Forest fire vehicle scheduling method for original forest region of northern great khingan

Abstract

The invention discloses a forest fire vehicle scheduling method for an original forest region of northern great khingan. The method comprises the following steps: step 1, collecting forest fire data; 2, determining constraint conditions of the model; 3, establishing a mixed integer linear programming model according to the constraint conditions determined in the step 2 by taking the aim of achieving the shortest rescue time of the fire fighting truck as a target; and 4, designing a genetic algorithm to solve the obtained mixed integer linear programming model to obtain an optimal path of the forest fire rescue vehicle. According to the improved genetic algorithm, a high-quality solution can be quickly obtained for the forest fire emergency rescue vehicle scheduling problem, rescue workers can be helped to make a quick response under the forest fire emergency situation, and the rescue vehicles can be scheduled for the forest fire at the first time according to a feasible scheme obtained through solving; the loss caused by the forest fire is reduced, and the method has important significance for improving the emergency rescue capability under the sudden forest fire condition.

Description

Technical field:

[0001] The present invention relates to the technical field of forest emergency rescue, and in particular to a method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling. Background technology:

[0002] As an important ecological protection area in northern my country, the Greater Khingan Range is also one of the many forest fire risk areas in my country and is deeply threatened by forest fires. The unique geographical location and climatic conditions of the Greater Khingan Range make this area particularly vulnerable to climate change. For example, the alternation of extreme low and high temperatures, coupled with the influence of the monsoon climate, makes vegetation more vulnerable to drought. In addition, the rich forest resources in the Greater Khingan Range have also attracted a large number of human activities, but irregular development and land use have increased the probability of fires.

[0003] Frequent forest fires in the Greater Khingan Range have led to a decrease in species diversity and a simpler community composition. Moreover, the smoke and dust caused by the fires have a negative impact on the air quality in the Greater Khingan Range, posing a long-term threat to the health and livelihoods of local residents. In addition, property losses, ecosystem collapse, and irreversible socio-economic impacts require comprehensive research and in-depth solutions to the forest fire problem in the Greater Khingan Range.

[0004] Due to the variability of fire terrain, the uncertainty of fire sources, and the urgency of time during the rescue process, how to efficiently and quickly dispatch fire trucks has become a complex and urgent research issue. Scientific and reasonable planning of fire truck dispatching can not only shorten the rescue response time and minimize losses, but also improve the overall rescue efficiency and ensure that the fire is controlled in a timely and effective manner.

[0005] At present, after a forest fire breaks out, deploying more firefighting and rescue vehicles can shorten the rescue time, but firefighting and rescue vehicles are not unlimited, especially in emergency situations, the available firefighting vehicles are limited. In the face of forest fires of varying severity, with limited resources, how to reasonably arrange firefighting vehicles to extinguish fires in a short period of time, properly dispatch firefighting vehicles, and reduce resource losses caused by forest fires is the primary goal. Summary of the invention:

[0006] The purpose of the present invention is to provide a method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling Mountains, which can quickly and efficiently carry out rescue work on forest fires.

[0007] The present invention is implemented by the following technical scheme: A method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling, comprising the following steps:

[0008] Step 1: Collect forest fire data;

[0009] Step 2: Determine the constraints of the model;

[0010] Step 3: To minimize the rescue time of the fire truck, a mixed integer linear programming model is established according to the constraints determined in step (2);

[0011] The mixed integer linear programming model is:

[0012]

[0013]

[0014] Where i, j are the fire point numbers; k is the fire truck number; K = {1, 2, ..., k} is the set of fire trucks; |K| is the number of fire trucks; d i,j is the distance between ignition points i and j; v k is the average speed of the fire truck; is the fire extinguishing speed of the fire truck; v i is the fire spreading speed of the fire point; T0 is the departure time of the fire truck; T k is the maximum driving time of the fire truck; M is a positive integer; is a decision variable. If fire truck k goes from node i to node j, then Otherwise, 0; is the time when fire truck k arrives at fire point i; t i The time taken to extinguish the fire; T max Rescue time for fire trucks; is the time when fire truck k arrives at fire point j; t j is the time taken to extinguish the fire point j; v k is the speed of fire truck k; V represents the set of fire base nodes; V' represents the set of fire point nodes;

[0015] Step 4: Design a genetic algorithm to solve the obtained mixed integer linear programming model to obtain the optimal path of forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

[0016] Furthermore, the constraints in step 2 include:

[0017] (1) Firefighting resources are limited, and the number of fire trucks is insufficient to cover all fire points;

[0018] (2) All fire fighting vehicles have the same performance and their fire extinguishing speed is higher than the fire spread speed to ensure that all fire points can be extinguished;

[0019] (3) In the event of a large-scale forest fire, each fire point is independent of each other and will not affect each other;

[0020] (4) The faster the fire spreads and the more serious the fire is, the higher the rescue priority level will be;

[0021] (5) Fire trucks depart from the fire base and return to the base after completing all firefighting tasks;

[0022] (6) The fire spread rate, number of fires at each fire point and the distance between them are known, and each fire point can only be served by one fire truck;

[0023] (7) For each fire truck, the first rescue target point is determined according to the rescue priority. Assuming that the number of fire trucks is K, the first K fire points with the highest priority are evenly allocated to each fire truck as the first rescue target fire point, and the subsequent rescue targets of each fire truck do not need to be sorted according to the rescue priority.

[0024] Furthermore, the designing of a genetic algorithm to solve the obtained mixed integer linear programming model includes:

[0025] (1) Individual coding

[0026] Take a sequence of N-bit integers (N1, N2, ..., N n ) is used to represent the gene encoding, where the N-bit integer covers the information related to the fire point and the fire truck; the integers from small to large represent the rescue priority in descending order; each position in the integer sequence represents the information of a fire point or a fire truck, thus forming a feasible solution; by dividing the number of fire points n by the number of fire trucks k and rounding down, the fire points are fairly allocated; from the perspective of gene encoding, it can be regarded as allocating a uniform number of genes to each rescue priority gene, each rescue priority gene and its allocated gene are regarded as a gene fragment of an individual, and the gene fragments corresponding to all rescue priority genes are composed of a complete individual;

[0027] (2) Initial population: the specific operation steps include:

[0028] S1: Map the gene coding value to the chaotic interval to obtain a chaotic mapping table; the coding value range is [a, b], and the chaotic interval is divided into k equal parts using formula (23) to obtain the chaotic interval mapping relationship of chaotic search;

[0029]

[0030] Where Z k Represents the chaotic interval, divide the chaotic interval into k equal parts, Z k is the interval boundary value; xk,n represents the chaos vector;

[0031] S2: Map the chaotic vector in the individual into a chaotic sequence through the chaotic interval mapping relationship, and then calculate the chaotic sequence through formula (24) to obtain the newly generated chaotic sequence after the current position iteration;

[0032] x k，n+1 = r·x k，n ·(1-x k，n )(k=0,1,2....,m) (24)

[0033] Where x k,n+1 Represents x k,n The value after a chaotic search, r∈[0,4];

[0034] Based on the power function carrier theory, the chaotic sequence x at both ends of the interval is processed by carrier to obtain the chaotic sequence x after carrier v ; As shown in formula (25), where v is a piecewise constant function, obtained based on the power function carrier theory, where the v value is x v The power value in ;

[0035]

[0036] S3: By determining the chaotic interval positions of the iteratively generated k chaotic sequences, the chaotic sequences are inversely mapped into the positive integer form of chaotic vectors;

[0037] S4: Repeat steps S1-S3 for the new individuals generated by the chaotic search until the number of individuals generated by the chaotic search reaches the maximum capacity of the initial population, and the operation of generating the initial population by the chaotic search based on the power function carrier ends;

[0038] (3) Establish elite populations and ordinary populations:

[0039] The initial population uses the roulette selection method to form a common population after a set number of iterations. There are already some excellent individuals in the common population after iterations, and these excellent individuals are called elite individuals. In the initial population iteration process, the top N elite individuals with higher fitness in the population are selected in each selection operation and added to the elite population. The maximum capacity of the elite population is half of the common population. When the population size of the elite population reaches the maximum, individuals are stopped from being added to form an elite population.

[0040] (4) The roulette wheel selection method is used to complete the selection operation of the common population;

[0041] (5) The elite population adopts the elite retention strategy selection method to complete the selection operation;

[0042] (6) After the selection operation, the common population and the elite population simultaneously perform crossover and mutation operations;

[0043] (7) Compare the optimal solution of the elite population after crossover and mutation operations with the optimal solution of the common population after crossover and mutation operations, and retain the better solution;

[0044] (8) Whether the number of iterations reaches the maximum number of iterations set value. If not, continue to execute steps (4)-(7). If the maximum number of iterations set value is reached, the iteration is terminated, and the optimal solution is output to obtain the optimal path for forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

[0045] Furthermore, the roulette wheel selection method: the cumulative probability Q of each individual being selected i The calculation formula is as follows:

[0046]

[0047] Among them, f(X i ) represents individual X i The fitness of P(X i ) represents the probability of each individual being selected; Q i represents the cumulative probability of each individual being selected.

[0048] Furthermore, the elite population adopts the elite retention strategy selection method to complete the selection operation, and the process is as follows:

[0049] Step 1: Evaluate individual fitness. Perform fitness evaluation on all individuals in the elite population to determine the fitness value of each individual.

[0050] Step 2: Select elite individuals, select several individuals with the highest fitness from the elite population as elite individuals;

[0051] Step 3: Keep the elite individuals and copy them directly to the next generation population. This ensures that at least some individuals in the next generation population are excellent, thus preventing the algorithm from falling into a local optimal solution.

[0052] Step 4: Fill the remaining population: In addition to the elite individuals, the remaining population vacancies can be filled through crossover mutation operations.

[0053] Furthermore, the crossover operation specifically includes the following steps:

[0054] STEP 1: Extract the rescue priority genes from the parent individuals and put them into the offspring individuals to obtain the basic structure of the offspring individuals;

[0055] STEP2: Take out the other genes except the rescue priority genes of the two parent individuals respectively, and arrange them closely according to their original order to form two new individuals NF1 and NF2, and then randomly select a gene from the new individuals as the initial crossover position of the circular crossover;

[0056] STEP3: Select a gene from the new individual NF1 or NF2 as the starting point of the cyclic crossover. Here, it is assumed that the selected crossover position is x in NF1. 1,3 , the starting point of the loop intersection is x 1,3 , then x 1,3 The x in the corresponding position in NF2 2,3 After the swap, return to the parent NF1 to find the same 2,3 The positions of the same coding in the gene are exchanged, and the cycle is repeated, and finally the parent NF2 is found to match x 1,3 The same coding position, the cyclic crossover step ends here, and the individuals NC1 and NC2 generated by the crossover of two new parents NF1 and NF2 are obtained;

[0057] STEP4: Distribute the genes in the offspring individuals NC1 and NC2 in the original order, according to the number of genes allocated to the rescue priority genes in the parent generations F1 and F2, and evenly distribute them to the offspring C1 and C2 to form complete offspring C1 and C2 with the rescue priority genes. The crossover operation is now completed and the final offspring C1 and C2 are output.

[0058] Furthermore, the mutation operation is to randomly select one of two mutation operators with equal probability; the two mutation operators are:

[0059] The first mutation operator uses a double-point exchange mutation. The specific operation method is to extract genes other than the rescue priority gene from the mutant individual, arrange them closely in the original order to form a new individual, and randomly select two mutation points x in the new individual. m With x n , then x m With x n The position of is exchanged to complete the mutation;

[0060] The second mutation operator adopts a region-based shift operator. The specific operation is as follows: assuming that the gene segment is {x1, x2, x3, x4}, randomly select a gene x2 in the gene segment, compare its relative distance d(x1, x2), d(x3, x2), d(x4, x2) with other genes in the gene segment, and then move gene x2 to the gene with the smallest relative distance d. The mutation is completed and the mutated part is put back to the original individual to obtain the mutated individual.

[0061] Furthermore, in order to achieve the shortest rescue time for the fire truck, the fitness calculation adopts the inverse of minimizing the rescue time function of the fire truck as the objective function, as shown in the following formula:

[0062]

[0063] Furthermore, the forest fire data includes: the number of fire points, wind speed, temperature at the fire points, wind coefficient, combustible correction coefficient, terrain slope coefficient, the distance between the fire points, the number of fire rescue vehicles, the geographical location of each fire point, the average speed of the fire rescue vehicles, and the extinguishing speed of the fire rescue vehicles.

[0064] Advantages of the present invention:

[0065] (1) The encoding method of the present invention can not only meet the needs of the research problem of the present invention, but also effectively improve the rescue efficiency of fire trucks and reduce the maximum time required for fire trucks to extinguish fires.

[0066] (2) The algorithm of the present invention adopts a power function carrier method to perform chaotic search on the basis of ordinary chaotic search to generate a genetic initial population, which can increase the diversity and ergodicity of the initial population and increase the probability of obtaining a high-quality solution when the initial population is generated.

[0067] (3) The improved genetic algorithm of the present invention can effectively avoid falling into the local optimal state. Compared with the IABC algorithm and the ordinary GA algorithm, it has significant superiority and effectiveness in solving the forest fire emergency rescue vehicle dispatch problem.

[0068] (4) The present invention adopts a dual population strategy of simultaneously performing crossover and mutation operations on the common population and the elite population to enhance the algorithm's optimization capability and, combined with the principle of partial rescue priority, expands the algorithm's solution scope.

[0069] (5) The improved genetic algorithm of the present invention can quickly obtain high-quality solutions to the problem of dispatching emergency rescue vehicles for forest fires, which can help rescue personnel to respond quickly to sudden forest fires and dispatch rescue vehicles for forest fires according to the feasible solutions obtained in the first place, thereby reducing the losses caused by forest fires. This is of great significance for improving emergency rescue capabilities in sudden forest fires. Description of the drawings:

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 Design example diagram for coding.

[0072] Figure 2 An example diagram of the crossover operation for improving the genetic algorithm.

[0073] Figure 3 An example diagram of the mutation operation to improve the genetic algorithm.

[0074] Figure 4 This is an image map of the study area.

[0075] Figure 5 This is the road network map of the study area.

[0076] Figure 6 It is the shortest path diagram between the fire base and the fire point 6 connection points.

[0077] Figure 7 This is the shortest path diagram between fire points 6 and 9.

[0078] Figure 8 This is the cost path diagram for fire points 1, 2, and 3.

[0079] Fig. 9 This is the convergence curve of Example 1.

[0080] Fig.10 This is the convergence curve of Example 4.

[0081] Fig.11 This is the convergence curve of Example 7.

[0082] Fig.12 This is a flow chart of the genetic algorithm of the present invention. Specific implementation method:

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] Example 1: In order to describe the problem more clearly, Table 1 gives the relevant variable parameters and meanings of this example.

[0085] Table 1 Model parameters and meanings

[0086]

[0087]

[0088] The method for dispatching vehicles for forest fire in the original forest area in the northern part of Daxinganling includes the following steps:

[0089] Step 1: Collect forest fire data;

[0090] Forest fire data include: number of fire points, wind speed, temperature at fire points, wind coefficient, combustible correction coefficient, terrain slope coefficient, distance between fire points, number of fire rescue vehicles, geographical location of each fire point, average speed of fire rescue vehicles, and extinguishing speed of fire rescue vehicles.

[0091] Step 2: Determine the constraints of the model;

[0092] The constraints include:

[0093] (1) Firefighting resources are limited, and the number of fire trucks is insufficient to cover all fire points;

[0094] (2) All fire fighting vehicles have the same performance and their fire extinguishing speed is higher than the fire spread speed to ensure that all fire points can be extinguished;

[0095] (3) In the event of a large-scale forest fire, each fire point is independent of each other and will not affect each other;

[0096] (4) The faster the fire spreads and the more serious the fire is, the higher the rescue priority level will be;

[0097] (5) Fire trucks depart from the fire base and return to the base after completing all firefighting tasks;

[0098] (6) The fire spread rate, number of fires at each fire point and the distance between them are known, and each fire point can only be served by one fire truck;

[0099] (7) For each fire truck, the first rescue target point is determined according to the rescue priority. Assuming that the number of fire trucks is K, the first K fire points with the highest priority are evenly allocated to each fire truck as the first rescue target fire point, and the subsequent rescue targets of each fire truck do not need to be sorted according to the rescue priority.

[0100] Step 3: To minimize the rescue time of the fire truck, a mixed integer linear programming model is established according to the constraints determined in step (2);

[0101] The mixed integer linear programming model is:

[0102]

[0103]

[0104] This embodiment uses an undirected graph to represent the forest network, V represents the fire base node set, and V' represents the fire point node set.

[0105] Among them, formula (5) represents the optimization objective, which means minimizing the rescue time of the fire truck; formula (6) means that each fire point can only be served by one fire truck; formulas (7) and (8) ensure that the fire truck starts from the fire rescue base and returns to the original base after completing the fire extinguishing; formula (9) shows that under the flow balance constraint, a fire truck provides service for one fire point and must leave after extinguishing the fire; formula (10) represents the driving time limit of the fire truck; formula (11) means that fire trucks are dispatched to extinguish the first |K| fire points first, and the rescue priority of these fire points is higher. ; Formula (12) represents the time when the fire truck departs from the rescue base; Formula (13) represents the calculation formula for the time taken to extinguish the fire at the fire point; Formulas (14) and (15) are used to calculate the time taken by the fire truck k to reach the fire point i; Formula (16) is the rescue time of the fire truck; Formula (17) describes the time when the fire truck arrives at the fire point i, which is affected by whether the fire truck arrives at the fire point; Formulas (18) and (19) represent effective inequalities, that is, each fire point can only be served by a fire truck once at most; Formulas (20), (21), and (22) are the value ranges of the decision variables.

[0106] The fire truck rescue time consists of two parts: one is the fire extinguishing time of the fire truck, and the other is the driving time of the fire truck. The two parts are superimposed to represent the fire truck rescue and fire extinguishing time. Since there is a single fire point and multiple fire trucks, the rescue time of the fire truck that takes the longest time among the fire trucks is used as the fire truck rescue time of the entire rescue operation.

[0107] Since the forest fire case studied in this embodiment occurred in the Greater Khingan Range area of ​​Heilongjiang Province, Wang Zhengfei's forest fire spread model was used. This model was obtained by Wang Zhengfei and others after years of research on the characteristics of forest fire spread in the Greater Khingan Range area. The model is more practical in the Greater Khingan Range area. The calculation formula of the model is as follows:

[0108] v0=aT+bW+c#(1)

[0109]

[0110] Among them, a, b, c represent regional coefficients. In the Northeast region, a=0.053, b=0.048, c=0.275. W represents the wind force level.

[0111] The calculation formula of fire extinguishing time and the fire spreading speed v at the fire point i i , the time when the fire truck arrives at the fire site Fire extinguishing speed of fire trucks The specific calculation formula is as follows:

[0112]

[0113] By transforming the above formula, the fire extinguishing time calculation formula of the fire point can be obtained as follows:

[0114]

[0115] The fire extinguishing speed of the fire truck must be greater than the fire spread speed at the ignition point to ensure that the fire is extinguished.

[0116] Step 4: Design a genetic algorithm to solve the obtained mixed integer linear programming model to obtain the optimal path of forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

[0117] Design a genetic algorithm to solve the obtained mixed integer linear programming model, and the specific steps include:

[0118] (1) Individual coding

[0119] For the emergency rescue problem of forest fire, the solution needs to be encoded. This encoding is divided into two key parts: the number of fire trucks and the number of fire points. A sequence of N-bit integers (N1, N2, ..., N n ) is used to represent the genetic code, where the N-bit integer covers the information related to the fire point and the fire truck. In this embodiment, it is assumed that there are 9 fire points (N=9) and 3 fire trucks (k=3). For the 9 fire points, the integers from 1 to 9 not only represent the fire point identification, but also reflect the rescue priority of the fire point, and the rescue priority decreases from 1 to 9. The higher the rescue priority, the faster the fire spreads. Each position in the integer sequence represents the information of a fire point or fire truck, thus forming a feasible solution. Its encoding form is as follows Figure 1 shown.

[0120] For the emergency rescue vehicle path planning, this embodiment adopts the principle of partial rescue priority. Figure 1 For example, there are 3 fire trucks and 9 fire points. For the 3 fire trucks, 3 routes need to be planned. Each route follows the principle of partial rescue priority and takes the first 3 fire points with the highest priority as the first rescue target of the 3 fire trucks, that is, the first rescue location after starting from the fire station on each route. The rescue targets of the subsequent fire points on the route do not need to be sorted according to the rescue priority. Figure 1 The three paths represented by the code are the first fire truck: 1-4-5, the second fire truck: 2-6-9, and the third fire truck: 3-8-7. For the gene that is the first rescue target in the code, this embodiment refers to it as a rescue priority gene.

[0121] In this embodiment, a uniform strategy is introduced, that is, the fire points are evenly distributed to the fire rescue vehicles by using the rounding function m=[n / k]. In other words, by dividing the number of fire points n by the number of fire trucks k and rounding down, a fair distribution of the fire points is achieved. From the perspective of gene coding, it can be regarded as allocating a uniform number of genes to each rescue priority gene, and each rescue priority gene and its assigned gene are regarded as a gene fragment of an individual, and the gene fragments corresponding to all rescue priority genes are composed of a complete individual. Through such a coding method, it can not only meet the needs of the research problem of this embodiment, but also effectively improve the rescue efficiency of fire trucks and reduce the rescue time of fire trucks.

[0122] (2) Initial population

[0123] The algorithm of this embodiment uses a power function carrier to perform chaotic search on the basis of ordinary chaotic search to generate the genetic initial population, which can increase the traversal and diversity of the population. Suppose the initially generated individual is x n (n=0), which contains m codes, each code is regarded as a chaotic vector, then the individual x n There are m chaotic vectors in x k,n (k=1, 2, 3, ..., m; n=0). In this embodiment, the code represents the order based on the vehicle's driving path. For example, the code (1, 4, 5, 2, 6, 9, 3, 8, 7) represents three paths, namely 1-4-5, 2-6-9, and 3-8-7. k,n Represents a chaotic vector, such as x 5,1 , represents the chaotic vector form of the fifth code in individual x5.

[0124] The specific steps for generating the initial population include:

[0125] S1: Since the encoding numbers in the individual are in the form of positive integers, and x in the chaotic search k,n The value range of x is (0, 1), and k,n It cannot be equal to 0, 0.25, 0.5, 0.75, or 1. Therefore, the genetic coding values ​​in individuals need to be mapped in a special way before use. k,n The range of values ​​is called the chaotic interval. The first step is to map the gene code value to the chaotic interval to obtain a chaotic mapping table. Take the individual x1 (3, 2, 1, 5, 4) as an example. The number of codes contained in it is 5, and the range of the code value [a, b] is [1, 5].

[0126] The chaotic interval k is divided into equal parts using formula (23), and the chaotic interval mapping relationship of chaotic search is obtained.

[0127]

[0128] Where Z k Represents the chaotic interval, divide the chaotic interval into k equal parts, Z k is the interval boundary value; x k,n represents the chaos vector;

[0129] For example: the interval boundary values ​​of individual x1 are 0, 0.2, 0.4, 0.6, and 0.8 respectively. Combined with the value range of the chaotic interval (0, 1), the final mapping relationship is: 1→[0,0.2], 2→[0.2,0.4], 3→[0.4,0.6], 4→[0.6,0.8], 5→[0.8,1].

[0130] S2: Map the chaotic vector in the individual into a chaotic sequence through the chaotic interval mapping relationship, and then calculate the chaotic sequence through formula (24) to obtain the newly generated chaotic sequence after the current position iteration;

[0131] x k，n+1 = r·x k，n ·(1-x k，n )(k=0,1,2...,m) (24)

[0132] Where x k,n+1 Represents x k,n The value after a chaotic search, r∈[0,4];

[0133] Based on the power function carrier theory, the chaotic sequence x at both ends of the interval is processed by carrier to obtain the chaotic sequence x after carrier v ; As shown in formula (25), where v is a piecewise constant function, obtained based on the power function carrier theory, where the v value is x v The power value in ;

[0134]

[0135] S3: By determining the chaotic interval positions of the iteratively generated k chaotic sequences, the chaotic sequences are inversely mapped into the positive integer form of chaotic vectors;

[0136] S4: Repeat steps S1-S3 for the new individuals generated by the chaotic search until the number of individuals generated by the chaotic search reaches the maximum capacity of the initial population, and the operation of generating the initial population by the chaotic search based on the power function carrier ends;

[0137] (3) Establish elite populations and ordinary populations:

[0138] The initial population generated in the previous step uses the roulette wheel selection method to form a common population after iterating and updating a set number of times (5 times in this embodiment);

[0139] There are already some excellent individuals in the ordinary population after iterative updates, and these excellent individuals are called elite individuals. In the initial population iteration process, the top N elite individuals with higher fitness in the population are selected in each selection operation and added to the elite population. The maximum capacity of the elite population is half of the ordinary population. When the population size of the elite population reaches the maximum, stop adding individuals to form an elite population.

[0140] (4) The roulette wheel selection method is used to complete the selection operation of the common population;

[0141] (5) The elite population adopts the elite retention strategy selection method to complete the selection operation;

[0142] (6) After the selection operation, the common population and the elite population simultaneously perform crossover and mutation operations;

[0143] (7) Compare the optimal solution of the elite population after crossover and mutation operations with the optimal solution of the common population after crossover and mutation operations, and retain the better solution;

[0144] (8) Whether the number of iterations reaches the maximum number of iterations set value. If not, continue to execute steps (4)-(7). If the maximum number of iterations set value is reached, the iteration is terminated, and the optimal solution is output to obtain the optimal path for forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

[0145] Among them, roulette selection method: the cumulative probability Q of each individual being selected i The calculation formula is as follows:

[0146]

[0147] Among them, f(X i ) represents individual X i The fitness of P(X i ) represents the probability of each individual being selected; Q i represents the cumulative probability of each individual being selected.

[0148] The elite population adopts the elite retention strategy selection method to complete the selection operation, and the process is as follows:

[0149] Step 1: Evaluate individual fitness. Perform fitness evaluation on all individuals in the elite population to determine the fitness value of each individual.

[0150] Step 2: Select elite individuals, select several individuals with the highest fitness from the elite population as elite individuals;

[0151] Step 3: Keep the elite individuals and copy them directly to the next generation population. This ensures that at least some individuals in the next generation population are excellent, thus preventing the algorithm from falling into a local optimal solution.

[0152] Step 4: Fill the remaining population: In addition to the elite individuals, the remaining population vacancies can be filled through crossover and mutation operations.

[0153] In this embodiment, according to the particularity of the problem of the present invention and the design of the coding format, each code in the individual gene is unique and the order of the codes must be based on the partial rescue priority principle of fire rescue. If the traditional crossover operation method is adopted, such as single-point crossover, multi-point crossover, etc., it will lead to code duplication and changes in the rescue priority of the fire point, resulting in the generation of some repeated and useless offspring individuals. The generated individual gene codes have the same genes, and the individuals composed of such gene codes are an infeasible solution. The gene codes in the offspring individuals do not follow the partial rescue priority principle, which not only cannot solve the actual needs of the problem in this article, but also degrade the feasible solution. In order to solve the above problems, this embodiment adopts special processing for the crossover operation to meet the needs of the research problem of the present invention and improve the search efficiency of the algorithm.

[0154] The specific steps of the crossover operation in this embodiment include:

[0155] The two parent generations are F1{x 1,1 ,x 1,2 ,x 1,3 ,...,x 1,9} and F2{x 2,1 ,x 2,2 ,x 2,3 ,...,x 2,9}, assuming that the number of gene codes in the parent individual is 9, and the rescue priority gene is x i,1 ,x i,4 ,x i,7 , representing the codes of the fire points with the fastest fire spread speed in the individuals. Then, follow the steps below to generate the offspring individuals C1 and C2:

[0156] STEP1: Extract the rescue gene x from the parent individual i,1 ,x i,4 ,x i,7 , put it into the offspring individual, and get the basic structure of the offspring individual.

[0157] STEP2: Take out the other genes except the rescue priority genes of the two parent individuals respectively, and arrange them closely according to their original order to form two new individuals NF1{x 2,2 ,x 2,3 ,x 2,5 ,x2,5 ,x 2,8 ,x 2,9}, NF2{x 1,2 ,x 1,3 ,x 1,5 ,x 1,6 ,x 1,8 ,x 1,9}, then randomly select a gene x from the new individual 1,j or x 2,j As the initial crossing position of the cyclic crossing.

[0158] STEP3: Choose x 1,j ,x 2,j One of them is used as the starting point of the cyclic crossover. Here, it is assumed that the selected crossover position is x 1,3 , the starting point of the loop intersection is x 1,3 , then x 1,3 With x 2,3 After the swap, return to the parent NF1 to find the same 2,3 The positions of the same coding in the gene are exchanged, and the cycle is repeated, and finally the parent NF2 is found to match x 1,3 The same coding position, the cyclic crossover step is completed, and the individuals NC1 and NC2 generated by the crossover of two new parents NF1 and NF2 are obtained.

[0159] STEP4: Evenly distribute the genes in the offspring individuals NC1 and NC2 to the offspring C1 and C2 in the original order according to the number of genes allocated to the rescue priority genes in the parent generations F1 and F2, and form complete offspring C1 and C2 with the rescue priority genes. The crossover operation is now complete. Output the final offspring C1 and C2.

[0160] like Figure 2 The figure shows the process demonstration of the crossover operation, where it is assumed that coding 1, 2, and 3 are rescue priority genes.

[0161] The traditional mutation operation modes in the genetic algorithm of this embodiment include random single-point mutation and multi-point mutation. As the length of the chromosome increases, the range of randomly selected mutation positions in the mutation operation also expands, which may lead to more invalid searches, thereby slowing down the iteration speed of the algorithm. In addition, the expansion of random mutation positions may also make it more difficult for the algorithm to achieve local optimization, affecting the performance of the algorithm. Since there are special rescue priority genes in the research problem of the present invention, the rescue priority genes cannot be changed during the entire genetic algorithm process. Based on the above problems and the characteristics of the research problem of the present invention, this embodiment adopts two mutation operators in the mutation operation, and one of the two mutation operators is randomly selected for use with equal probability. The specific operation steps of the two mutation operators are as follows: Figure 3 shown.

[0162] The first mutation operator uses a double-point exchange mutation. The specific operation method is to extract genes other than the rescue priority gene from the mutant individual, arrange them closely in the original order to form a new individual, and randomly select two mutation points x in the new individual. m With x n , then x m With x n The positions of the two are exchanged to complete the mutation.

[0163] The second mutation operator adopts a region-based shift operator. The specific operation is as follows: assuming that the gene segment is {x1, x2, x3, x4}, randomly select a gene x2 in the gene segment, compare its relative distance d(x1, x2), d(x3, x2), d(x4, x2) with other genes in the gene segment, and then move gene x1 to the gene with the smallest relative distance d. The mutation is completed and the mutated part is put back to the original individual to obtain the mutated individual.

[0164] In this embodiment, the goal is to achieve the shortest rescue time for fire trucks. The calculation of fitness uses the inverse of the fire truck rescue time function as the objective function, as shown in the following formula:

[0165]

[0166] Example 2: Experimental simulation and result analysis

[0167] This embodiment uses a real fire case that occurred in July 2020 in the jurisdiction of the Wuma Forestry Bureau and the Yong'anshan Forestry Bureau in the Greater Khingan Range for experimental simulation. Due to the lack of relevant examples of suitable forest fire emergency rescue vehicle dispatching problems, large-scale random generation of fire points is also used as a simulation example for experimental simulation. The relevant data of the real fire case in this embodiment was obtained by the local management and protection bureau. The large-scale simulation example randomly generates 20, 30, and 40 fire points under the same example settings, and uses the IABC (Improved Artificial Bee Colony) improved bee colony algorithm, GA (Genetic Algorithm) genetic algorithm, and IGA (Improved Genetic Algorithm) improved genetic algorithm, that is, the genetic algorithm of the present invention, to solve the problem. The specific parameter settings are described in detail in the subsequent content.

[0168] The algorithms used and designed in this embodiment are all implemented in Python language. The operating system used in the experiment is Windows 10, the CPU is 2.40 GHz, and the memory is 16 GB.

[0169] 1. Simulation Data Processing

[0170] The case studied in this embodiment is a number of fires that occurred in July 2020 in the jurisdiction of the Wuma Forestry Bureau and the Yong'anshan Forestry Bureau in the Greater Khingan Range. Since the fires occurred on the same day and the time interval between the fires was short, they can be regarded as a concurrent forest fire. This forest fire is a typical lightning forest fire. It is a fire caused by natural lightning strikes in summer. This fire involved 9 fire points. Let N = 9. The forest farms involved in these 9 fire points are distributed in Anya Forest Farm, Jixinggou Forest Farm, Maohe Forest Farm 79 Forest Class, Maohe Forest Farm 96 Forest Class, Wuyuan Forest Farm, Yakruchi Forest Farm, Yimuhe Forest Farm 34 Forest Class, Yimuhe Forest Farm 42 Forest Class, and Wuma Forest Farm. Figure 4 This is an image of the distribution of fire points. The image comes from 91 Satellite Assistant (https: / / www.91weitu.com / ).

[0171] The road network density in the northern original forest area is low, and the fire point is far away from the road. The fire truck cannot directly reach the fire point through the existing road. For this reason, this embodiment selects a location on the road around the forest area as the connection point between the road and the forest area, such as Figure 5 As shown in the figure, the green marked points are connection points. The entire path from the fire base to the fire point is divided into two parts by the connection point. The first part takes the fire base as the starting point and the connection point between the highway and the forest area as the end point. The second part takes the connection point as the starting point and the fire point as the end point. The two parts are combined to form the complete path from the fire base to the fire point. The planning of these two parts of the path is obtained by obtaining the road network data and DEM data of the study area, extracting the terrain slope and undulation information of the study area in the DEM data, and using ArcGIS to perform shortest path analysis and cost path analysis. The shortest path analysis can plan the shortest path from the fire base to the connection point, and the cost path analysis can plan the path with the least obstruction from the connection point to the fire point.

[0172] The road network data of this embodiment comes from Openstreetmap (http: / / www.openstreetmap.org / ).

[0173] The DEM data comes from the Geospatial Data Cloud Platform of Computer Network Information Center of the Chinese Academy of Sciences (http: / / www.gscloud.cn).

[0174] 2. Shortest Path Analysis

[0175] The shortest path analysis is to solve the shortest path between the fire base and the connection point, and between different fire points and the connection point through the Network Analyst tool in ArcGIS. Taking fire point 6 and fire point 9 as an example, Figure 6 and Figure 7 The path plan after the shortest path analysis is shown. Figure 6 It is the shortest path from the fire base to the fire point 6 connection point. Figure 7 It is the shortest path from the connection point of ignition point 6 to the connection point of ignition point 9.

[0176] 3. Cost Path Analysis

[0177] ArcGIS was used to perform a cost path analysis between the fire point and the connection point. The weight factors used in this example were slope and undulation. The calculation formula after evenly distributing the weights was as follows: "slope"*0.5+"undulation"*0.5. The final path was generated using the "cost path" of "distance analysis" in the Spatial Analyst tool using the end point data, cost distance data, and backlink data. The generated path is shown in the figure below. Figure 8 shown.

[0178] 4. Fire rescue priority division

[0179] The fire spread speed at the ignition point calculated according to Wang Zhengfei's forest fire spread model is shown in Table 2.

[0180] Table 2 Spread speed of each fire point (unit: m / min)

[0181]

[0182] Then, they are sorted according to the fire spread speed. After sorting, fire point 1 is the fire point with the highest rescue priority, and the rescue priorities of the other fire points decrease step by step. The sorting is shown in Table 3.

[0183] Table 3 Spread speed between fire points according to rescue priority (unit: m / min)

[0184]

[0185] Table 4 Distances between fire points according to rescue priority (unit: kilometers)

[0186]

[0187] 5. Simulation Results and Analysis

[0188] In the case of this embodiment, different scenarios are set by changing the fire extinguishing speed and the driving speed of the rescue vehicle. The number of rescue vehicles in all scenarios is 3, and the scenarios are set as follows:

[0189] Scenario 1: The fire extinguishing speed is 8 min / m and the rescue vehicle travels at a speed of 54 km / h.

[0190] Scenario 2: The fire extinguishing speed is 8 min / m and the rescue vehicle travels at a speed of 74 km / h.

[0191] Scenario 3: The fire extinguishing speed is 10 min / m and the rescue vehicle travels at a speed of 54 km / h.

[0192] Scenario 4: The fire extinguishing speed is 10 min / m and the rescue vehicle travels at a speed of 54 km / h.

[0193] Scenario 5: The fire extinguishing speed is 12 min / m and the rescue vehicle travels at a speed of 54 km / h.

[0194] Scenario 6: The fire extinguishing speed is 12 min / m and the rescue vehicle travels at a speed of 54 km / h.

[0195] Based on the small scale of simulation cases, setting the number of iterations too large cannot well reflect the effectiveness of the algorithm. Therefore, the number of iterations of the algorithm used in all scenarios is set to 10. The purpose is to compare the advantages and disadvantages of the solutions obtained in the shortest time in small-scale cases and better reflect the performance of the algorithm.

[0196] After solving the six scenarios set in the experimental simulation using IABC, GA, and IGA (the improved genetic algorithm of the present invention) algorithms respectively, the calculation results are shown in the following table, where Best is the optimal value of the objective function obtained after the algorithm runs 100 times, Avg is the average value of the objective function, D is the driving distance of the entire rescue journey, and CPU is the average running time of the program.

[0197] Table 5 Calculation results of IABC algorithm under six scenarios

[0198]

[0199]

[0200] Table 6 Calculation results of GA algorithm under six scenarios

[0201]

[0202] Table 7 Calculation results of IGA algorithm under 6 scenarios

[0203]

[0204]

[0205] Table 8 Fire truck dispatching schemes under six scenarios of IABC algorithm

[0206]

[0207] Table 9 Fire truck dispatching schemes under 6 scenarios of GA algorithm

[0208]

[0209] Table 1 Fire truck dispatching schemes under six scenarios of IGA algorithm

[0210]

[0211]

[0212] In Tables 8, 9, and 10, A1, A2, and A3 are fire truck dispatch solutions obtained by the IABC algorithm, B1, B2, and B3 are solutions obtained by the GA algorithm, and C1, C2, and C3 are solutions obtained by the IGA algorithm. It can be seen from the table that in the six scenarios set, the optimal value and average value calculated by the IGA algorithm are improved compared with the ordinary GA algorithm, and the Best value is reduced by about 4%-8%. The experimental results show that the IGA algorithm can obtain a better solution than the ordinary GA algorithm, reducing the rescue time of the entire emergency rescue process. Although the Best value is slightly improved compared with the IABC algorithm, the overall difference is very small. This is due to the small scale of the case problem. The principle of partial rescue priority is adopted, and the rescue order of other genes except the rescue priority gene is not restricted, which expands the scope of the solution, so that the IGA algorithm can avoid falling into the local optimal state. The average program running time of the IGA algorithm is also reduced, which shows that the IGA algorithm can quickly obtain high-quality solutions.

[0213] VI. Large-Scale Simulation Results and Analysis

[0214] Since the scale of the fire in the above real case is relatively small, the requirements for the algorithm's solving performance are not high, and the algorithm performance cannot be well compared. In order to better verify the solving performance of the algorithm of the present invention, simulation examples with 20, 30, and 40 ignition points are randomly generated. Under the same parameter settings, IABC, GA, and IGA (the improved genetic algorithm of the present invention) algorithms are used for solving them respectively. The generation method is as follows: the fire spread speed is randomly generated within [1,7] m / min, and the distance between each ignition point is randomly generated within [50, 100] km.

[0215] The detailed scenario settings and solution results of the large-scale experimental simulation are shown in the following table. Compared with the previous experimental simulation, due to the large number of ignition points, the number of vehicles K is added as a new parameter variable.

[0216] Table 2 Large-scale experimental simulation results of three algorithms

[0217]

[0218] In the above table, K is the number of fire trucks, Best is the optimal objective function value obtained in 50 runs of the algorithm, and Avg is the average objective function value obtained in 50 runs of the algorithm. From the calculation results in the table, it can be found that the Avg value and Best value calculated by the IGA algorithm in this embodiment are better than the IABC algorithm and the GA algorithm, and the Avg value and Best value calculated by the IABC algorithm are not much different. The Best value obtained by the IGA algorithm is compared with the Best value obtained by the IABC algorithm. The maximum reduction in the total rescue time is no more than 5%, and the average is about 3.5%. Compared with the GA algorithm, the Best value is reduced by a maximum of 10.9%, and the Avg value is reduced by 8.8%. In all large-scale example settings, the IGA algorithm has improved the results obtained by the algorithm compared with the IABC algorithm and the ordinary GA algorithm, which shows that the IGA algorithm of the present invention performs well in the calculation results. In terms of the running time of the program, the IGA algorithm performs better, which shows that the IGA algorithm can obtain a scheduling plan more quickly in the emergency of forest fires and reduce the losses caused by forest fires.

[0219] For three large-scale examples with a scenario setting of 54km / h vehicle speed and 12m / min fire extinguishing speed, the convergence curves of each example solved by GA, IABC, and IGA algorithms are compared. Figure 9-11 shown.

[0220] From the above convergence curves, it can be seen that the IGA algorithm can obtain a better solution than the other two algorithms at the beginning of the iteration process. It proves that the generation of the initial solution through chaotic search based on the power function carrier can increase the diversity and ergodicity of the initial population, and increase the probability of obtaining a high-quality solution when the initial population is generated. At the beginning of the iteration process, the three algorithms can converge at a faster speed and reach a relatively stable solution after a certain number of iterations. From the convergence graph, it can be seen that the ordinary GA algorithm is easy to fall into the local optimum and it is difficult to search for a better solution. The IGA algorithm can effectively avoid falling into the local optimum after being improved on the basis of the ordinary GA algorithm. Comparing the three algorithms, the IGA algorithm has obtained a better solution than the other two algorithms in the final results obtained by convergence in the three groups of experiments. By further comparing the convergence curves of the three algorithms, the superiority and effectiveness of the IGA algorithm in solving the forest fire emergency rescue vehicle dispatch problem are verified.

[0221] Aiming at the particularities of road conditions in the northern virgin forest area and the vehicle dispatching problem of forest fires, the present invention establishes a vehicle dispatching model with the goal of minimizing the rescue time. Taking into account the actual road conditions in the northern virgin forest area of ​​Daxinganling, the road network density is low, and the fire point is far away from the road, and the fire truck cannot directly reach the fire point for rescue through the road in the road network. Based on this problem, the present invention downloads the DEM data of the study area, extracts the slope and undulation information as weight factors, and plans the least obstructed path before solving the problem, and then combines this path with the roads in the road network and introduces it into the problem model for solving it. Specifically, in the model, a parameter required is the distance length of the path. The distance length of this path is between two fire points, or between the fire base and a certain fire point. How to determine this path requires certain planning. The method adopted in this embodiment is to divide a path into two parts. Take the fire base to a certain fire point as an example. To reach the fire point, the fire base must first drive to the vicinity of the fire point through the roads existing in the road network. The road position near the fire point can be selected in advance as a connection point, and the shortest path from the fire base to this connection point is used as the shortest path. Because the fire point cannot be directly reached through the existing roads in the road network, when the rescue vehicle arrives near the fire point, a path with less obstruction can be planned. By using regional elevation data and forest terrain for cost path analysis, ArcGIS tools can be used to plan a path with less undulation and slope to reach the fire point for fire extinguishing. This part is called the cost path. The combination of the cost path and the shortest path is a complete path for the fire truck to reach the fire point from the rescue base. Then the sum of the distance lengths of the two paths is calculated to obtain the length of the combined path, which is introduced into the model for solution. This narrows the difference between the forest fire rescue vehicle dispatch problem and the traditional vehicle dispatch problem, ensures the authenticity and reliability of the research results, provides a new idea for the forest fire emergency rescue vehicle dispatch problem, and makes the research more realistic.

[0222] The present invention introduces a chaotic search method based on a power function carrier to generate an initial population, designs an IGA algorithm, and conducts experimental simulations on real fire cases and randomly generated large-scale examples, compares the solution results of different algorithms, and verifies the feasibility and effectiveness of the IGA algorithm in this paper. The IGA algorithm can quickly obtain high-quality solutions to the forest fire emergency rescue vehicle dispatch problem, which can help rescue personnel to respond quickly to forest fire emergencies, dispatch rescue vehicles for forest fires according to the feasible solutions obtained in the first place, reduce the losses caused by forest fires, and is of great significance for improving emergency rescue capabilities in the event of sudden forest fires.

[0223] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling, characterized in that: The following steps are involved: Step 1: Collect forest fire data; Step 2: Determine the constraints of the model; Step 3: To minimize the rescue time of the fire truck, a mixed integer linear programming model is established according to the constraints determined in step (2); The mixed integer linear programming model is: Where i is the fire point number; k is the fire truck number; K = {1, 2, ..., k} is the set of fire trucks; |K| is the number of fire trucks; d i,j is the distance between ignition points i and j; v k is the average speed of the fire truck; is the fire extinguishing speed of the fire truck; v i is the fire spreading speed of the fire point; T0 is the departure time of the fire truck; T k is the maximum driving time of the fire truck; M is a positive integer; is a decision variable. If fire truck k goes from node i to node j, then Otherwise, 0; is the time when fire truck k arrives at fire point i; t i The time taken to extinguish the fire; T max Rescue time for fire trucks; is the time when fire truck k arrives at fire point j; t j is the time taken to extinguish the fire point j; v k is the speed of fire truck k; V represents the set of fire base nodes; V' represents the set of fire point nodes; Step 4: Design a genetic algorithm to solve the obtained mixed integer linear programming model to obtain the optimal path of forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

2. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 1 is characterized in that: The constraints in step 2 include: (1) Firefighting resources are limited, and the number of fire trucks is insufficient to cover all fire points; (2) All fire fighting vehicles have the same performance and their fire extinguishing speed is higher than the fire spread speed to ensure that all fire points can be extinguished; (3) In the event of a large-scale forest fire, each fire point is independent of each other and will not affect each other; (4) The faster the fire spreads and the more serious the fire is, the higher the rescue priority level will be; (5) Fire trucks depart from the fire base and return to the base after completing all firefighting tasks; (6) The fire spread rate, number of fires at each fire point and the distance between them are known, and each fire point can only be served by one fire truck; (7) For each fire truck, the first rescue target point is determined according to the rescue priority. Assuming that the number of fire trucks is K, the first K fire points with the highest priority are evenly allocated to each fire truck as the first rescue target fire point, and the subsequent rescue targets of each fire truck do not need to be sorted according to the rescue priority.

3. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 1 is characterized in that: The designing of a genetic algorithm to solve the obtained mixed integer linear programming model comprises: (1) Individual coding Take a sequence of N-bit integers (N1, N2, ..., N n ) is used to represent the gene encoding, where the N-bit integer covers the information related to the fire point and the fire truck; the integers from small to large represent the rescue priority in descending order; each position in the integer sequence represents the information of a fire point or a fire truck, thus forming a feasible solution; by dividing the number of fire points n by the number of fire trucks k and rounding down, the fire points are fairly allocated; from the perspective of gene encoding, it can be regarded as allocating a uniform number of genes to each rescue priority gene, each rescue priority gene and its allocated gene are regarded as a gene fragment of an individual, and the gene fragments corresponding to all rescue priority genes are composed of a complete individual; (2) Initial population: the specific operation steps include: S1: Map the gene coding value to the chaotic interval to obtain a chaotic mapping table; the coding value range is [a, b], and the chaotic interval is divided into k equal parts using formula (23) to obtain the chaotic interval mapping relationship of chaotic search; Where Z k Represents the chaotic interval, divide the chaotic interval into k equal parts, Z k is the interval boundary value; x k,n represents the chaos vector; S2: Map the chaotic vector in the individual into a chaotic sequence through the chaotic interval mapping relationship, and then calculate the chaotic sequence through formula (24) to obtain the newly generated chaotic sequence after the current position iteration; x k，n+1 =r·x k，n ·(1-x k，n ) (k=0,1,2....,m) (24) Where x k,n+1 Represents x k,n The value after a chaotic search, r∈[0,4]; Based on the power function carrier theory, the chaotic sequence x at both ends of the interval is processed by carrier to obtain the chaotic sequence x after carrier v ; As shown in formula (25), where v is a piecewise constant function, obtained based on the power function carrier theory, where the v value is x v The power value in ; S3: By determining the chaotic interval positions of the iteratively generated k chaotic sequences, the chaotic sequences are inversely mapped into the positive integer form of chaotic vectors; S4: Repeat steps S1-S3 for the new individuals generated by the chaotic search until the number of individuals generated by the chaotic search reaches the maximum capacity of the initial population, and the operation of generating the initial population by the chaotic search based on the power function carrier ends; (3) Establish elite populations and ordinary populations: The initial population uses the roulette wheel selection method to form a common population after a set number of iterations. There are already some excellent individuals in the common population after iterations, and these excellent individuals are called elite individuals. In the initial population iteration process, the top N elite individuals with higher fitness in the population are selected in each selection operation and added to the elite population. The maximum capacity of the elite population is half of the common population. When the population size of the elite population reaches the maximum, individuals are stopped from being added to form an elite population. (4) The roulette wheel selection method is used to complete the selection operation of the common population; (5) The elite population adopts the elite retention strategy selection method to complete the selection operation; (6) After the selection operation, the common population and the elite population simultaneously perform crossover and mutation operations; (7) Compare the optimal solution of the elite population after crossover and mutation operations with the optimal solution of the common population after crossover and mutation operations, and retain the better solution; (8) Whether the number of iterations reaches the maximum number of iterations set value. If not, continue to execute steps (4)-(7). If the maximum number of iterations set value is reached, the iteration is terminated, and the optimal solution is output to obtain the optimal path for forest fire rescue vehicles, that is, the number and order of the fire points that each fire truck needs to visit.

4. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 3 is characterized in that: The roulette wheel selection method: The cumulative probability Q of each individual being selected i The calculation formula is as follows: Among them, f(X i ) represents individual X i The fitness of P(X i ) represents the probability of each individual being selected; Q i represents the cumulative probability of each individual being selected.

5. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 3 is characterized in that: The elite population adopts the elite retention strategy selection method to complete the selection operation, and the process is as follows: Step 1: Evaluate individual fitness. Perform fitness evaluation on all individuals in the elite population to determine the fitness value of each individual. Step 2: Select elite individuals, select several individuals with the highest fitness from the elite population as elite individuals; Step 3: Keep the elite individuals and copy them directly to the next generation population. This ensures that at least some individuals in the next generation population are excellent, thus preventing the algorithm from falling into a local optimal solution. Step 4: Fill the remaining population: In addition to the elite individuals, the remaining population vacancies can be filled through crossover mutation operations.

6. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 3 is characterized in that: The crossover operation specifically includes: STEP 1: Extract the rescue priority genes from the parent individuals and put them into the offspring individuals to obtain the basic structure of the offspring individuals; STEP2: Take out the other genes except the rescue priority genes of the two parent individuals respectively, and arrange them closely according to their original order to form two new individuals NF1 and NF2, and then randomly select a gene from the new individuals as the initial crossover position of the circular crossover; STEP3: Select a gene from the new individual NF1 or NF2 as the starting point of the cyclic crossover. Here, it is assumed that the selected crossover position is x in NF1. 1,3 , the starting point of the loop intersection is x 1,3 , then x 1,3 The x in the corresponding position in NF2 2,3 After the swap, return to the parent NF1 to find the same 2,3 The positions of the same coding in the gene are exchanged, and the cycle is repeated, and finally the parent NF2 is found to match x 1,3 The same coding position, the cyclic crossover step ends here, and the individuals NC1 and NC2 generated by the crossover of two new parents NF1 and NF2 are obtained; STEP4: Distribute the genes in the offspring individuals NC1 and NC2 in the original order, according to the number of genes allocated to the rescue priority genes in the parent generations F1 and F2, and evenly distribute them to the offspring C1 and C2 to form complete offspring C1 and C2 with the rescue priority genes. The crossover operation is now completed and the final offspring C1 and C2 are output.

7. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 3 is characterized in that: The mutation operation is to randomly select one of the two mutation operators with equal probability; the two mutation operators are: The first mutation operator uses a double-point exchange mutation. The specific operation method is to extract genes other than the rescue priority gene from the mutant individual, arrange them closely in the original order to form a new individual, and randomly select two mutation points x in the new individual. m With x n , then x m With x n The position of is exchanged to complete the mutation; The second mutation operator adopts a region-based shift operator. The specific operation is as follows: assuming that the gene segment is {x1, x2, x3, x4}, randomly select a gene x2 in the gene segment, compare its relative distance d(x1, x2), d(x3, x2), d(x4, x2) with other genes in the gene segment, and then move gene x2 to the gene with the smallest relative distance d. The mutation is completed and the mutated part is put back to the original individual to obtain the mutated individual.

8. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 4 or 5, characterized in that: To achieve the shortest rescue time for fire trucks, the fitness calculation adopts the inverse of minimizing the rescue time function of fire trucks as the objective function, as shown in the following formula:

9. The method for dispatching vehicles for forest fires in the original forest area in the northern part of Daxinganling according to claim 1, characterized in that: The forest fire data include: the number of fire points, wind speed, temperature at the fire points, wind coefficient, combustible correction coefficient, terrain slope coefficient, the distance between the fire points, the number of fire rescue vehicles, the geographical location of each fire point, the average speed of the fire rescue vehicles, and the extinguishing speed of the fire rescue vehicles.

Images