the key buildings that affect the spread of fire, laying a theoretical foundation for realizing accurate fire prevention and control in traditional villages.

[0018] (2) The method of the present invention comprehensively considers the complex combustion properties of traditional villages to determine the main influencing factors. Based on the principle of fire dynamics and the complex combustion properties of traditional villages, the index layer factors are sorted out to ensure that the selected factors fully cover the potential disaster factors of fire spread between buildings in traditional villages. Based on the binary logistic regression analysis model, the main influencing factors affecting the spread of fire in traditional villages are screened out. The fire spread probability calculation formula retains the main influencing factors of fire spread with significant correlation, eliminates non-critical factors, and increases the accuracy and scientificity of the fire spread probability prediction.

[0019] (3) The present invention proposes a lightweight fire spread probability calculation formula and fire spread simulation model. The fire spread probability calculation formula based on the binary logistic regression analysis model proposed in this method condenses the main influencing factors of the fire spread probability calculation. The fire spread simulation model constructed by this method uses the neighborhood as the minimum spread unit in each calculation stage, which simplifies the calculation steps of the fire spread sequence while ensuring the efficiency and effect of the fire spread model. The present invention can reduce redundant calculation steps, save manpower and material resources, realize efficient evaluation of the fire spread sequence of traditional villages, and improve the efficiency of fire protection deployment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 It is a flow chart of a lightweight traditional village fire spread prediction method disclosed in the present invention; Figure 2 It is a schematic diagram of a neighborhood-based fire spread pattern disclosed in the present invention; Figure 3Schematic diagram of six calculation stages of fire spread prediction in Example 1 of the present invention; Figure 4 Schematic diagram of fire spread prediction result and actual situation in Example 1 of the present invention; Figure 5 is a distribution diagram of key buildings in multiple spreading paths in Example 1 of the present invention; Figure 6 Schematic diagram of six calculation stages of fire spread prediction in Example 2 of the present invention; Figure 7 is a schematic diagram of fire spread prediction results in Example 2 of the present invention; Figure 8 is a distribution map of key buildings in multiple spreading paths in Example 2 of the present invention; Figure 9a This is a diagram of the actual fire spread sequence in Jinzhuzhuang Village. Figure 9b is a schematic diagram of the prediction result of the actual fire spread sequence of Jinzhuzhuang Village in Example 1 of the present invention, Fig.9c This is a schematic diagram of the prediction results of the actual fire spread sequence of Jinzhuzhuang Village in Example 2 of the present invention; Fig.10 is a structural block diagram of a lightweight traditional village fire spread prediction system in Example 3 of the present invention; Fig.11 It is a structural block diagram of an electronic device in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.

[0023] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments.

[0024] Example 1 Figure 1 This is a flow chart of a lightweight traditional village fire spread prediction method disclosed in the present invention. This embodiment discloses a lightweight traditional village fire spread prediction method, comprising the following steps: S1. Create a traditional village fire spread data set and verification set: import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, where the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; In the above step S1, this embodiment first uses CAD software to map the building blocks and village terrain of Wengding Village, Xiaozhai Village and Jinzhu Zhuang Village, which are traditional villages affected by large-scale fires in the mountainous area of ​​southern China, and imports them into the geographic information processing software GIS platform, and assigns building fire spread related attributes. The data set T contains two traditional village samples, Wengding Village and Xiaozhai Village, with a total of 557 buildings. The building assignment includes 6 indicator factors of actual disaster situation, building combustible characteristic attributes and combustion environment characteristic attributes, specifically: actual disaster situation, fixed fire load density Qi , mobile fire load density Q , building surface area S f , Building opening area S 0 , as well as the building proximity D, the building relative wind direction position R; the validation set Y contains a traditional village sample of Jinzhu Zhuangzhai, with a total of 61 buildings, including 6 indicator factors that constitute the characteristic attributes of building combustibles and combustion environment, specifically: fixed fire load density Qi , mobile fire load density Q , building surface area S f , Building opening area S 0 , building proximity D, building relative wind direction position R, the partial fire spread data set T is shown in Table 1, and the partial fire spread verification set Y is shown in Table 2.

[0025] Table 1. Fire spread data set T data table of some examples 1

[0026] Table 2. Data table of fire spread verification set Y in some examples 1

[0027] S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; In the above step S2, the data set T in step S1 is put into a binary logistic regression analysis model, the actual disaster situation is used as the dependent variable, and the six indicator layer factors are used as independent variables to calculate the probability of fire spread when the target building is ignited by the burning building, wherein the calculation formula is as follows: Formula (1) In the formula, y =1 indicates that an event has occurred, P ( y =1) represents the probability value of an event occurring; x i is the ith independent variable, i=1,2,3,…n; n is the total number of independent variables, Based on regression calculation x i The corresponding logistic regression coefficients; In this embodiment, χ is calculated. 2 =131.241, the significance value of the model is <0.005, the binary logistic regression analysis model is statistically significant, and the independent variable factor mobile fire load density Q , building surface area S f The significance value of is greater than 0.05, and there is no significant correlation in the analysis (Table 3). Therefore, these two indicator layer factors are eliminated and not included in the construction of the subsequent fire spread probability calculation formula.

[0028] Table 3. Binary logistic regression analysis results of fire spread data in traditional villages

[0029] S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; In the above step S3, the fire spread probability calculation formula is determined, and the process includes: 1. The main influencing factors of fire spread screened in step S2 are put into the binary logistic regression analysis model as independent variables, and the actual disaster situation is used as the dependent variable to calculate the probability of fire spread when the target building is ignited by the burning building. The calculation formula is the same as the above formula (1), where: Based on regression calculation x i Corresponding logistic regression coefficients; independent variables x i Corresponding to the main influencing factors of fire spread obtained through screening, the logistic regression coefficients of each independent variable are calculated as shown in Table 4; Table 4. Binary logistic regression analysis results of the main influencing factors of fire spread

[0030] 2. According to formula (1), the fire spread probability calculation formula is constructed as follows: Formula (3) In the formula, y =1 means a building is ignited by a burning building, P ( y =1) indicates the probability of a building being ignited by a burning building.

[0031] S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability step by step in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, and then the simulation stops; In the above step S4, the fire spread simulation mode is staged spread, which simplifies the fire spread process between buildings into a staged outward expansion. The building neighborhood range is defined as the farthest range of each burning building when the fire spreads directly outward in each stage, and the buffer zone extending 10m outward from the outer contour of the burning building is formed as the building neighborhood; when a building is on fire, the fire spread neighborhood caused by the burning building is formed, and it is further judged whether the buildings located in the neighborhood of the burning building meet the condition that they have not caught fire; if they have not caught fire, the fire spread probability from the burning building to the unburned building is calculated according to formula (3); among them, if a building is only partially located in the neighborhood of the burning building, the building is also regarded as a potential target for the fire spread of the burning building, and it is necessary to judge whether the building is on fire. If the condition is met, the fire spread probability of the building in the neighborhood is calculated; If there are multiple unburned buildings with a spread probability P ≥ 0.5 in a neighborhood, they are defined as being ignited in the same calculation stage, that is, when the buildings in the neighborhood of the burning building are ignited, the fire spread neighborhood calculation of the next stage has been generated at the same time, forming the neighborhood of the burning building in the new calculation stage; if an unburned building is in the neighborhood of two burning buildings at the same time in a certain calculation stage, the spread probability of the unburned building in the neighborhood of the two burning buildings is calculated respectively, and it is judged whether the fire of the unburned building will be spread by the two buildings at the same time; in the building neighborhood of the new calculation stage, if a building has been spread in the previous calculation stage, the spread probability calculation is not repeated, that is, it has been determined that it does not meet the condition of not yet on fire; if there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is P < 0.5, the simulation stops. In this embodiment, the actual burning building in Jinzhu Zhuangzhai in 2017 of the verification set Y is set as the burning building in the prediction model, and the fire spread prediction is performed, forming 6 calculation stages, and a total of 25 buildings are spread by fire ( Figure 3 ). Through statistical analysis of the model's predicted values ​​and actual values, the accuracy of the prediction for buildings that were ignited was 75.7%, the accuracy of the prediction for buildings that were not ignited was 83.8%, and the accuracy of the prediction for all building samples was 79.4%.

[0032] S5. Establish fire spread judgment criteria: In the same calculation stage, the fire spread order of buildings in the neighborhood of the same burning building that are not on fire and have a probability ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. In the above step S5, the size of the buildings that meet the two conditions of no fire and fire spread probability P≥0.5 in the same building neighborhood in the same calculation stage is compared, and the one with a larger probability is placed in the front of the spread sequence, and the one with a smaller probability is placed in the back of the spread sequence; if there are multiple neighborhoods of burning buildings in the same calculation stage, the spread sequence of the buildings that meet the above two conditions in the neighborhood of the burning building with a higher spread sequence in the previous stage is ranked first. After the fire spread prediction is performed in this embodiment, the 25 buildings spread by the fire are sorted, and the predicted spread sequence is basically the same as the actual spread sequence, the number of buildings spread by the fire is the same, and there is only a difference in the local spread sequence.

[0033] S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths of traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths; In the above step S6, when the probability of fire spread of an unburned building in the neighborhood of a burned building is P≥0.5, a spread path is marked between the unburned building and the burned building. After the fire spread prediction simulation stops, the number of fire spread paths generated by the traditional village buildings is counted, and the fire spread sequence diagram of the target traditional village and the key building distribution diagram of multiple spread paths are drawn; if there are n unburned buildings with a probability of P≥0.5 in the neighborhood of the same burned building, n outward spreading paths are formed; if a certain unburned building is in the neighborhood of n burned buildings at the same time in a certain calculation stage, and the corresponding spread probability P≥0.5, there are n spread paths pointing to the building. In this embodiment, the direction of the fire spread path and the sequence of building fires are marked, the number of fire spread paths of the traditional village buildings is counted, and the fire spread sequence diagram of the target traditional village is obtained ( Figure 4 ) and key building distribution map ( Figure 5 ), determine that there are 2 key buildings that generate more than 3 spread paths.

[0034] Example 2 Based on steps S1 to S6 of the above embodiment 1, this embodiment further discloses a method for predicting the spread of fire in a lightweight traditional village, including the following steps: S1, this step refers to step S1 in embodiment 1; S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; In the above step S2, the calculation result of this embodiment is the same as that of embodiment 1, and the obtained value is 2 =131.241, the significance value of the model is <0.005, and the logistic regression model is statistically significant. Q , building surface area S f The significance value is greater than 0.05, which is not significantly correlated in the analysis, but the mobile fire load density Q , building surface area S f The correlation coefficient is small (Table 5), which has little impact on the probability calculation, so it is also included in the construction of the subsequent probability calculation formula.

[0035] Table 5. Binary logistic regression analysis results of the six indicator layer factors

[0036] S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; In the above step S3, the fire spread probability calculation formula is determined, and the process includes: the six indicator layer factors and their correlation coefficients in step S2 are used to construct a fire spread probability model according to formula (1), and the constructed model is as follows: Formula (4) In the formula, y =1 means a building is ignited by a burning building, P ( y =1) indicates the probability of a building being ignited by a burning building.

[0037] S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability in stages according to formula (4), define the building with probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, then the simulation stops; In the above step S4, this embodiment sets the actual fire buildings in Jinzhuzhuang Village in 2017 in the validation set Y as the fire buildings in the prediction model, and performs fire spread prediction, forming 6 calculation stages, with a total of 18 buildings being spread by fire ( Figure 6 ). Through statistical analysis of the model's predicted values ​​and actual values, the accuracy of the prediction for buildings that were ignited was 65.2%, the accuracy of the prediction for buildings that were not ignited was 91.9%, and the accuracy of the prediction for all building samples was 81.7%.

[0038] S5. Establish fire spread judgment criteria: In the same calculation stage, the fire spread order of buildings in the neighborhood of the same burning building that are not on fire and have a probability ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. In the above step S5, after the fire spread prediction is performed in this embodiment, the 18 buildings affected by the fire are sorted. The predicted spread sequence is quite different from the actual spread sequence. The fire spread prediction of the 8 buildings in the northern area of ​​Jinzhu Zhuangzhai is opposite to the actual situation.

[0039] S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths of traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths; In the above step S6, this embodiment marks the direction of the fire spreading path and the sequence of the buildings on fire, counts the number of fire spreading paths of the traditional village buildings, and obtains the fire spreading sequence diagram of the target traditional village ( Figure 7 ) and key building distribution map ( Figure 8 ), determine whether there is a key building that generates more than 3 spread paths.

[0040] The method of this embodiment is roughly the same as that of the provided embodiment 1, and the main difference is that when determining the fire spread probability calculation formula, the index layer factors that do not have significant correlation are not eliminated, and the 6 index layer factors and their correlation coefficients are included in the formula. The calculation results of embodiment 1 and embodiment 2 are listed as shown in Table 6. The accuracy of embodiment 1 in predicting the samples of ignited buildings is 75.7%, and the accuracy of embodiment 2 in predicting the samples of unignited buildings is 65.2%. From the data, although the accuracy of embodiment 2 for all types of predictions is higher than that of embodiment 1, its prediction of the probability of building fire spread is generally low, resulting in a low accuracy rate in predicting that the building will be ignited. This result will cause the fire intervention measures to fail to fully cover buildings with the risk of fire spread, so the operation of embodiment 1 is better than embodiment 2.

[0041] Table 6. Data comparison table of different embodiments

[0042] Screening the index layer factors based on the correlation with the actual fire damage situation will affect the simplicity and accuracy of the constructed fire spread probability calculation formula. Although the index layer factors that do not have significant correlation are not eliminated and directly included in the consideration of formula construction, it can ensure that no information is lost, but it may cause a large number of redundant or irrelevant features in the data set, and the model has noise interference, resulting in overfitting of the probability model and failure to accurately predict the probability. By comparing Example 1 with the actual fire spread prediction diagram (the actual result and the prediction result comparison diagram are shown in FIG. Figure 9a , Figure 9b and Fig.9c As shown in the figure, it is found that when determining the fire spread probability calculation formula in Example 2, since the indicator layer factors that do not have significant correlation are not eliminated, the possibility of fire spread is over-judged in the actual prediction, which is far from the actual fire spread situation. The probability of the predicted building being spread is too small, and a large number of spread paths and spread buildings are missing, resulting in inaccurate fire spread prediction results.

[0043] Example 3 Reference Fig.10 This embodiment provides a lightweight traditional village fire spread prediction system, which includes a data set preparation module 301, an influencing factor screening module 302, a spread probability calculation module 303, a fire spread simulation model construction module 304, a fire spread judgment standard formulation module 305 and a data visualization module 306, which are sequentially connected. The data set making module 301 is used to import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, wherein the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; The influencing factor screening module 302 is used to put the data set T into a binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; The fire spread probability calculation module 303 is used to put the main fire spread influencing factors with significant correlation into a binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct a fire spread probability calculation formula; The fire spread simulation mode construction module 304 is used to determine the building neighborhood range, calculate the fire spread probability in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the unburned buildings in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building P < 0.5, and the simulation stops; The fire spread judgment standard formulation module 305 is used for the buildings that are not on fire and have a probability P ≥ 0.5 in the same calculation stage and in the neighborhood of the same burning building. The fire order is determined according to the probability. The buildings with a large probability are ranked first in the spread sequence, and the buildings with a small probability are ranked last in the spread sequence. The data visualization module 306 is used to visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths generated by traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the key building distribution diagram of multiple spread paths.

[0044] Example 4 This embodiment provides an electronic device, which may be a computer. Fig.11 As shown, a processor 402, a memory, an input device 403, a display 404 and a network interface 405 connected via a system bus 401 are provided. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium 406 and an internal memory 407. The non-volatile storage medium 406 stores an operating system, a computer program and a database. The internal memory 407 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the processor 402 executes the computer program stored in the memory, a lightweight traditional village fire spread prediction method proposed in the above-mentioned embodiment 1 is implemented. The lightweight traditional village fire spread prediction method includes the following steps: S1. Create a traditional village fire spread data set and verification set: import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, where the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability step by step in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, and then the simulation stops; S5. Establish fire spread judgment criteria: In the same calculation stage, the fire order of buildings in the neighborhood of the same burning building that are not on fire and have a probability P ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths of traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

[0045] Example 5 This embodiment provides a storage medium, which is a computer-readable storage medium, storing a computer program. When the computer program is executed by a processor, a lightweight traditional village fire spread prediction method proposed in the above embodiment 1 is implemented. The lightweight traditional village fire spread prediction method includes the following steps: S1. Create a traditional village fire spread data set and verification set: import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, where the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability step by step in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, and then the simulation stops; S5. Establish fire spread judgment criteria: In the same calculation stage, the fire order of buildings in the neighborhood of the same burning building that are not on fire and have a probability P ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths of traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

[0046] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0047] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A lightweight traditional village fire spread prediction method, characterized in that: The prediction method comprises the following steps: S1. Create a traditional village fire spread data set and verification set: import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, where the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; S2. Screening the main influencing factors of fire spread: Put the data set T into the binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; S3. Determine the fire spread probability calculation formula: put the main fire spread influencing factors with significant correlation into the binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct the fire spread probability calculation formula; S4. Construct a neighborhood-based fire spread simulation model: determine the building neighborhood range, calculate the fire spread probability step by step in stages, define the building that meets the probability P ≥ 0.5 as the burning building in the next calculation stage, form the building neighborhood of the new calculation stage, and perform probability calculation on the buildings that are not on fire in the building neighborhood of the new calculation stage until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building is less than 0.5, and then the simulation stops; S5. Establish fire spread judgment criteria: In the same calculation stage, the fire order of buildings in the neighborhood of the same burning building that are not on fire and have a probability P ≥ 0.5 is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. S6. Data visualization: Visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths generated by traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

2. The method for predicting the spread of fire in a lightweight traditional village according to claim 1 is characterized in that: In step S1, the buildings in the data set T are assigned actual disaster conditions, building combustible characteristic attributes, and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned building combustible characteristic attributes and combustion environment characteristic attributes, wherein the actual disaster conditions are binary variables, and if the statistical target building is ignited by the burning building, it is marked as 1, and if the statistical target building is not ignited by the burning building, it is marked as 0; The characteristic properties of building combustibles are composed of four index layer factors: fixed fire load density Qi , mobile fire load density Q , building surface area S f , Building opening area S 0 The combustion environment characteristic attributes are composed of two index layer factors: building proximity D and building relative wind direction position R; among them, building relative wind direction position R is a categorical variable, and the other five index layer factors are continuous variables, and the values ​​are obtained from actual observation and measurement. The specific statistical standards are as follows: ①Fixed fire load density Qi It is defined as: the fire load density of building structures and components, which is determined by the materials and structure of the building, where the reference type values ​​are: reinforced concrete structure: 0MJ / ㎡, masonry structure: 200MJ / ㎡, rammed earth structure: 300MJ / ㎡, brick-wood structure / stone-wood structure: 1000MJ / ㎡, all-wood structure: 1500MJ / ㎡, grass-wood structure: 2000MJ / ㎡; ②Mobile fire load density Q It is defined as: the total fire load density of indoor items in a building, which is determined by the function of the building, with reference types of values: office buildings: 400 MJ / ㎡, public exhibition buildings: 500 MJ / ㎡, residential buildings: 780 MJ / ㎡, leisure and entertainment buildings: 1000 MJ / ㎡, commercial buildings: 1500 MJ / ㎡, educational buildings: 2500 MJ / ㎡; ③Building surface area S f It is defined as: the sum of the building's facade and roof area, obtained by actual measurement; ④Building opening area S 0 It is defined as: the sum of the areas of doors, windows and openings on the building facade, obtained by actual measurement; ⑤ The building proximity D is defined as: the closest spatial distance between the target building facade and the burning building facade, which is obtained by actual measurement and is the square root of the sum of the square of the horizontal spacing value and the square of the vertical spacing value of the facade; ⑥ The relative wind direction position R of the building is defined as: under a certain wind direction condition, the position relationship between the target building and the burning building, when it is in the upwind direction, it is recorded as 1, and when it is in the downwind direction, it is recorded as 0; in both the data set T and the validation set Y, the wind direction on the day of the fire incident is used as the benchmark for statistics, and when simulating and predicting traditional villages that are not affected by the fire, the annual average wind direction of the target village is used as the benchmark for statistical calculation; The dataset T contains ≥2 traditional villages and ≥120 buildings.

3. The method for predicting the spread of fire in a lightweight traditional village according to claim 2 is characterized in that: In step S2, the calculation formula of the binary logistic regression analysis model is as follows: Formula (1) In the formula, y =1 indicates that an event has occurred, P ( y =1) indicates the probability value of an event occurring; x i is the ith independent variable, i=1,2,3,…n; n is the total number of independent variables, Based on regression calculation x i The corresponding logistic regression coefficients; Put the data set T formed in step S1 into formula (1), take the actual disaster situation as the dependent variable and the six indicator layer factors as the independent variables, and calculate the probability of fire spread when the target building is ignited by the burning building; It is stipulated that in the comprehensive test of the coefficients of the binary logistic regression analysis model, when the significance value of the model is <0.005, the binary logistic regression model is statistically significant. When the significance value of the independent variable is ≤0.05, it has a significant correlation, which means that this independent variable can be used as the main influencing factor of fire spread and included in the construction of the subsequent fire spread probability calculation formula. Otherwise, the independent variable is eliminated.

4. The method for predicting the spread of fire in a lightweight traditional village according to claim 3 is characterized in that: The process of determining the fire spread probability calculation formula in step S3 is as follows: The main influencing factors of fire spread screened in step S2 are put into the binary logistic regression analysis model as independent variables, and the actual disaster situation is used as the dependent variable to calculate the probability of fire spread when the target building is ignited by the burning building. The calculation formula is the same as the above formula (1), where the independent variable x i Corresponding to the main influencing factors of fire spread obtained by screening, the fire spread probability calculation formula is constructed according to formula (1), which is as follows: Formula (2) In the formula, y =1 means a building is ignited by a burning building, P ( y =1) indicates the probability value of a building being ignited by a burning building; is the logistic regression coefficient corresponding to the main influencing factors of fire spread obtained based on regression calculation, i=1,2,3,…n, n is the total number of independent variables, n=6.

5. The method for predicting the spread of fire in a lightweight traditional village according to claim 1 is characterized in that: The fire spread simulation mode in step S4 is staged spread, which simplifies the fire spread process between buildings into a staged outward expansion. The building neighborhood range is defined as the farthest range of each burning building when the fire spreads directly outward in each stage, and the buffer zone formed by expanding 10m outward from the outer contour of the burning building is the building neighborhood; When a building is on fire, a fire spreading neighborhood caused by the burning building is formed, and further judgment is made as to whether the buildings located in the neighborhood of the burning building meet the condition of not being on fire; If there is no fire, the probability of fire spreading from the burning building to the unburned building is calculated according to formula (2). If a building is only partially in the neighborhood of the burning building, the building is also regarded as a potential target for the fire to spread to the burning building. It is necessary to determine whether the building is on fire. If the conditions are met, the probability of fire spreading of the building in this neighborhood is calculated. If there are multiple unburned buildings with a spread probability P ≥ 0.5 in a neighborhood, they are defined as being ignited in the same calculation phase, that is, when a building in the neighborhood of a burning building is ignited, the fire spread neighborhood calculation for the next phase is already generated at the same time, forming the neighborhood of the burning building in the new calculation phase; If a certain unburned building is in the neighborhood of two burning buildings at a certain calculation stage, the spread probability of the unburned building in the neighborhood of the two burning buildings is calculated respectively, and it is determined whether the fire of the unburned building will be spread by the two burning buildings at the same time; In the building neighborhood of the new calculation phase, if a building has been spread in the previous calculation phase, the spread probability calculation is not repeated, that is, it has been determined that it does not meet the condition of not yet on fire; If there is no burning building in the building neighborhood in the new calculation stage or the fire spread probability of the building P is less than 0.5, the simulation stops.

6. The method for predicting the spread of fire in a lightweight traditional village according to claim 1 is characterized in that: In step S5, the buildings that meet the two conditions of no fire and fire spread probability P ≥ 0.5 in the same building neighborhood in the same calculation stage are compared in size, and the buildings with a larger probability of fire spread are placed in the front of the spread sequence, and the buildings with a smaller probability of fire spread are placed in the back of the spread sequence; If there are multiple neighborhoods of burning buildings in the same calculation stage, the spreading sequence of the buildings that meet the two conditions of no fire and fire spread probability P ≥ 0.5 in the neighborhood of the burning building at the front of the spreading sequence in the previous stage will be ranked first.

7. The method for predicting the spread of fire in a lightweight traditional village according to claim 1 is characterized in that: In step S6, when the probability of fire spread of an unburned building in the neighborhood of a burned building is P≥0.5, a spread path is marked between the unburned building and the burned building. After the fire spread prediction simulation is stopped, the number of fire spread paths generated by the traditional village buildings is counted, and a fire spread sequence diagram of the target traditional village and a distribution diagram of key buildings with multiple spread paths are drawn; If there are n unburned buildings with probability P ≥ 0.5 in the neighborhood of the same burning building, n outward spreading paths are formed; If a non-burning building is in the neighborhood of n burning buildings at a certain calculation stage, and the corresponding spread probability P ≥ 0.5, then there are n spread paths pointing to the building.

8. A lightweight traditional village fire spread prediction system, used to execute the lightweight traditional village fire spread prediction method according to any one of claims 1 to 7, characterized in that: The fire spread prediction of lightweight traditional villages includes: The data set making module is used to import the surveyed traditional village CAD building block plan into the geographic information processing software GIS platform to form a data set T and a verification set Y, wherein the buildings in the data set T are assigned with actual disaster conditions, building combustible characteristic attributes and combustion environment characteristic attributes, and the buildings in the verification set Y are assigned with building combustible characteristic attributes and combustion environment characteristic attributes; The influencing factor screening module is used to put the data set T into a binary logistic regression analysis model to screen out the main influencing factors of fire spread with significant correlation; The fire spread probability calculation module is used to put the main fire spread influencing factors with significant correlation into a binary logistic regression analysis model, calculate the correlation coefficients corresponding to the influencing factors, and construct a fire spread probability calculation formula; The fire spread simulation model construction module is used to determine the scope of the building neighborhood, calculate the fire spread probability in stages, and define the buildings that meet the probability P ≥ 0.5 as the burning buildings in the next calculation stage, forming the building neighborhood of the new calculation stage. The probability calculation of the unburned buildings in the building neighborhood of the new calculation stage is performed until there is no burning building in the building neighborhood of the new calculation stage or the fire spread probability of the building P < 0.5, and the simulation stops; The fire spread judgment standard formulation module is used for the buildings that are not on fire and have a probability P ≥ 0.5 in the same calculation stage and in the neighborhood of the same burning building. The fire order is determined according to the probability. The buildings with a large probability will be ranked first in the spread sequence, and the buildings with a small probability will be ranked last in the spread sequence. The data visualization module is used to visualize the fire spread sequence, mark the direction of the fire spread path and the sequence of building fires, count the number of fire spread paths generated by traditional village buildings, and obtain the fire spread sequence diagram of the target traditional village and the distribution map of key buildings with multiple spread paths.

9. An electronic device comprising a processor and a memory for storing a program executable by the processor, characterized in that: When the processor executes the program stored in the memory, it implements the lightweight traditional village fire spread prediction method described in any one of claims 1 to 7.

10. A storage medium storing a program, characterized in that: When the program is executed by a processor, the lightweight traditional village fire spread prediction method described in any one of claims 1 to 7 is implemented.

Images