Chemical industrial park fire spreading prediction method based on cellular automaton model

Abstract

The invention discloses a chemical industrial park fire spreading prediction method based on a cellular automaton model, and belongs to the technical field of fire prediction. Setting an initial state and a state conversion rule; simulating fire spreading; and refining the fire area. Through the steps of scene information acquisition, initial state and state conversion rule setting, fire spreading simulation, fire area refinement and the like, related personnel can effectively predict the fire spreading trend in advance, the problem of inaccurate prediction caused by uncertain factors is eliminated, and harm possibly brought by a fire is greatly reduced. Meanwhile, the method reduces the increase of the emergency processing cost caused by inaccurate fire prediction, provides a more reliable and accurate guarantee for the safety production of the chemical industry park, and improves the overall safety and economic benefits of the park.

Description

Technical Field

[0001] The present invention relates to the technical field of fire prediction, and particularly to a method for predicting the spread of fire in a chemical industrial park based on a cellular automaton model. Background Art

[0002] At present, the technology for predicting the spread of fire in chemical industrial parks plays a crucial role in the field of fire prevention and control and is widely used in the safety management system of chemical industrial parks. Due to the presence of numerous flammable and explosive hazardous chemicals, complex technological processes, and storage facilities in chemical industrial parks, once a fire occurs, the fire spreads rapidly and the path is complex and changeable, being affected by multiple factors interactively. In traditional methods for predicting the spread of fire, although some key factors have been considered, due to the high complexity and uncertainty of actual fire scenarios, such as sudden changes in wind direction, changes in the combustion characteristics of chemical raw materials, etc., there are often large deviations between the prediction results and the actual fire spread situation. This deviation seriously affects the accuracy and effectiveness of emergency rescue decisions. Currently, the means to improve the accuracy of fire spread prediction mainly include improving the accuracy and coverage of monitoring equipment, optimizing numerical simulation algorithms, and introducing more meteorological data, etc. However, these measures often require high equipment procurement and maintenance costs, complex algorithm debugging, and a large amount of data processing work.

[0003] After retrieval, the existing fire spread models can be mainly divided into three categories. The first type of fire spread model is the physical model, which is based on the law of conservation of energy. After W.R. Fons first created the forest fire spread model in 1946, other models were derived. However, the physical model requires many parameters and has a complex model structure. The second type is the empirical model. The most widely used forest fire spread model is the Australian forest and grassland fire spread model. The calculations involved in the empirical model are simple and easy to perform, but these methods require long-term actual measurement data. Predicting fire spread using the empirical model depends on the measurement accuracy of the images used, so the practicality of the empirical model is poor. The third model type is the semi-empirical fire spread model. This model is derived based on the physical model and the empirical model. Among them, the Rothermel model proposed by Richard C. Rothermel in 1972 is the most widely used model in the United States. However, this model must operate under some simplified requirements, such as a continuous combustion bed, uniform combustibles, etc. Therefore, it is difficult to use the Rothermel model in practical applications. These models usually rely on statistical or phenomenological descriptions of observed fire behavior and require a large amount of data that are difficult to find from historical fires. The computational cost for real-time prediction of large-scale fires is huge. The selection of the fire spread simulation algorithm plays an important role.

[0004] In the field of fire prevention and control in chemical parks, accurately predicting the trend of fire spread has always been a key problem that needs to be solved urgently. Traditional fire spread prediction methods often have many limitations and cannot meet the high-precision prediction requirements in actual complex scenarios. Therefore, a fire spread prediction method for chemical parks based on a cellular automaton model is proposed. Summary of the invention

[0005] The technical problem to be solved by the present invention is how to solve the problems existing in existing prediction methods, such as insufficient consideration of complex environmental factors, lack of adaptability to real-time changing factors, and balancing calculation efficiency and accuracy. A method for predicting the spread of fire in a chemical park based on a cellular automaton model is provided.

[0006] The present invention solves the above technical problems through the following technical solutions, and the present invention comprises the following steps:

[0007] S1: Scene information acquisition

[0008] Determine the geographical layout, building structure and road distribution information of the chemical park and obtain the weather information of the day, then determine the size of the cells in the cellular model and set the prediction time according to user needs;

[0009] S2: Set the initial state and state transition rules

[0010] Determine the nature of the areas within the chemical park, divide them into combustible areas and non-combustible areas, and divide the combustible areas into production areas, storage areas, public engineering areas, management and office areas, and living areas; then set the initial state of the cells in each area, and the initial state of the cells in different areas is different; formulate the cell state conversion rules based on the speed, direction, and potential restrictions of the fire spread;

[0011] S3: Simulating fire spread

[0012] The maze algorithm is used to simulate the spread of fire and calculate the fire spread rate and probability. The cells in the fire area affect the spread rate of the surrounding cells according to the calculated spread rate, and the real-time status of the cells is updated in time.

[0013] S4: Fire area refinement

[0014] Based on step S3, the fire radius in 36 directions from the fire starting point is calculated, and the fire range is refined using the cubic spline interpolation algorithm.

[0015] Furthermore, in step S1, the specific processing process is as follows:

[0016] S11: Divide the scene

[0017] Divide the chemical industrial park scene by grid division method to determine the size of the cell, i.e., the grid.

[0018] S12: Obtain weather information

[0019] Obtain the wind speed and direction, air humidity, and temperature data of the local area on the same day.

[0020] Furthermore, in the step S2, the initial states of the cells are as follows:

[0021] State - 1: Represents a non - combustible cell;

[0022] State 0: Represents a combustible cell that has not been ignited yet;

[0023] State 1: Represents a combustible cell that has just started burning but does not yet have the ability to spread outward;

[0024] State 2: Represents a combustible cell that is burning violently and can spread outward;

[0025] State 3: Represents a combustible cell in the late stage of combustion that does not have the ability to ignite other cells;

[0026] State 4: Represents a cell on the outer wall of a building;

[0027] State 5: Represents an inflammable and explosive cell;

[0028] State 6: Represents a combustible cell that has finished burning.

[0029] Furthermore, in the step S2, the cell state transition rules are formulated as follows:

[0030] Rule 1: IF (i, j, t) state = 0 or 1, Then the states of (i±1, j±1, t + 1) and (i±2, j±2, t + 1) are not affected;

[0031] Rule 2: IF (i, j, t) state = 2, Then the states of (i±1, j±1, t + 1) and (i±2, j±2, t + 1) change with a spread probability p burn change;

[0032] Rule 3: IF (i, j, t) state = 3, Then the states of (i±1, j±1, t + 1) and (i±2, j±2, t + 1) are not affected;

[0033] Rule 4: IF (i, j, t) state = 4, Then the states of (i±1, j±1, t + 1) and (i±2, j±2, t + 1) change with a spread probability p burn change;

[0034] Rule 5: IF the state of (i, j, t) = 5, Then the spread probability p burn increases;

[0035] Rule 6: IF the state of (i, j, t) = 6 or -1, Then the states of (i±1, j±1, t+1) and (i±2, j±2, t+1) are not affected;

[0036] Among them, (i, j, t) represents the state of the cell at time t with coordinates (i, j), and the states of (i±1, j±1, t+1) and (i±2, j±2, t+1) represent the states of the surrounding cells at time t+1 with coordinates (i±1, j±1) and (i±2, j±2).

[0037] Furthermore, in the step S3, the specific processing process is as follows:

[0038] S31: Initialization

[0039] Add the cell corresponding to the fire starting point position to a queue and mark it as state 1;

[0040] S32: Traverse the queue

[0041] Judge whether the queue is empty. If the queue is empty, it means there is no burning area and the spread ends. If not, start traversing the cell nodes in the queue;

[0042] S33: Access the node

[0043] Take out a cell node from the queue and access this cell node. If this cell node has the ability to ignite, calculate the spread rate R and the spread probability p burn , judge the influence on the surrounding cells, update the state, otherwise increase the burning time of this cell node;

[0044] S34: Update

[0045] Update the cell state, add the burning cell node to the queue, and reduce the prediction time by 1 second;

[0046] S35: Repeat the steps

[0047] Repeat steps S32 to S34 until the prediction time is reduced to 0 or the queue is empty.

[0048] Furthermore, in the step S33, the process of calculating the influence on the surrounding cells is as follows:

[0049] S331: Calculate the distance between the current cell and the surrounding cells:

[0050]

[0051] Among them, a is the side length of the cell, i and j are the row number and column number of the current cell, and m and k are the row and column numbers of the affected cells;

[0052] S332: Calculate the spread speed R:

[0053] R 0 = 0.030T + 0.047W + 0.009(100 - h) - 0.304

[0054]

[0055] Among them, R 0 is the initial spread speed, cosθ is the slope correction coefficient, K s is a correction coefficient used to characterize the chemical properties, flammability and combustion configuration pattern of combustibles, K w represents the wind force correction coefficient, represents the slope correction coefficient, v represents the wind speed, is the angle formed when rotating clockwise in the uphill direction to coincide with the wind direction, T is the daily maximum temperature; h is the daily minimum humidity; W is the average wind level at noon;

[0056] S333: Calculate the spread probability p burn :

[0057] p burn = R(1 + p veg )(1 + p den ) / p h

[0058] Among them, p h is affected by the spread blocked by the building, p den , p veg are the fire spread probabilities that depend on the type and size of the combustion area respectively, and R is the spread rate, which is related to weather, terrain and combustibles;

[0059] S334: After calculating the above-mentioned spread rate R and spread probability p burn , judge the influence on the surrounding cells. For each surrounding cell of the current cell, generate a random number r ∈ [0, 1). If r ≤ p burn , then trigger the state conversion operation, and update the corresponding cell state according to the formulated cell state conversion rules.

[0060] Furthermore, in the step S4, centered on the fire starting point position, obtain the coordinates corresponding to the burning cell farthest from the combustion center every 10°, a total of 36 coordinates from 0° to 360°. Based on the points corresponding to the 36 coordinates, use the cubic spline interpolation algorithm to obtain a smooth fire area.

[0061] Furthermore, the cubic spline interpolation algorithm requires at least four sample points to generate a cubic interpolation polynomial. Therefore, the 36 coordinate points (x i , y i ) are divided into 12 groups, with four coordinate points in each group. Each group of data is split, and after grouping the data, interpolation is performed to smooth the curve, thereby obtaining a smooth fire area.

[0062] The present invention has the following advantages compared with the prior art:

[0063] (1) The method for simulating and predicting the spread of fire in a chemical industrial park based on cellular automata of the present invention can accurately obtain the scene information of the chemical industrial park (including geographical layout, building structure, road distribution, meteorological data, etc.), reasonably determine the cell scale and prediction duration, and accurately simulate the spread of fire according to the scientifically set state transition rules during the simulation process. It can improve the accuracy of predicting the spread of fire in the chemical industrial park, enhance the reliability of judging the development trend of the fire, thereby providing a more effective decision-making basis for fire prevention and control in the park, ensuring the safety of the park, and reducing the losses that may be caused by the fire.

[0064] (2) The method for simulating and predicting the spread of fire in a chemical industrial park based on cellular automata of the present invention uses a reasonable algorithm (such as the maze algorithm) to simulate the process of fire spread. On the basis of accurately modeling the chemical industrial park (dividing combustible and non-combustible areas, etc.), it can accurately present the spread path and speed of the fire in the park, so that relevant personnel can make emergency preparations and resource allocation in advance, improving the efficiency of dealing with fires in chemical industrial parks.

[0065] (3) In the process of refining the fire area of the method for simulating and predicting the spread of fire in a chemical industrial park based on cellular automata of the present invention, the cubic spline interpolation algorithm is used to calculate the fire influence radius with high precision. By accurately depicting the fire coverage area, it can provide more accurate reference for the protection of key facilities and personnel evacuation in the park, and at the same time help optimize the layout and use strategy of fire-fighting facilities to ensure that the fire can be controlled to the greatest extent during a fire, guaranteeing the normal production order and the safety of personnel's lives and property in the park. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 is a schematic flow chart of the method for predicting the spread of fire in a chemical industrial park based on the cellular automata model in an embodiment of the present invention;

[0067] Figure 2 is a schematic diagram of the process of smoothing and refining the fire area in an embodiment of the present invention, where (a) is before smoothing and (b) is after smoothing;

[0068] Figure 3 is a local simulation spread example diagram at time t1 in an embodiment of the present invention;

[0069] Figure 4 It is a local simulation spread example diagram at time t2 in the embodiment of the present invention;

[0070] Figure 5 It is a local simulation spread example diagram at time t3 in the embodiment of the present invention. Specific Embodiments

[0071] The following will give a detailed description of the embodiments of the present invention. These embodiments are implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following embodiments.

[0072] As Figure 1 shown, this embodiment provides a technical solution: a method for predicting the spread of fire in a chemical industrial park based on a cellular automaton model, including the following steps:

[0073] S1: Scene Information Acquisition

[0074] Taking a chemical industrial park in a certain area as the simulation site for fire spread, the geographical layout, building structure, and road distribution details of the chemical industrial park are obtained through consulting a large amount of materials and actual investigations; accurately collect the weather information of the day, and the meteorological data comes from the China Meteorological Science Data Sharing Service Network; reasonably determine the scale specification of the cells (lattices) in the cellular model to lay a solid foundation for subsequent simulation operations; set and obtain the corresponding prediction duration according to user needs.

[0075] The specific process of scene information acquisition is as follows:

[0076] S1.1: Scene Division

[0077] The scale of the chemical industrial park is 282.62m × 121.42m, and the side length of a cell is set to 1 / 4m. The entire scene is divided into approximately five hundred and fifty thousand cells. The complex scene of the chemical industrial park is orderly segmented based on the layout of various facilities in the park, topographical features, and simulation accuracy requirements.

[0078] S1.2: Weather Information Acquisition

[0079] Collect detailed local meteorological data of the day, such as the numerical value of wind speed and the specific direction of wind direction, clarify the quantitative index of air humidity, and accurately record the specific temperature of the day. These meteorological elements will directly affect the direction and speed of fire spread. Air humidity affects the combustion efficiency, and temperature is related to the intensity of the combustion reaction. Collecting and integrating them comprehensively and accurately can provide parameter support for fire spread simulation, greatly enhancing the authenticity and credibility of the prediction results.

[0080] S2: Set the Initial State and State Transition Rules

[0081] Precisely define the characteristics of each area within the chemical industrial park, clearly divide it into combustible areas and non-combustible areas, and further subdivide the combustible areas into building areas, flammable and explosive areas; the initial states of cells in different nature areas vary; carefully formulate comprehensive and detailed state transition rules, fully considering the rate, direction of fire spread, and potential limiting conditions.

[0082] Among them, there are eight kinds of initial states set for the cells.

[0083] State -1: This state corresponds to non-combustible cells. It always remains non-ignited throughout the fire spread process, serving as a natural barrier to fire propagation and playing the role of a boundary or isolation area in the model.

[0084] State 0: Represents combustible cells that have not been ignited yet. This is the potential area where the fire may further develop. These cells are in a safe unburned state at the initial moment, but during the fire development process, they may turn into a burning state once certain conditions are met.

[0085] State 1: Indicates combustible cells that have just started burning but do not yet have the ability to spread outward. This state usually appears in a local area at the initial stage of the fire, such as the cells corresponding to the initial ignited substances near the fire source point. At this time, the fire is still in the initial development stage and has not had an obvious heat radiation and ignition effect on the surrounding cells.

[0086] State 2: Used to mark combustible cells that are burning violently and can spread outward. This is an important manifestation of the fire being in the rapid development stage. At this time, the heat generated by the combustion, the height of the flame, etc. are sufficient to have a significant heat transfer and ignition effect on the surrounding combustible cells.

[0087] State 3: Represents combustible cells that have entered the later stage of combustion and have lost the ability to ignite other cells. At this time, although this part of the substance is still burning, due to the consumption of its own combustible components, heat dissipation, etc., it no longer poses a direct threat of fire spread to the surrounding environment.

[0088] State 4: Specifically refers to the cells on the outer wall of the building, which have the characteristic of hindering the spread of combustion to a certain extent. Due to the material and structural characteristics of the building outer wall, it can play a certain role in blocking and delaying the fire during the fire spread process. For example, the cells corresponding to the outer wall of the brick-concrete structure factory building and the firewall in the park belong to this state, and it is one of the key factors affecting the fire spread path and speed in the model.

[0089] State 5: Represents cells with extremely high flammable and explosive hazards. Once the fire spreads to this type of cell, the system will immediately trigger a safety warning mechanism.

[0090] State 6: Combustible cells indicating that the combustion has ended, which marks the complete end of the fire process in the corresponding area. The substances have been fully combusted or successfully extinguished by fire extinguishing measures and no longer pose a fire risk.

[0091] The cell state transition rules are set as follows:

[0092] If at time t, the state of cell (i, j) is 0 or 1 (i.e., combustible cells that have not been ignited), then at time t+1, the cells surrounding cell (i, j) will not be affected by this cell. This means that unignited combustible cells will not cause changes in the states of their surrounding cells at the current time.

[0093] If at time t, the state of cell (i, j) is 2 (i.e., combustible cells that are burning violently and can spread outwards), then at time t+1, there is a certain probability (spread probability p burn ) that the states of the surrounding cells will increase (possibly changing from a lower combustion state to a more intense combustion state or from an unburned state to a burning state). This reflects that violently burning cells will affect their surrounding cells with a certain probability, intensifying their combustion states.

[0094] If at time t, the state of cell (i, j) is 3 (i.e., combustible cells that have reached the late stage of combustion and lost the ability to ignite other cells), then at time t+1, the states of the surrounding cells will not be affected by this cell. This indicates that cells in the late stage of combustion can no longer affect their surrounding cells.

[0095] If at time t, the state of cell (i, j) is 4 (i.e., cells on the building exterior wall), then at time t+1, there is a probability that the states of the surrounding cells will increase. This shows that when cells on the building exterior wall are affected by certain conditions, there is a certain probability that their surrounding cells will change states, and the building exterior wall has a certain degree of property to hinder the spread of combustion.

[0096] If at time t, the state of cell (i, j) is 5 (i.e., cells with extremely high explosion and fire hazard), then the spread probability p burn will increase. This means that once the fire spreads to such cells, the fire will be more likely to spread to the surrounding areas, highlighting the promoting effect of flammable and explosive substances on the spread of the fire.

[0097] If at time t, the state of cell (i, j) is 6 (combustible cells with combustion ended) or -1 (non-combustible cells), then at that time, the states of the surrounding cells will not be affected by the cell.

[0098] The surrounding cells in the above rules are the cells in the two layers surrounding this cell (cell (i, j)).

[0099] S3: Simulate the spread of the fire

[0100] Use the maze algorithm to realistically simulate the spread of fire, accurately calculate the fire spread rate and spread probability. The cells in the fire area affect the spread rates of the surrounding cells according to the calculated spread rate, and the real-time state of the cells is updated in a timely manner.

[0101] Among them, the maze algorithm is specifically as follows:

[0102] S3.1. Initialization

[0103] Put the cell corresponding to the fire starting point into a queue, and at the same time mark this cell as state 1, that is, the state of just starting to burn and not having the ability to spread outward.

[0104] S3.2. Traverse the queue

[0105] Check whether this queue is empty. If the queue is empty, it means that there is no burning area and the fire spread process is over; if the queue is not empty, then start processing the cell nodes in the queue one by one.

[0106] S3.3. Access the node

[0107] Take out a cell node from the queue for inspection. If this cell node has the ability to ignite, calculate the spread rate R and spread probability p burn , judge the influence on the surrounding cells and update the state; if this cell node does not have the ability to ignite, then increase the burning time of this cell node.

[0108] S3.4. Update

[0109] Update the state of the cells, add the burning cell nodes to the queue, and at the same time reduce the prediction time by 1 second.

[0110] S3.5. Repeat the steps

[0111] Repeatedly execute steps S3.2 to S3.4 until the prediction time is reduced to 0 or the queue is empty.

[0112] In step S3.3, the specific process of updating the cell state is as follows:

[0113] S3.3.1: Calculate the distance between the current cell and the surrounding cells:

[0114]

[0115] Among them, a is the side length of the cell, i and j are the row and column numbers of the current cell, and m and k are the row and column numbers of the affected cell;

[0116] S3.3.2: Calculate the spread speed R:

[0117] R 0 = 0.030T + 0.047W + 0.009(100 - h) - 0.304

[0118]

[0119] where, R 0 is the initial spread speed, cosθ is the slope correction coefficient, K s is a correction coefficient used to characterize the chemical properties, flammability and combustion configuration pattern of combustibles, K w represents the wind correction coefficient, represents the slope correction coefficient, v represents the wind speed, is the angle formed when rotating clockwise in the uphill direction and coinciding with the wind direction, T is the daily maximum temperature; h is the daily minimum humidity; W is the average wind force at noon;

[0120] S3.3.3: Calculate the spread probability p burn :

[0121] p burn = R(1 + p veg )(1 + p den ) / p h

[0122] where, p h is affected by the spread blocked by buildings, p den , p veg are the fire spread probabilities that depend on the type and size of the combustion area respectively. R is the spread rate, which is related to weather, terrain and combustibles;

[0123] S3.3.4: After calculating the above-mentioned spread rate R and spread probability p burn , determine the impact on the surrounding cells. For each surrounding cell of the current cell, generate a random number r ∈ [0, 1). If r ≤ p burn , then trigger the state transition operation and update the corresponding cell state according to the formulated cell state transition rules.

[0124] S4: Fire area refinement: Based on the results of step S3, accurately calculate the fire influence radius in 36 directions of the fire starting point, and use the cubic spline interpolation algorithm to perform high-precision refinement processing on the fire coverage area.

[0125] More specifically, a comprehensive analysis of the surrounding environment of the fire origin point is required. The specific operation is to obtain information in the 360° direction around the fire origin point, find the burning point farthest from the ignition point every 10°, and accurately calculate the coordinate positions of these burning points (burning cells). Divide the 36 sample points (x i , y i ) into 12 groups, with four sample points in a group, namely (x 0 , y 0 , x 1 , y 1 , x 2 , y 2 , x 3 , y 3 ), (x 3 , y 3 , x 4 , y 4 , x 5 , y 5 , x 6 , y 6 ). And so on. Since the interpolation algorithm needs to form a single-valued function, each group of data will have the following situations that need to be split: (taking the first group as an example)

[0126] x 0 < x 1 < x 2 , x 2 > x 3 , split the four points into ((x 0 , y 0 ), (x 1 , y 1 ), (x 2 , y 2 ), ((x 2 , y 2 ), (x 3 , y 3 ));

[0127] x 0 < x 1 , x 1 > x 2 > x 3 , split the four points into ((x 0 , y 0 ), (x 1 , y 1 ), ((x 1 , y 1 ), (x 2 , y 2 ), (x 3 , y 3 ));

[0128] x 0 <x 1 ,x 1 >x 2 ,x 2 <x 3 ,split the four points into ((x 0 ,y 0 ),(x 1 ,y 1 )),((x 1 ,y 1 ),(x 2 ,y 2 )),((x 2 ,y 2 ),(x 3 ,y 3 ));

[0129] x 0 >x 1 >x 2 ,x 2 <x 3 ,split the four points into ((x 0 ,y 0 ),(x 1 ,y 1 ),(x 2 ,y 2 )),((x 2 ,y 2 ),(x 3 ,y 3 ));

[0130] x 0 >x 1 ,x 1 <x 2 <x 3 ,split the four points into ((x 0 ,y 0 ),(x 1 ,y 1 )),((x 1 ,y 1 ),(x 2 ,y 2 ),(x 3 ,y 3 ));

[0131] x 0 <x 1 ,x 1 >x 2 ,x 2 <x 3 ,split the four points into ((x 0 ,y0 ),(x 1 ,y 1 ))),((x 1 ,y 1 ),(x 2 ,y 2 ))),((x 2 ,y 2 ),(x 3 ,y 3 ));

[0132] In addition, if the abscissas of two points are the same in a set of data, then the second point needs to be removed.

[0133] Finally, interpolation and smoothing curves are performed on the data after the above grouping and processing. In this way, the fire spread area can be simulated more carefully and accurately, so as to better understand the development trend and scope of the fire, as Figure 2 shown.

[0134] As Figures 3 to 5 shown, it is an example of fire spread prediction by the method of the present invention. Figure 3 It is a local simulation spread example diagram at time t1. Figure 4 It is a local simulation spread example diagram at time t2. Figure 5 It is a local simulation spread example diagram at time t3.

[0135] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting fire spread in a chemical park based on a cellular automaton model, characterized in that: The following steps are involved: S1: Scene information acquisition Determine the geographical layout, building structure and road distribution information of the chemical park and obtain the weather information of the day, then determine the size of the cells in the cellular model and set the prediction time according to user needs; S2: Set the initial state and state transition rules Determine the nature of the areas within the chemical park, divide them into combustible areas and non-combustible areas, and divide the combustible areas into production areas, storage areas, public engineering areas, management and office areas, and living areas; Then, the initial state of the cells in each area is set, and the initial state of the cells in different areas is different; based on the speed, direction and potential restrictions of the fire spread, the cell state transition rules are formulated; S3: Simulating fire spread The maze algorithm is used to simulate the spread of fire and calculate the fire spread rate and probability. The cells in the fire area affect the spread rate of the surrounding cells according to the calculated spread rate, and the real-time status of the cells is updated in time. S4: Fire area refinement Based on step S3, the fire radius in 36 directions from the fire starting point is calculated, and the fire range is refined using the cubic spline interpolation algorithm.

2. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 1, characterized in that: In step S1, the specific processing process is as follows: S11: Divide the scene The chemical park scene is divided into two parts using a grid division method to determine the size of the cell, i.e. the grid; S12: Get weather information Get the local wind speed and direction, air humidity and temperature data for the day.

3. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 1, characterized in that: In step S2, the initial states of the cells are as follows: State-1: represents non-flammable cells; State 0: represents a combustible cell that has not yet been ignited; State 1: represents the combustible cells that have just started to burn but have not yet been able to spread outwards; State 2: represents a combustible cell that burns violently and can spread outward; State 3: represents the combustible cells in the late stage of combustion that do not have the ability to ignite other cells; State 4: cells representing the building's exterior wall; State 5: represents flammable and explosive cells; State 6: Combustible cells representing the end of combustion.

4. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 3, characterized in that: In step S2, the cell state transition rule is formulated as follows: Rule 1: IF (i, j, t) state = 0 or 1, then the states of (i ± 1, j ± 1, t + 1) and (i ± 2, j ± 2, t + 1) are not affected; Rule 2: IF (i, j, t) state = 2, then (i ± 1, j ± 1, t + 1) and (i ± 2, j ± 2, t + 1) states spread with probability p burn change; Rule 3: IF (i, j, t) state = 3, then (i±1, j±1, t+1) and (i±2, j±2, t+1) states are not affected; Rule 4: IF (i, j, t) state = 4, then (i ± 1, j ± 1, t + 1) and (i ± 2, j ± 2, t + 1) states spread with probability p burn change; Rule 5: IF (i, j, t) state = 5, then the spread probability is p burn Increase; Rule 6: IF (i, j, t) state = 6or-1, then (i±1, j±1, t+1) and (i±2, j±2, t+1) states are not affected; Among them, (i,j,t) represents the state of the cell at time t and coordinates (i,j), and the states (i±1,j±1,t+1)and(i±2,j±2,t+1) represent the states of the surrounding cells at time t+1 and coordinates (i±1,j±1) and (i±2,j±2).

5. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 4, characterized in that: In step S3, the specific processing process is as follows: S31: Initialization Add the cell corresponding to the fire starting point to a queue and mark it as state 1; S32: Traversing the queue Determine whether the queue is empty. If the queue is empty, it means there is no burning area and the spread is over. If it is not empty, start traversing the cell nodes in the queue. S33: Access Node Take a cell node from the queue and visit it. If the cell node has the ability to ignite, calculate the spread rate R and spread probability p. burn , determine the impact on the surrounding cells and update the status, otherwise increase the burning time of the cell node; S34: Update Update the cell state, add the burning cell node to the queue, and reduce the predicted time by 1 second; S35: Repeat step Repeat steps S32 to S34 until the predicted time is reduced to 0 or the queue is empty.

6. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 5, characterized in that: In step S33, the process of calculating the impact on surrounding cells is as follows: S331: Calculate the distance between the current cell and the surrounding cells: Among them, a is the side length of the cell, i, j are the number of rows and columns of the current cell, and m, k are the number of rows and columns of the affected cell; S332: Calculate the spreading speed R: R0=0.030T+0.047W+0.009(100-h)-0.304 Where R0 is the initial spreading velocity, cosθ is the slope correction coefficient, K s A correction factor used to characterize the chemical properties, flammability and combustion configuration of combustibles, K w represents the wind correction factor, represents the slope correction coefficient, v represents the wind speed, It is the angle formed when the uphill direction rotates clockwise and coincides with the wind direction. T is the maximum temperature of the day. h is the minimum humidity of the day. W is the average wind level at noon. S333: Calculate the spread probability p burn : p burn =R(1+p veg )(1+p den ) / p h Among them, p h is affected by the building's obstruction of spread, p den 、p veg is the probability of fire spread, which depends on the type and size of the burning area, and R is the spread rate, which is related to weather, terrain, and combustibles; S334: Calculate the above-mentioned spreading rate R and spreading probability p burn After that, determine the impact on the surrounding cells. For each surrounding cell of the current cell, generate a random number r∈[0,1). If r≤p burn , then the state transition operation is triggered, and the corresponding cell state is updated according to the formulated cell state transition rule.

7. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 1, characterized in that: In step S4, with the fire starting point as the center, the corresponding coordinates of the combustion cell farthest from the combustion center are obtained every 10°, with a total of 36 coordinates from 0° to 360°. A cubic spline interpolation algorithm is used based on the points corresponding to the 36 coordinates to obtain a smooth fire area.

8. A method for predicting fire spread in a chemical park based on a cellular automaton model according to claim 7, characterized in that: The cubic spline interpolation algorithm requires at least four sample points to generate a cubic interpolation polynomial; therefore, the 36 coordinate points (x i ,y i ) is divided into 12 groups, with four coordinate points as one group. Each group of data is split, and the data is grouped and then interpolated to smooth the curve to obtain a smooth fire area.

Images