A fire spread simulation method for forest-urban interface based on spatiotemporal semantics

Abstract

The present invention relates to the field of semantic modeling technology, and focuses on a method for simulating the spread of fire in forest-urban interface areas based on spatiotemporal semantics driving. In view of the complexity of fire spread in forest-urban interface areas (WUI), this method innovatively introduces a spatiotemporal semantic driving mechanism on the basis of traditional fire spread models, and combines the regional characteristics of WUI to solve the shortcomings of traditional simulation methods in terms of accuracy, timeliness and adaptability to complex environments. The method includes: a WUI fire spread model based on cellular automata, the construction of a WUI fire scenario simulation knowledge meta-model and knowledge graph, and the construction of a fire spatiotemporal semantic query and simulation framework. By integrating semantic information in the time and space dimensions, the present invention can more accurately simulate the spread path and speed of fire under different terrains, building distributions and meteorological conditions, provide a more reliable decision-making basis for fire prevention, emergency response and resource allocation, and effectively improve the prevention and control capabilities of fires in forest-urban interface areas.

Description

Technical Field

[0001] The present invention relates to the technical field of semantic modeling and fire simulation, and in particular to a method for simulating fire spread in the forest-town boundary area based on spatiotemporal semantics. Background Art

[0002] With the acceleration of urbanization, the scale of forest-urban interfaces (WUIs) continues to expand globally. Due to their unique geographical location, WUIs are often located at the intersection of natural vegetation and human settlements, making them a key area where forest fire and urban fire risks overlap. In recent years, climate change and the frequent occurrence of extreme weather events have further exacerbated the probability and severity of WUI fires.

[0003] The spread of these fires typically exhibits complex spatiotemporal dynamics, including multidimensional propagation paths, uneven burning rates, and high sensitivity to human activity. Traditional fire spread modeling methods are mostly based on physical simulations or empirical models, often relying on various assumptions and mathematical principles to simulate fire spread behavior. However, these methods ignore the complexity of spatiotemporal interactions during fire spread, which limits the applicability and accuracy of models, particularly in WUI fires, which are highly dependent on the built environment and human activity.

[0004] Existing fire spread modeling methods, when applied to WUI fires, fail to fully consider the critical role of human activities and the built environment in fire propagation, making it difficult for the models to dynamically capture the complex spatiotemporal interactions of fire spread. Furthermore, in emergencies, physical fire spread models suffer from long response times and poor flexibility, limiting their application in rapidly changing disaster environments. These methods fail to meet the timeliness and flexibility demands of modern emergency response and fire management. To address this, we propose a fire spread simulation method for the forest-urban interface, driven by spatiotemporal semantics. Summary of the Invention

[0005] This paper proposes a fire spread method for the forest-urban interface based on spatiotemporal semantics. Based on semantic modeling technology, this method deeply integrates multi-source heterogeneous data with mathematical models, constructs a semantic fire knowledge graph, and introduces spatiotemporal semantic information to intelligently analyze the key scenarios and driving mechanisms of fire spread. It aims to integrate multi-source spatiotemporal data, complete intelligent fire spread modeling, support emergency response and dynamic decision-making, and provide scientific basis and technical support for fire risk assessment, early warning and emergency response.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics, the method comprising:

[0007] The WUI fire spread model based on cellular automata is constructed. The cellular automata model, namely cellularautomata CA, is used to discretize the fire spread process. Based on the previous research, the knowledge of forest fire and building fire spread simulation is integrated. Different cellular states, state transformation rules and time evolution are carefully divided and summarized. The WUI fire spread simulation model based on cellular automata is proposed.

[0008] The WUI fire scenario simulation knowledge meta-model was constructed. The spatiotemporal elements involved in the WUI fire spread process, their attributes, and mapping relationships were sorted out. The knowledge meta-framework and attribute model were defined from five dimensions: emergencies, hazard factors, disaster-prone environments, disaster-bearing carriers, and emergency management. Based on the WUI fire spread model simulation process, the mapping relationships of spatiotemporal elements involved in WUI fires were sorted out to construct a knowledge meta-relationship model. The knowledge meta-framework, attributes, and relationship models were integrated, and the knowledge related to WUI fire simulation was integrated and expressed using geographic cells as knowledge meta-units, completing the construction of the WUI fire scenario simulation knowledge meta-model.

[0009] The WUI fire scenario simulation ontology was constructed based on the WUI fire scenario simulation knowledge meta-model and the SSN, TIME, and GEOSPARQL ontology standards published by W3C. It integrates WUI fire domain knowledge with spatiotemporal information of sudden fire events. It defines a WUI fire element monitoring sub-ontology for monitoring and updating spatiotemporal data, and a WUI fire knowledge meta-ontology for storing and visualizing fire scenario knowledge. This integration then forms a complete WUI fire scenario simulation knowledge system. Finally, Protégé software was used to complete the ontology modeling of the conceptual model, logical foundation, and specification constraints, realizing the construction of the WUI fire scenario simulation ontology.

[0010] The WUI fire knowledge graph was constructed using the Neo4j graph database for integration and formal display of ontology associations. By processing multi-source fire data, the multi-source information of the experimental area was integrated and dynamically updated. The instance nodes of relevant geographic elements and knowledge elements in the WUI fire scenario were created and updated, and the WUI fire scenario simulation knowledge graph, namely WUIFFSSKG, was constructed.

[0011] FireSTS, a fire spatiotemporal semantic query and simulation framework, uses a relational database to store spatial geometry information and a graph database to store attribute data. It also builds a JDBC driver that integrates access interfaces for both relational and graph databases, along with WUI fire spread simulation and multi-source information integration methods. With a graph-side-led execution model and a collaborative execution model between the relational and graph sides, FireSTS converts hybrid queries that integrate graph semantics and spatiotemporal computing into executable subqueries and simulations for GDBMS and RDBMS through the JDBC driver.

[0012] Semantic-driven simulation uses Cypher language in the WUI fire knowledge graph to complete the setting of fire simulation scenarios such as ignition point, simulation time and simulation conditions, and calls the WUI fire spread simulation method encapsulated in FireSTS to realize semantic-driven WUI fire spread simulation. The simulation results are updated to the corresponding knowledge unit node attributes in the knowledge graph, and output as an EXCEL file with a single cell as a record.

[0013] Preferably, in the WUI fire spread model method based on cellular automata, the state of the cells changes as the fire spreads, and the cell state characteristics are as follows:

[0014] When the cell state = 0, that is, the cell state S0: the cell is in the fire isolation area, there is no combustible material in the cell, and the cell state remains unchanged;

[0015] When the cell state = 1, that is, the cell state S1: the combustible material in the cell has not been ignited, and the cell is in a combustible but not burned state;

[0016] When the cell state = 2, that is, the cell state S2: the combustible material in the cell has just started to burn and has not been completely burned. The cell is in an internal combustion state and has not ignited the power of other cells.

[0017] When the cell state = 3, that is, the cell state S3: the combustible material in the cell is completely burned, the cell is in a completely burned state, begins to spread outward and has the ability to ignite the remaining cells;

[0018] When the cell state = 4, that is, the cell state S4: the combustible material in the cell has been burned, the cell is in an extinguished state, and has no ability to ignite the remaining cells;

[0019] Preferably, the rule for converting the cell state from S0 to S1 defines cells covering fire-isolated areas such as water areas and roads as S0, and cells covering non-fire-isolated areas such as forests and grasslands as S1; the rule for converting the cell state from S1 to S2 defines that the combustible but unburned S1 state cell is ignited by the surrounding fully burned S3 state cells, and the cell is converted from the combustible but unburned S1 state to the internal combustion state S2; since the WUI area is the boundary between the built-up area and the forest area, the spread mechanism of forest fire and building fire is different. This paper divides the ignition mode into four cases according to the internal spread mechanism of different ground fires, namely:

[0020] Case 1: Forest cells spread to forest cells. The burning rate R of forest cells is calculated using Wang Zhengfei's forest fire spread model, and the probability of forest cells catching fire at different times is calculated based on the changes in the ignition state of adjacent cells. When the probability of fire Greater than the forest cell fire threshold P FrThat is, it is ignited, and the cells in the S1 state are ignited and converted into S2;

[0021] Case 2: Spread from building cell to building cell; the cell is affected by the adjacent burning house, that is, the completely burned S3 state building cell. It is necessary to first calculate the contribution rate P of a single burning house to the ignition of the cell. mn , and then calculate the probability of fire of the cell due to the cumulative influence of the surrounding burning houses I n , when the probability of cell fire is greater than the building cell fire threshold P Br that is, to be ignited;

[0022] Case 3: Spread from forest cells to building cells; when a building cell is within the influence range of the forest cell in state S3, the burning probability In of the building cell is calculated. When the burning probability is greater than the threshold P Br that is, ignited;

[0023] Case 4: Spread from building cells to forest cells, when the forest cell is in the influence range P of the building cell in state S3 Amn Inside, it can be ignited;

[0024]

[0025] in is the cumulative value of the ignition transfer from the adjacent cells to the current cell after the next time interval Δt; is the spreading rate of cells in state S3 with spreading capability in the Moore neighborhood of the current cell at time t; Δt is the simulation time step and is set to 1 min, L is the horizontal distance between cells (m), and i and j are the coordinate indices of the current cell.

[0026] Preferably, the transition rule from the cell state S2 to S3 defines the time process of the cell from the internal combustion S2 state to the complete combustion S3 state as Δt1. When the combustion duration of the cell in the S2 state is t cost When Δt1 is greater than Δt1, the cell state changes from S2 to S3. Because the internal spread mechanisms of building fire and forest fire are different, the internal combustion time of forest cells in this paper is calculated by the ratio of cell burning distance and rate. The development of house fire is affected by many factors. According to the experience of fire experts, this process is simplified to the time t from the beginning of combustion to complete combustion of building cells. 23 ;

[0027]

[0028] where R IN is the internal forest fire spread rate of the cell. The S2 state cell is converted to S3 state after Δt1 and has the ability to spread to the neighborhood.

[0029] Preferably, the transition rule of the cell state S3 to S4 defines the time process of the cell from the complete combustion S3 state to the extinguished S4 state as Δt2, and the Δt2 depends on the continuous combustion of the fuel, external fire extinguishing measures and environmental conditions. The flame combustion duration is related to the time Δt required for the complete combustion of the fuel. burn and fire extinguishing intervention time Δt inter Related;

[0030] Δt2=Δt burn +Δt inter ;

[0031]

[0032] Where W is the fuel load within the cell and is expressed in kg / m 2 , m is the mass burning rate of the combustible material, and the unit is kg / (m 2 min); t 34 It is the time from complete combustion to extinguishment of building cells, obtained from the experience of fire experts, and the unit is min.

[0033] Preferably, the WUI fire scenario simulation knowledge meta-model covers the definitions of the fire knowledge meta-framework model, attribute model and relationship model.

[0034] Preferably, the WUI fire scenario simulation knowledge meta-ontology includes a definition of a WUI fire element monitoring sub-ontology for realizing spatiotemporal data monitoring and updating of sudden fire events; and a definition of a WUI fire knowledge meta-ontology for realizing inductive storage and visualization of fire scenario knowledge.

[0035] Preferably, the storage and mapping of spatiotemporal data in the fire spatiotemporal semantic query and simulation framework stores the text attributes of the spatiotemporal data in GDMS and the geographic attributes in RDBNS. The entity identifier ID in GDBM is used to associate with the corresponding entity in RDBMS, and the entity class in GDBM is used to associate with the corresponding table name in RDBMS.

[0036] Preferably, the implementation and calling of the fire spatiotemporal semantic query and simulation framework encapsulates the corresponding fire simulation method and multi-source data processing method, as well as the interface between the relational database and the graph database in a Java Maven project, compresses the Java Maven project into a Jar file and loads it into the neo4j graph database, and uses the Cypher language in neo4j to complete the calling of its encapsulated method.

[0037] Preferably, the semantically driven simulation creates and modifies scenario simulation instance nodes in the WUI fire scenario simulation knowledge graph through the Cypher language, completes the setting of fire simulation scenarios such as ignition point and simulation time, and calls the fire spread simulation method encapsulated in the FireSTS framework in neo4j by inputting the node name of the simulation scenario and the end time parameters of the simulation, thereby realizing the semantically driven WUI fire spread simulation.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. This invention integrates multi-source information and WUI fire domain knowledge, uses knowledge graphs as the information organization form, and implements fire spread simulation modeling based on semantic-driven fusion of multi-source heterogeneous data. It can update and process real-time meteorological data and adjust the fire spread simulation, thereby enhancing the adaptability and real-time performance of the model and making it suitable for dealing with complex and rapidly changing fire environments.

[0040] 2. The present invention can simulate multiple possible paths of fire spread, timely predict the spread trend of fire, help formulate targeted prevention and control strategies, and provide timely decision-making support for fire departments through visualizing dynamic fire spread maps, thereby reducing casualties and property losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of the cell state matrix of the present invention;

[0042] Figure 2 This is a cellular transformation flow chart of the WUI fire spread of the present invention;

[0043] Figure 3 This is a schematic diagram of the WUI fire scenario simulation body of the present invention;

[0044] Figure 4 The WUI fire scenario simulation knowledge graph construction process of the present invention;

[0045] Figure 5 This is a flow chart for monitoring and updating the spatiotemporal elements of WUI fires according to the present invention;

[0046] Figure 6 Schematic diagram of entity nodes of the knowledge graph of the present invention, (a) schematic diagram of element entity nodes, (b) schematic diagram of knowledge meta-nodes;

[0047] Figure 7 Schematic diagram of the construction and calling of the fire spatiotemporal semantic query and simulation framework (FireSTS) of the present invention;

[0048] Figure 8 This is a schematic diagram of the research area in an embodiment of the present invention;

[0049] Figure 9 Schematic diagram of the semantic-driven WUI fire spread simulation operation of the present invention, (a) a schematic diagram of the semantic-based fire simulation scenario setting, (b) a schematic diagram of the spread simulation implemented by calling the FireSTS framework method;

[0050] Figure 10 The fire spread simulation diagrams of the Yaji Mountain Scenic Area on March 30, 2019 at different times are shown in the present invention: (a) fire spread simulation diagram at 13:00, (b) fire spread simulation diagram at 15:00, and (c) fire spread simulation diagram at 17:00.

[0051] Figure 11 Flowchart of the entire present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0053] See also Figures 1 to 7 and Figure 11 The present invention provides an embodiment of a method for simulating fire spread in a forest-urban interface based on spatiotemporal semantics, the method comprising:

[0054] The WUI fire spread model based on cellular automata is constructed. The cellular automata model, namely cellularautomata CA, is used to discretize the fire spread process. Based on the previous research, the knowledge of forest fire and building fire spread simulation is integrated. Different cellular states, state transformation rules and time evolution are carefully divided and summarized. The WUI fire spread simulation model based on cellular automata is proposed.

[0055] The WUI fire scenario simulation knowledge meta-model was constructed. The spatiotemporal elements involved in the WUI fire spread process, their attributes, and mapping relationships were sorted out. The knowledge meta-framework and attribute model were defined from five dimensions: emergencies, hazard factors, disaster-prone environments, disaster-bearing carriers, and emergency management. Based on the WUI fire spread model simulation process, the mapping relationships of spatiotemporal elements involved in WUI fires were sorted out to construct a knowledge meta-relationship model. The knowledge meta-framework, attributes, and relationship models were integrated, and the knowledge related to WUI fire simulation was integrated and expressed using geographic cells as knowledge meta-units, completing the construction of the WUI fire scenario simulation knowledge meta-model.

[0056] The WUI fire scenario simulation ontology was constructed based on the WUI fire scenario simulation knowledge meta-model and the SSN, TIME, and GEOSPARQL ontology standards published by W3C. It integrates WUI fire domain knowledge with spatiotemporal information of sudden fire events. It defines a WUI fire element monitoring sub-ontology for monitoring and updating spatiotemporal data, and a WUI fire knowledge meta-ontology for storing and visualizing fire scenario knowledge. This integration then forms a complete WUI fire scenario simulation knowledge system. Finally, Protégé software was used to complete the ontology modeling of the conceptual model, logical foundation, and specification constraints, realizing the construction of the WUI fire scenario simulation ontology.

[0057] The WUI fire knowledge graph was constructed using the Neo4j graph database for integration and formal display of ontology associations. By processing multi-source fire data, the multi-source information of the experimental area was integrated and dynamically updated. The instance nodes of relevant geographic elements and knowledge elements in the WUI fire scenario were created and updated, and the WUI fire scenario simulation knowledge graph, namely WUIFFSSKG, was constructed.

[0058] FireSTS, a fire spatiotemporal semantic query and simulation framework, uses a relational database to store spatial geometry information and a graph database to store attribute data. It also builds a JDBC driver that integrates access interfaces for both relational and graph databases, along with WUI fire spread simulation and multi-source information integration methods. With a graph-side-led execution model and a collaborative execution model between the relational and graph sides, FireSTS converts hybrid queries that integrate graph semantics and spatiotemporal computing into executable subqueries and simulations for GDBMS and RDBMS through the JDBC driver.

[0059] Semantic-driven simulation uses Cypher language in the WUI fire knowledge graph to complete the setting of fire simulation scenarios such as ignition point, simulation time and simulation conditions, and calls the WUI fire spread simulation method encapsulated in FireSTS to realize semantic-driven WUI fire spread simulation. The simulation results are updated to the corresponding knowledge unit node attributes in the knowledge graph, and output as an EXCEL file with a single cell as a record.

[0060] See also Figure 1 In the WUI fire spread model method based on cellular automata, the state of the cells will change as the fire spreads. The characteristics of the cell state are as follows:

[0061] When the cell state = 0, that is, the cell state S0: the cell is in the fire isolation area, there is no combustible material in the cell, and the cell state remains unchanged;

[0062] When the cell state = 1, that is, the cell state S1: the combustible material in the cell has not been ignited, and the cell is in a combustible but not burned state;

[0063] When the cell state = 2, that is, the cell state S2: the combustible material in the cell has just started to burn and has not been completely burned. The cell is in an internal combustion state and has not ignited the power of other cells.

[0064] When the cell state = 3, that is, the cell state S3: the combustible material in the cell is completely burned, the cell is in a completely burned state, begins to spread outward and has the ability to ignite the remaining cells;

[0065] When the cell state = 4, that is, the cell state S4: the combustible material in the cell has been burned, the cell is in an extinguished state, and has no ability to ignite the remaining cells;

[0066] See also Figure 2 , the transition rule of cell state S0 to S1, the cells covering fire-isolated areas such as water areas and roads are defined as S0, and the cells covering non-fire-isolated areas such as forests and grasslands are defined as S1; the transition rule of cell state S1 to S2, the combustible but unburned S1 state cells are ignited by the surrounding completely burned S3 state cells, and the cells are converted from the combustible but unburned S1 state to the internal combustion state S2; since the WUI area is the boundary between the built-up area and the forest area, the spread mechanism of forest fire and building fire is different. According to the internal spread mechanism of different ground objects, this paper divides the ignition mode into four cases, namely:

[0067] Case 1: Forest cells spread to forest cells. The burning rate R of forest cells is calculated using Wang Zhengfei's forest fire spread model, and the probability of forest cells catching fire at different times is calculated based on the changes in the ignition state of adjacent cells. When the probability of fire Greater than the forest cell fire threshold P Fr That is, it is ignited, and the cells in the S1 state are ignited and converted to S2;

[0068] Case 2: Spread from building cell to building cell; the cell is affected by the adjacent burning house, that is, the completely burned S3 state building cell. It is necessary to first calculate the contribution rate P of a single burning house to the ignition of the cell. mn , and then calculate the probability of fire of the cell due to the cumulative influence of the surrounding burning houses I n , when the probability of cell fire is greater than the building cell fire threshold P Br that is, ignited;

[0069] Case 3: Spread from forest cells to building cells; when a building cell is within the influence range of the forest cell in state S3, the burning probability In of the building cell is calculated. When the burning probability is greater than the threshold P Br that is, ignited;

[0070] Case 4: Spread from building cells to forest cells, when the forest cell is in the influence range P of the building cell in state S3 AmnInside, it can be ignited;

[0071]

[0072] in is the cumulative value of the ignition transfer from the adjacent cells to the current cell after the next time interval Δt; is the spreading rate of cells in state S3 with spreading capability in the Moore neighborhood of the current cell at time t; Δt is the simulation time step and is set to 1 min, L is the horizontal distance between cells (m), and i and j are the coordinate indices of the current cell.

[0073] The transition rule from the cell state S2 to S3 is defined as the time process from the internal combustion S2 state to the complete combustion S3 state as Δt1. When the combustion duration of the cell in the S2 state is t cost When Δt1 is greater than Δt1, the cell state changes from S2 to S3. Because the internal spread mechanisms of building fire and forest fire are different, the internal combustion time of forest cells in this paper is calculated by the ratio of cell burning distance and rate. The development of house fire is affected by many factors. According to the experience of fire experts, this process is simplified to the time t from the beginning of combustion to complete combustion of building cells. 23 ;

[0074]

[0075] where R IN is the internal forest fire spread rate of the cell. The S2 state cell is transformed into the S3 state after Δt1 and has the ability to spread to the neighborhood.

[0076] The transition rule from cell state S3 to S4 defines the time process of the cell from the fully burned S3 state to the extinguished S4 state as Δt2. Δt2 depends on the continuous combustion of the fuel, external fire extinguishing measures and environmental conditions. The duration of flame combustion is related to the time Δt required for complete combustion of the fuel. burn and fire extinguishing intervention time Δt inter Related;

[0077] Δt2=Δt burn +Δt inter ;

[0078]

[0079] Where W is the fuel load within the cell and is expressed in kg / m 2 , m is the mass burning rate of the combustible material, and the unit is kg / (m 2 min); t 34 It is the time from complete combustion to extinguishment of building cells, obtained from the experience of fire experts, and the unit is min.

[0080] The fire prevention and control area is divided into several geographic cells. The geographic cells are used as knowledge units to integrate the relevant characteristics and knowledge of the WUI fire field, forming a multi-dimensional description of the WUI fire field knowledge and the spatiotemporal information of sudden fire events, and realizing the set framework model K m , attribute model K a and relational model K r The WUI fire scenario simulation knowledge element model is constructed. The WUI fire scenario knowledge element framework and attribute model are shown in Table 1 below, and the WUI fire scenario knowledge element relationship model and the inference rule set of the WUI fire spread simulation knowledge element are shown in Table 2 below.

[0081] K=(K m ,K a ,K r );

[0082] K m =(N m ,A m ,R m ),

[0083] K a =(p a ,d a ,f a ),

[0084]

[0085] Where M is the set of all knowledge element frameworks, A is the set of all knowledge element attributes, R is the set of all relationships between attributes, and N is the set of all knowledge element attributes. m Indicates the name and concept of the thing corresponding to the knowledge element, A m is a set of quantitative or qualitative attributes of things, R m is the set of relationships between attributes of things, p a is the attribute value of the thing, d a is the dimension of attribute value, f a is a time-varying function of the attribute and is non-empty only if the attribute is measurable and varies with time, p r is the mapping relationship type, p r Including linear, nonlinear and logical, Enter the attribute state set for the relationship, Output attribute state set for the relation, f r is the specific mapping function of the relationship.

[0086] Table 1

[0087]

[0088]

[0089] Table 2

[0090]

[0091]

[0092]

[0093] See also Figure 3 , a top-down approach is used to define the WUI fire scenario simulation ontology (see Figure 3 ), organized it into two modules: the WUI fire element monitoring module and the WUI fire knowledge element module. By referencing the Semantic Sensor Network Ontology (SSN) and other ontologies published by W3C, such as TIME and GEOSPARQL, and combining them with the relevant requirements of WUI fire element monitoring and management, the SSN ontology was reconstructed to realize the construction of the WUI fire element monitoring ontology, which is used for spatiotemporal data monitoring and updating of sudden fire events. Based on the WUI fire scenario knowledge element model, the geographic cell is used as the knowledge element unit for integration and induction for discrete abstract expression, and the WUI fire knowledge element ontology for the inductive storage and visualization of fire scenario knowledge is constructed.

[0094] See also Figure 4 Protégé software is used to complete the formal expression and ontology knowledge modeling of the conceptual model, logical basis, and specification constraints involved in the WUI fire simulation. The Neo4j graph database is used to integrate and formally display the ontology associations. Through multi-source fire data processing, the instance creation of the relevant fire ontology class is completed, and then the WUI fire scenario simulation knowledge graph is constructed.

[0095] See also Figure 5 、 6 , using remote sensing images, DEM (digital elevation model), meteorological data, mobile signals and other data sources as sensor instances in the ontology, using the executor in the ontology, namely the FireSTS framework, to process different data sources, complete the extraction of surface features and knowledge element monitoring attributes, and then complete the creation and update of surface feature entities and knowledge element entities (nodes) in the knowledge graph, realizing the construction of a knowledge graph that supports real-time analysis and response of complex fire scenes. The dynamic update process of the knowledge graph is as follows: Figure 5 As shown, the surface feature entities (such as Figure 6 -(a)) and knowledge meta-entities (such as Figure 6-(b)), where EntityID is the unique identification field of surface features and knowledge meta-entities, EntityClass is the mapping between graph database nodes and relational database elements, and WKT stores the geometric shapes of surface features.

[0096] See also Figure 7 , storage and mapping of spatiotemporal data in the fire spatiotemporal semantic query and simulation framework, storing the text attributes of spatiotemporal data in GDMS, and the geographic attributes in RDBNS, the entity identifier ID in GDBM is used to associate with the corresponding entity in RDBMS, and the entity class in GDBM is used to associate with the corresponding table name in RDBMS; implementation and calling in the fire spatiotemporal semantic query and simulation framework, encapsulating the corresponding fire simulation method and multi-source data processing method in the Java Maven project, and the interface between the relational database and the graph database, compressing the Java Maven project into a Jar file and loading it into the neo4j graph database, and using the Cypher language in neo4j to complete the call of its encapsulation method.

[0097] Example:

[0098] This paper takes the 2019 Yaji Mountain Scenic Area "3.30" fire as an example, and selects the Yaji Mountain Scenic Area and its surrounding areas as the research area (such as Figure 8 As shown in the figure, by calling the semantic-driven fire spread simulation method proposed in this paper, the simulation of the "3.30" fire spread scenario was realized. The fire spread simulation situation was compared with the relevant reports when the fire occurred. The effectiveness of the method proposed in this study in predicting the spread of WUI fire was verified, providing a scientific basis for fire management and emergency decision-making.

[0099] Based on the FireSTS framework, information extraction and semantic modeling were performed on the multi-source data of Yaji Mountain. Different simulation scenarios were set to simulate the fire conditions under different scenarios through semantic driving. Figure 9 (a) and call the corresponding method in the FireSTS framework to implement fire spread simulation, such as Figure 9 (b).

[0100] This paper starts the simulation from the fire outbreak at 12:23 on March 30, 2019, and simulates the fire spread trend at different times in the study area, such as Figure 10As shown in the figure, the fire spread to Pinggu District at 13:00 and approached the ancient Bixia Yuanjun Temple at around 17:00. This simulation result is consistent with the fire spread trend reported in the news report on the fire: "At 13:06 on March 30, 2019, the fire spread to Pinggu District and approached the Yaji Mountain Scenic Area; at 15:30, the open fire in Miyun District was extinguished, but the fire continued to spread in Pinggu, and in the afternoon, the fire approached the ancient Bixia Yuanjun Temple." It can be concluded that the model can relatively accurately simulate the spread process in the study area, providing critical early warning information for fire emergency decision-making.

[0101] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A fire spread simulation method for the forest-urban interface based on spatiotemporal semantics, characterized by: The method comprises: The WUI fire spread model based on cellular automata is constructed. The cellular automata model, namely cellularautomata CA, is used to discretize the fire spread process. The knowledge of two fire spread simulations, forest fire and building fire, is integrated. Different cellular states, state transformation rules and time evolution are carefully divided and summarized. The WUI fire spread simulation model based on cellular automata is proposed. The WUI fire scenario simulation knowledge meta-model was constructed. The spatiotemporal elements involved in the WUI fire spread process, as well as their attributes and mapping relationships, were sorted out. A knowledge meta-framework and attribute model were defined from five dimensions: emergencies, hazard factors, disaster-prone environments, disaster-bearing carriers, and emergency management. Based on the WUI fire spread model simulation process, the mapping relationships of spatiotemporal elements involved in WUI fires were sorted out to construct a knowledge meta-relationship model. The knowledge meta-framework, attributes, and relationship models were integrated, and WUI fire simulation-related knowledge was integrated and expressed using geographic cells as knowledge meta-units, realizing the construction of the WUI fire scenario simulation knowledge meta-model. The WUI fire scenario simulation ontology was constructed based on the WUI fire scenario simulation knowledge meta-model and the SSN, TIME, and GEOSPARQL ontology standards published by W3C. It integrates WUI fire domain knowledge with spatiotemporal information of sudden fire events. It defines a WUI fire element monitoring sub-ontology for monitoring and updating spatiotemporal data, and a WUI fire knowledge meta-ontology for storing and visualizing fire scenario knowledge. This integration then forms a complete WUI fire scenario simulation knowledge system. Finally, Protégé software was used to complete the ontology modeling of the conceptual model, logical foundation, and specification constraints, realizing the construction of the WUI fire scenario simulation ontology. The WUI fire scenario simulation knowledge graph uses the Neo4j graph database for integration and formal display of ontology associations. By processing multi-source fire data, the integration and dynamic update of multi-source information in the experimental area are achieved, and the instance nodes of relevant geographical elements and knowledge elements in the WUI fire scenario are created and updated, thereby constructing the WUI fire scenario simulation knowledge graph, namely WUIFFSSKG. FireSTS, a fire spatiotemporal semantic query and simulation framework, uses a relational database to store spatial geometry information and a graph database to store attribute data. It also builds a JDBC driver that integrates access interfaces for both relational and graph databases, along with WUI fire spread simulation and multi-source information integration methods. With a graph-side-led execution model and a collaborative execution model between the relational and graph sides, FireSTS converts hybrid queries that integrate graph semantics and spatiotemporal computing into executable subqueries and simulations for GDBMS and RDBMS through the JDBC driver. Semantic-driven simulation uses the Cypher language to set the ignition point, simulation time, and simulation conditions of the fire simulation scenario in the WUI fire knowledge graph, and calls the WUI fire spread simulation method encapsulated in FireSTS to implement semantic-driven WUI fire spread simulation. The simulation results are updated to the corresponding knowledge unit node attributes in the knowledge graph, and are output as an EXCEL file with a single cell as a record.

2. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 1 is characterized by: In the WUI fire spread model method based on cellular automata, the state of the cells changes as the fire spreads. The characteristics of the cell state are as follows: When the cell state = 0, that is, the cell state S0: the cell is in the fire isolation area, there is no combustible material in the cell, and the cell state remains unchanged; When the cell state = 1, that is, the cell state S1: the combustible material in the cell has not been ignited, and the cell is in a combustible but not burned state; When the cell state = 2, that is, the cell state S2: the combustible material in the cell has just started to burn and has not been completely burned. The cell is in an internal combustion state and has not ignited the power of other cells. When the cell state = 3, that is, the cell state S3: the combustible material in the cell is completely burned, the cell is in a completely burned state, begins to spread outward and has the ability to ignite the remaining cells; When the cell state = 4, that is, the cell state S4: the combustible material in the cell has been burned, the cell is in an extinguished state, and does not have the ability to ignite the remaining cells.

3. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 2 is characterized by: The cell state S0 to S1 conversion rule defines the fire-isolated area covering water areas and roads as S0, and the cell covering the non-fire-isolated area of ​​forests and grasslands as S1. The cell state S1 to S2 conversion rule defines that the combustible but unburned S1 state cell is ignited by the surrounding fully burned S3 state cells, and the cell is converted from the combustible but unburned S1 state to the internal combustion state S2. Since the WUI area involves the boundary between the built-up area and the forest area, the spread mechanism of forest fire and building fire is different. According to the internal spread mechanism of different ground fires, the ignition mode is divided into four cases, namely: Case 1: Forest cells spread to forest cells. The burning rate R of forest cells is calculated using Wang Zhengfei's forest fire spread model, and the probability of forest cells catching fire at different times is calculated based on the changes in the ignition state of adjacent cells. , when the probability of ignition Greater than the forest cell fire threshold That is, it is ignited, and the cell in the S1 state is ignited and converted into S2; Case 2: Spread from building cell to building cell; the cell is affected by the adjacent burning house, that is, the completely burned S3 state building cell. It is necessary to first calculate the contribution rate of a single burning house to the ignition of the cell. , and then calculate the probability of fire of the cell due to the cumulative impact of the surrounding burning houses , when the probability of cell fire is greater than the building cell fire threshold that is, ignited; Case 3: Spread from forest cells to building cells; when a building cell is within the influence range of the forest cell in state S3, the burning probability In of the building cell is calculated. When the burning probability is greater than the threshold that is, ignited; Case 4: Spread from building cells to forest cells, when the forest cells are within the influence range of the building cells in state S3 Inside, it can be ignited; ； ； ； in For the next time interval The cumulative value of the ignition transfer from the adjacent cells to the current cell; is the spreading rate of cells in the Moore neighborhood with spreading capability S3 at the current cell at time t; is the simulation time step and is set to 1 min, L is the horizontal distance between cells in meters, and i and j are the current cell coordinate indices.

4. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 3 is characterized by: The cell state S2 to S3 transition rule defines the time process of the cell from the internal combustion S2 state to the complete combustion S3 state as , when the combustion duration of the S2 state cell is Greater than When , the cell state changes from S2 to S3. Because the internal spread mechanism of building fire and forest fire is different, the internal combustion time of forest cells is calculated by the ratio of cell burning distance and rate. The development of house fire is affected by many factors. According to the experience of fire experts, this process is simplified to the time from the beginning of burning to the complete burning of building cells. ; ； in is the internal forest fire spread rate of the cell, the S2 state cell is Then it transforms into S3 state and has the ability to spread to the neighboring areas.

5. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 4 is characterized by: The cell state S3 to S4 transition rule defines the time process of the cell transitioning from the fully burned S3 state to the extinguished S4 state as , The duration of the flame and the time required for the fuel to burn completely depend on the duration of the fuel burning, external fire extinguishing measures and environmental conditions. and firefighting intervention time Related; ； ； Where W is the fuel load within the cell and is expressed in kg / m², m is the mass burning rate of the combustible material and is expressed in kg / (m²・min), It is the time from complete combustion to extinguishment of building cells, obtained from the experience of fire experts, and the unit is min.

6. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 5 is characterized by: The WUI fire scenario simulation knowledge meta-model covers the definitions of the fire knowledge meta-frame model, attribute model and relationship model.

7. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 6 is characterized by: The WUI fire scenario simulation knowledge meta-ontology includes the definition of the WUI fire element monitoring sub-ontology for realizing spatiotemporal data monitoring and updating of sudden fire events; and the definition of the WUI fire knowledge meta-ontology for realizing nano-storage and visualization of fire scenario knowledge.

8. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 7 is characterized by: The storage and mapping of spatiotemporal data in the fire spatiotemporal semantic query and simulation framework stores the text attributes of the spatiotemporal data in the GDMS and the geographic attributes in the RDBMS. The entity identifier ID in the GDBM is used to associate with the corresponding entity in the RDBMS, and the entity class in the GDBM is used to associate with the corresponding table name in the RDBMS.

9. The method for simulating fire spread in forest-urban interface areas based on spatiotemporal semantics according to claim 8, characterized in that: The implementation and call of the fire spatiotemporal semantic query and simulation framework encapsulates the corresponding fire simulation method and multi-source data processing method, as well as the interface between the relational database and the graph database, in a Java Maven project. The Java Maven project is compressed into a Jar file and loaded into the neo4j graph database. The Cypher language in neo4j is used to complete the call of its encapsulated method. The semantic-driven simulation creates and modifies scenario simulation instance nodes in the WUI fire scenario simulation knowledge graph through the Cypher language, completes the setting of the fire simulation scenario of the ignition point and simulation time, and calls the fire spread simulation method encapsulated in the FireSTS framework in neo4j by inputting the node name of the simulation scenario and the end time parameters of the simulation, thereby realizing the semantic-driven WUI fire spread simulation.

Images