An emergency resource overall arrangement and deployment method based on improved NSGA-Ⅱ algorithm

Abstract

The application discloses an emergency resource overall arrangement method based on an improved NSGA-Ⅱ algorithm, which comprises the following steps: 1, initializing parameters; 2, executing a generalized reverse learning strategy to generate an optimal initial population; 3, performing non-dominated sorting and genetic operation on the population to generate a sub-population; 4, merging the parent and child populations and eliminating repeated solutions; 5, performing non-dominated sorting on the merged population, calculating the crowding degree of individuals and cyclically deleting the worst individual in the crowding degree, and reserving suitable individuals as a new parent population; 6, calculating the probabilities of adaptive crossover and adaptive mutation; 7, performing selection, crossover and mutation operations on the population to generate a sub-population; 8, repeating steps 4, 5, 6 and 7 until the maximum iteration number is reached; and 9, outputting a pareto solution set, and a decision maker can select a corresponding non-dominated solution as an optimal scheduling scheme according to an actual emergency situation. The application can obtain an optimal scheme for overall arrangement of rescue materials, thereby reducing rescue costs and improving the comprehensive satisfaction of people in disaster areas with rescue.

Description

Technical Field

[0001] The present invention belongs to the field of emergency logistics technology, and more specifically, relates to an emergency resource coordination and allocation method based on an improved NSGA-II algorithm. The present invention can be used to fairly and reasonably allocate emergency supplies in the event of sudden large-scale disasters. Background Art

[0002] The coordinated allocation and dispatch of emergency resources is a core component of emergency logistics management. This research focuses on how to quickly and effectively utilize intelligent decision-making theory and computer technology to select effective emergency material allocation plans and correct delivery routes after a large-scale disaster. This ensures that the collected materials can be delivered to the affected areas in a timely manner, minimizing the significant casualties and economic losses caused by the disaster. Therefore, government emergency management departments and emergency management experts and scholars must develop effective emergency resource coordination and dispatch plans after a disaster to ensure the timely and effective allocation of emergency resources.

[0003] In the coordinated deployment of emergency resources, few researchers have studied the impact of multi-period resource scheduling on the psychological perceptions of disaster victims. This is reflected in model construction. Traditional emergency resource scheduling methods only measure objective economic indicators such as material quantity or time, and then select appropriate algorithms based on the model. In this model construction process, scholars place disaster victims in a passive position and fail to consider the impact of the emergency material scheduling process on their subjective psychology. This can lead to a deviation in the actual perception of the disaster site during post-disaster rescue, failing to truly improve actual rescue efficiency.

[0004] When a disaster strikes, the public suffers enormous losses and pain, manifested in the loss of life and immense psychological trauma. After a disaster, victims urgently need various medical supplies, daily necessities, and rescue supplies to maintain their vital signs. The allocation of emergency resources can not only alleviate the physical pain of victims, but also, to a certain extent, alleviate their psychological trauma. However, the variety and quantity of supplies required in the disaster-stricken areas are vast, making it difficult to raise them in a short period of time, resulting in a situation where supply exceeds demand. At this time, untimely, unreasonable, and unfair allocation plans can further stimulate victims, leading to psychological collapse or actions that disrupt post-disaster rescue order. Therefore, rationally considering the post-disaster psychological perception of victims and coordinating the allocation of emergency supplies based on the actual post-disaster scenario has great application value and research significance in shortening rescue time, improving post-disaster rescue efficiency, and preventing post-disaster rescue activities from deviating from the psychological perception of victims.

[0005] In its patent application number 2017111727992, Hefei University of Technology disclosed a modular scheduling method for emergency materials based on the improved NSGA-II algorithm. The implementation steps of this application are: the first step is to obtain the genetic variables of each chromosome in the initial population based on the relationship between the number of material modules transported from the material storage depot to the distribution center, the capacity limit of each emergency material storage depot, and the transportation volume limit of the rescue materials in transit; the second step is to initialize the algorithm parameters and generate the initial population. Each chromosome in the population represents an emergency material distribution plan. Crossover and mutation iterative operations are performed on the existing population, and the neighborhood set of each chromosome in the population is calculated at the same time; the third step is to introduce a neighborhood-based chromosome elimination mechanism to improve the NSGA-II algorithm, eliminate some chromosomes, and after multiple iterations, finally obtain the Pareto optimal solution set. The shortcomings of this application are: first, in actual disaster relief scenarios, emergency supplies will be in short supply, and various resources urgently need to be transported to the disaster-stricken areas to play their due role. It is impossible to carry out modular packaging at the distribution center, and the packaging process in reality consumes a lot of time, which is not conducive to timely rescue at the disaster-stricken areas; secondly, this method only considers cost indicators and material unsatisfied indicators, does not combine the objective indicators of disaster victims with the objective indicators of rescue, and does not consider the realistic factors of road repair during transportation and multimodal transport after the disaster. It is an idealized model, and the resulting emergency dispatch plan has scenario limitations in actual application. Summary of the Invention

[0006] 1. Technical problem to be solved by the invention

[0007] The purpose of the present invention is to address the deficiencies of the above-mentioned existing technologies and provide an emergency resource coordination and allocation method based on an improved NSGA-Ⅱ algorithm. The present invention solves the coordinated allocation problem in multi-period scheduling of emergency resources, thereby reducing rescue costs, improving the perceived satisfaction of disaster-stricken points, avoiding the deviation of post-disaster emergency rescue from the most intuitive feelings of disaster victims, and providing decision-making support for the optimization of the emergency resource scheduling process.

[0008] 2. Technical solution

[0009] In order to achieve the above object, the technical solution provided by the present invention is:

[0010] The present invention provides an emergency resource coordination and allocation method based on an improved NSGA-II algorithm, comprising the following steps:

[0011] Step 1: Initialize various parameters in the NSGA-Ⅱ algorithm, including: population size popsize, evolutionary generation Gen, number of iterations maxgen, crossover probability P c1 、P c2 , mutation probability P m1 、P m2 , generalization mutation iteration number M;

[0012] Step 2: Initialize the current evolutionary generation Gen=1 and generate a better initial population P0;

[0013] Step 3: Calculate the set of the i-th chromosome in the Gen-th generation under the d-th fitness function in represents the fitness value of the i-th chromosome in the th generation under the d-th fitness function, then the objective function set is F = {f1, f2, ..., f d ,…,f D}, d = 1, 2, ... D;

[0014] Step 4: For the generation population with a population size of N Perform non-dominated sorting, selection, crossover, and mutation operations to generate the Gen-generation population The offspring population The population size is also N;

[0015] Step 5. Calculate the generation population according to step 3 The offspring population The fitness set of chromosome i under multiple objective functions

[0016] Step 6: For the offspring population With the parent population Perform a merge operation to form Its population size is 2N, and the fitness value of the merged population is calculated And delete the duplicate chromosomes with the same fitness value to generate the merged population of the Gen generation Its population size N≤popsize≤2N;

[0017] Step 7: Merge the population Perform fast non-dominated sorting, calculate population crowding, and compare crowding;

[0018] Step 8. Merge the population from the first generation Select a suitable chromosome population of size N as the parent population of the Gen+1 generation

[0019] Step 9, calculate the probability of chromosome adaptive crossover and adaptive mutation;

[0020] Step 10: Collect the Gen+1 parent population Perform selection, crossover, and mutation operations to generate offspring populations

[0021] Step 11: Assign Gen+1 to Gen and determine whether Gen≥maxgen holds. If not, proceed to step 6. Otherwise, it means that maxgen iterations have been completed and population A is finally obtained.

[0022] Step 12: The first level of population A All chromosomes in are output as the Pareto optimal solution set, and the material coordination and allocation plan corresponding to the Pareto optimal solution set is used as the optimal scheduling plan.

[0023] 3. Beneficial effects

[0024] Compared with the existing known technologies, the technical solution provided by the present invention has the following significant effects:

[0025] (1) The present invention constructs a satisfaction function of the time it takes for a distribution center to reach a disaster site, and combines the comparison factor of the disaster site after the disaster to construct a comprehensive time perception satisfaction function of the disaster site; at the same time, considering the shortage of supplies, the loss of unsatisfied supplies and the comparison damage between disaster sites are integrated, and the comprehensive material loss function of the disaster site is constructed in combination with the material transportation loss; and calculates the multi-period emergency material transportation cost, and establishes a multi-objective optimization model to solve the problem of deviation between the actual rescue process and the perception of disaster victims in the multi-period scheduling of emergency materials.

[0026] (2) The present invention improves the initial population generation process in the NSGA-Ⅱ algorithm by generalizing the reverse learning strategy, ensuring the diversity of chromosomes in the initial population and improving the quality of the initial population and the convergence performance of the algorithm.

[0027] (3) Before performing genetic operations, the present invention introduces the steps of calculating the adaptive crossover and adaptive mutation probabilities. Among them, the adaptive crossover probability enables the algorithm to focus on global search in the early stage and gradually focus on local search in the later stage. The generalized polynomial mutation can effectively improve the algorithm's local search ability, maintain population diversity, and improve the algorithm's search performance as a whole. The adaptive mutation probability gradually increases the chromosome mutation probability in the later stage of iteration, which increases the population diversity. After repeated iterations, a Pareto solution in the optimal solution set is selected as the optimal emergency resource allocation plan. The NSGA-Ⅱ algorithm solves the problem of timely and fair distribution of emergency material allocation plans.

[0028] (4) The present invention designs a generalized polynomial mutation mechanism based on the polynomial mutation process, uses generalized reverse learning to generate more neighborhood solutions, prompts the algorithm to jump out of the local optimum, improves the algorithm's local search ability, maintains population diversity, and overall improves the algorithm's search performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1This is a flow chart of the emergency resource coordination and allocation method using the improved NSGA-Ⅱ algorithm adopted in the present invention;

[0030] Figure 2 It is a process diagram of the multi-cycle emergency resource coordinated deployment in the present invention. DETAILED DESCRIPTION

[0031] In order to further understand the content of the present invention, the present invention is described in detail with reference to the accompanying drawings and embodiments.

[0032] Example 1

[0033] The emergency resource coordination and allocation method based on the improved NSGA-II algorithm in this embodiment has the following process: Figure 1 As shown in the figure, based on the characteristics of the collection, type and quantity of relief supplies, a multi-cycle allocation method for emergency resources under disaster conditions is studied. That is, the distribution point collects supplies, transports the supplies to the distribution center by vehicles, and then the distribution center uses multiple transportation methods to allocate them to the disaster site. The overall process is as follows Figure 2 This embodiment uses an improved NSGA-II algorithm to solve the problem, thereby obtaining a timely, reasonable, and fair rescue material dispatching plan, thereby reducing the total cost of multi-cycle rescue work, improving rescue efficiency, and improving the perceived satisfaction of disaster victims at the disaster site.

[0034] Specifically, it is applied to a three-level emergency dispatch environment consisting of a set of rescue material types E = {1, 2, ..., e} consisting of e kinds of rescue materials, a set of l emergency material distribution points H = {1, 2, ..., l}, a set of m emergency material distribution centers I = {1, 2, ..., m}, a set of J disaster-stricken points J = {1, 2, ..., n}, a set of K transportation tools K = K1∪K2, and a set of T periods T = {1, 2, ..., w}. Considering the emergency material dispatch transportation volume, damage volume, transportation time, transportation distance, and road repair conditions, a comprehensive cost function and a disaster victim comprehensive perception function are constructed, where K1 = {1, 2, ...k1} represents the vehicle transportation set, K2 = {1, 2, ...k2} represents the helicopter transportation set, and there is no road network damage between the distribution point H and the distribution center I, so vehicle transportation can be used. The fixed transportation and unit transportation costs are respectively There may be road damage between distribution center I and the disaster site. A combined transportation method of vehicles and helicopters is used. The specific transportation method is determined by the road parameters. The fixed cost and unit transportation cost are: The maximum transport volume of materials e from distribution center i to disaster site j in the tth period is The actual demand for emergency supplies e at disaster site j in period t is

[0035] The steps of the emergency resource coordination and deployment method are as follows:

[0036] Step 1: Initialize various parameters in the NSGA-Ⅱ algorithm, including: population size popsize, evolutionary generation Gen, number of iterations maxgen, crossover probability P c1 、P c2 , mutation probability P m1 、P m2 , generalization mutation iteration number M;

[0037] Step 2: Initialize the current evolutionary generation Gen = 1 and generate a better initial population P0, which is the P participating in the iteration in the Gen generation. parent ;

[0038] Step 2.1: Use real number matrix coding to describe the amount of materials distributed from the distribution center to the disaster site in each cycle. The matrix A = (a1, a2, ..., a j ,…,a J ) is the solution in the target solution space, and the column vector in the matrix The number of rows in the matrix represents the distribution center i, the number of columns in the matrix represents the disaster site j, and the chromosome encoding gene is represented by the vector Indicates that the gene encoded in the vector It represents the quantity of corresponding type of materials e transported by distribution center i to disaster site j;

[0039] Step 2.2: First, randomly and uniformly generate each gene value according to certain constraints to form the initial population P1. Then, adopt the generalized reverse learning strategy to form the generalized population P2. The initial population P1 and the generalized population P2 are merged. After comparing the non-dominated sorting and crowding, the first N chromosomes with better fitness are selected to form the initial population P0. In generating the initial population, the diversity and quality of the initial population are guaranteed. Specifically,

[0040] Step 2.2.1. Generate a real number matrix encoding the gene values ​​of A, i.e. the number of emergency supplies e in the i-th distribution center Make in This generates the distribution center's emergency material transportation volume set Satisfying the relations (1) and (2), the chromosome matrix column vectors are generated in turn The initial population P1 can be obtained;

[0041]

[0042]

[0043] Step 2.2.2, encode the column vector a on the matrix of the initial population P1 j , which is in n-dimensional space, Considered as zi , then z i is a real number between [0, B], It is converted into Where γ∈[0,1], for out-of-range The generalized population P2 can be obtained by making corrections according to formula (3).

[0044]

[0045] Step 2.2.3: After non-dominated sorting and crowding calculation, the initial population P1 and the generalized population P2 are compared for crowding to select chromosomes with better non-dominated order or crowding to enter the initial population P0.

[0046] Step 3: Calculate the set of the i-th chromosome in the Gen-th generation under the d-th fitness function in represents the fitness value of the i-th chromosome in the th generation under the d-th fitness function, then the objective function set is F = {f1, f2, ..., f d ,…f D}, d = 1, 2, ... D;

[0047] Step 4: For the generation population with a population size of N Perform non-dominated sorting, selection, crossover, and mutation operations to generate the Gen-generation population The offspring population The population size is also N;

[0048] Step 5. Calculate the generation population according to step 3 The offspring population The fitness set of chromosome i under multiple objective functions

[0049] Step 6: For the offspring population With the parent population Perform a merge operation to form Its population size is 2N, and the fitness value of the merged population is calculated And delete the duplicate chromosomes with the same fitness value to generate the merged population of the Gen generation Its population size N≤popsize≤2N;

[0050] Step 7: Merge the population Perform fast non-dominated sorting, calculate population crowding, and compare crowding;

[0051] Step 7.1: In the emergency resource allocation problem, let the objective function set F = {f1, f2, f3}, where f1 represents the maximization of disaster victims' perceived time satisfaction, f2 represents the minimization of the perceived loss of comprehensive materials, and f3 represents the minimization of comprehensive cost. Then D = 3, d∈{1, 2, 3}; the objective function F = {f1, f2, f3} in the emergency resource allocation problem is calculated as follows:

[0052]

[0053]

[0054] f3=C1+C2 (6)

[0055] In the above formula (4), f1 represents the comprehensive perceived time satisfaction of all disaster-affected points in all periods, It is the comprehensive perceived time satisfaction that integrates waiting time satisfaction and comparison level; in formula (5), f2 represents the comprehensive material loss pain effect of all disaster-stricken points in all periods. The comprehensive material perception loss of the disaster site that integrates the material waiting loss and the comparison loss, V(ρ j ) is the perceived loss of material damage at the disaster site under prospect theory; in formula (6), f3 represents the material transportation cost of all disaster sites in all periods, C1 represents the multi-period material collection cost and transportation cost, and C2 represents the multi-period post-disaster road repair cost.

[0056] Step 7.2: Merge the population of the first generation Perform fast non-dominated sorting to obtain the set of chromosome levels in represents the kth level in the Gen-th generation, and n represents the number of levels divided;

[0057] Step 7.3: For the chromosome set of level k The crowding degree of the boundary chromosome is infinite, and the crowding degree of the remaining chromosomes is calculated according to formula (7);

[0058]

[0059] In formula (7), d i represents the crowdedness of chromosome i, is the dth objective function value of individual i+1, i-1, f dmax With f dmin is the maximum and minimum value of the corresponding objective function;

[0060] Step 7.4: Chromosome obtains non-dominated order k and crowding distance d i Attributes, use these two chromosome attributes to compare any two chromosomes, if chromosome i is in the set If the level is smaller, i is dominant; if the two chromosomes are in the same chromatin level set But the crowdedness of chromosome i is d i If it is larger, it can also be determined that chromosome i is dominant;

[0061] Step 8: Merge the population from the Gen generation Select a suitable chromosome population of size N as the parent population of the Gen+1 generation

[0062] Step 8.1, initialize k=1;

[0063] Step 8.2: Select all chromosomes in the kth level and retain them in the Gen+1th generation parent population set middle;

[0064] Step 8.3: Determine the Gen+1 generation parent population set Is the total number of chromosomes in greater than N? If not, proceed to step 8.5; if so, proceed to step 8.4;

[0065] Step 8.4: For the k-th level chromosome Calculate the crowding degree and perform descending sorting, delete the individual with the smallest crowding degree within the k level; then recalculate The chromosome crowding degree in the set is sorted in descending order, and the individuals with the smallest crowding degree are deleted until the sum of the number of individuals retained in this level and the number of chromosomes in the first k-1 levels is N, and then jump to step 9;

[0066] Step 8.5: Assign k+1 to k and continue with step 8.2.

[0067] Step 9. Calculate the adaptive crossover and adaptive mutation probabilities of chromosomes. The adaptive crossover probability enables the algorithm to focus on global search in the early stage and gradually focus on local search in the later stage, which improves the algorithm search performance. The adaptive mutation probability gradually increases the chromosome mutation probability in the later stage of iteration, increasing population diversity. The specific calculation of the adaptive crossover and mutation probabilities is shown in the following formula.

[0068]

[0069] Among them, p c is the adaptive crossover probability of the current generation, p c1 、p c2 is the maximum and minimum crossover probability, Gen is the current iteration number, and maxgen is the maximum iteration number;

[0070]

[0071] Among them, p mis the probability of chromosome mutation in the current generation, p m1 、p m2 is the maximum and minimum mutation probability, Gen is the current iteration number, and maxgen is the maximum iteration number.

[0072] Step 10: Collect the Gen+1 parent population Perform selection, crossover, and mutation operations to generate offspring populations Utilize generalized reverse learning to generate more neighborhood solutions, prompt the algorithm to jump out of the local optimum, improve the algorithm's local search capability, and maintain population diversity. The generalized polynomial mutation steps used in step 10 are as follows:

[0073] Step 10.1: For the chromosome i after crossover, loop through the gene positions of each chromosome;

[0074] Step 10.2: Generate a random number rand for the gene position on the chromosome and compare it with the adaptive mutation probability p m Size, if rand<p m , then execute step 10.4; otherwise, do not mutate the gene position and continue to step 10.7;

[0075] Step 10.3: Let t = 1, M = 10;

[0076] Step 10.4: Perform polynomial mutation operation on the position gene. The position gene is Mutation Perform generalized reverse learning operations on the mutated individuals to generate

[0077] Step 10.5: Store the polynomially mutated and generalized chromosomes in the set S. Determine the size of t and M. If t is less than M, assign t+1 to t and proceed to step 10.4. Otherwise, proceed to step 10.7.

[0078] Step 10.6: Place the original chromosome i into the set S, perform a non-dominated sort on the chromosomes in the set S, select the best chromosome in the set S to replace the original chromosome i, and output the best chromosome.

[0079] Step 10.7: Continue traversing the next gene position and execute step 10.2;

[0080] Step 11: Assign Gen+1 to Gen and determine whether Gen≥maxgen holds. If not, proceed to step 6. Otherwise, it means that maxgen iterations have been completed and population A is finally obtained.

[0081] Step 12: The first level of population A All chromosomes in are output as the Pareto optimal solution set, and the material coordination and allocation plan corresponding to the Pareto optimal solution set is used as the optimal scheduling plan.

[0082] The above is a schematic description of the present invention and its embodiments, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs a structure and embodiment similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A method for coordinated deployment of emergency resources based on an improved NSGA-II algorithm, characterized in that: Here are the steps: Step 1: Initialize various parameters in the NSGA-II algorithm, including: population size popsize, evolutionary generation Gen, number of iterations maxgen, crossover probability P c1 、P c2 , mutation probability P m1 、P m2 , generalization mutation iteration number M; Step 2: Initialize the current evolutionary generation Gen=1 and generate a better initial population P0; Step 3: Calculate the set of the i-th chromosome in the Gen-th generation under the d-th fitness function in represents the fitness value of the i-th chromosome in the th generation under the d-th fitness function, then the objective function set is F = {f1, f2, ..., f d ,…f D }, d = 1, 2, ... D; Step 4: For the generation population with a population size of N Perform non-dominated sorting, selection, crossover, and mutation operations to generate the Gen-generation population The offspring population The population size is also N; Step 5. Calculate the generation population according to step 3 The offspring population The fitness set of chromosome i under multiple objective functions Step 6: For the offspring population With the parent population Perform a merge operation to form Its population size is 2N, and the fitness value of the merged population is calculated And delete the duplicate chromosomes with the same fitness value to generate the merged population of the Gen generation Its population size N≤popsize≤2N; Step 7: Merge the population Perform fast non-dominated sorting, calculate population crowding, and compare crowding; Step 8. Merge the population from the first generation Select a suitable chromosome population of size N as the parent population of the Gen+1 generation Step 9, calculate the probability of chromosome adaptive crossover and adaptive mutation; Step 10: Collect the Gen+1 parent population Perform selection, crossover, and mutation operations to generate offspring populations Among them, the parent population set Perform tournament selection, simulated binary crossover, and generalized polynomial mutation operations in sequence; The generalized polynomial mutation steps used are as follows: Step 10.1: For the chromosome i after crossover, loop through the gene positions of each chromosome; Step 10.2: Generate a random number rand for the gene position on the chromosome and compare it with the adaptive mutation probability p m Size, if rand <p m , then execute step 10.4; otherwise, do not mutate the gene position and continue to step 10.7; Step 10.3: Let t = 1, M = 10; Step 10.4: Perform polynomial mutation operation on the position gene. The position gene is Mutation Perform generalized reverse learning operations on the mutated individuals to generate Step 10.5: Store the polynomially mutated and generalized chromosomes in the set S. Determine the size of t and M. If t is less than M, assign t+1 to t and proceed to step 10.

4. Otherwise, proceed to step 10.

7. Step 10.6: Place the original chromosome i into the set S, perform a non-dominated sort on the chromosomes in the set S, select the best chromosome in the set S to replace the original chromosome i, and output the best chromosome. Step 10.7: Continue traversing the next gene position and execute step 10.2; Step 11: Assign Gen+1 to Gen and determine whether Gen≥maxgen holds. If not, proceed to step 6. Otherwise, it means that maxgen iterations have been completed and population A is finally obtained. Step 12: The first level of population A All chromosomes in are output as the Pareto optimal solution set, and the material coordination and allocation plan corresponding to the Pareto optimal solution set is used as the optimal scheduling plan.

2. The method for coordinating and allocating emergency resources based on the improved NSGA-II algorithm according to claim 1 is characterized in that: The process of generating the initial population P0 in step 2 is as follows: Step 2.1: Use real number matrix coding to describe the amount of materials distributed from the distribution center to the disaster site in each cycle. The matrix A = (a1, a2, ..., a j ,…,a J ) is the solution in the target solution space, and the column vector in the matrix The number of rows in the matrix represents the distribution center i, the number of columns in the matrix represents the disaster site j, and the chromosome encoding gene is represented by the vector Indicates that the gene encoded in the vector It represents the quantity of corresponding type of materials e transported by distribution center i to disaster site j; Step 2.2: First, randomly and uniformly generate gene values ​​according to certain constraints to form the initial population P1. Then, adopt the generalized reverse learning strategy to form the generalized population P2. Merge the initial population P1 with the generalized population P2. After comparing the non-dominated sorting and crowding, select the first N chromosomes with better fitness to form the initial population P0.

3. The method for coordinating and allocating emergency resources based on the improved NSGA-II algorithm according to claim 2 is characterized in that: The specific steps of step 2.2 are as follows: Step 2.2.

1. Generate a real number matrix encoding the gene values ​​of A, i.e. the number of emergency supplies e in the i-th distribution center Make in This generates the distribution center's emergency material transportation volume set Satisfying the relations (1) and (2), the chromosome matrix column vectors are generated in turn The initial population P1 can be obtained; Step 2.2.2, encode the column vector a on the matrix of the initial population P1 j , which is in n-dimensional space, Considered as z i , then z i is a real number between [0, B], It is converted into Where γ∈[0,1], for out-of-range According to formula (3), the generalized population P2 can be obtained by making corrections; Step 2.2.3: After non-dominated sorting and crowding calculation, the initial population P1 and the generalized population P2 are compared for crowding to select chromosomes with better non-dominated order or crowding to enter the initial population P0.

4. The method for coordinating and allocating emergency resources based on the improved NSGA-II algorithm according to claim 3 is characterized in that: The process of step 7 is: Step 7.1: In the emergency resource allocation problem, let the objective function set F = {f1, f2, f3}, where f1 represents the maximization of disaster victims' perceived time satisfaction, f2 represents the minimization of the perceived loss of comprehensive materials, and f3 represents the minimization of comprehensive cost. Then D = 3, d∈{1, 2, 3}; Step 7.2: Merge the population of the first generation Perform fast non-dominated sorting to obtain the set of chromosome levels in represents the kth level in the Gen-th generation, and n represents the number of levels divided; Step 7.3: For the chromosome set of level k The crowdedness of the boundary chromosomes is infinite, and the crowdedness of the remaining chromosomes is calculated according to formula (4); d i represents the crowdedness of chromosome i, is the dth objective function value of individual i+1, i-1, f dmax With f dmin is the maximum and minimum value of the corresponding objective function; Step 7.4: Chromosome obtains non-dominated order k and crowding distance d i Attributes, use these two chromosome attributes to compare any two chromosomes, if chromosome i is in the set If the level is smaller, i is dominant; if the two chromosomes are in the same chromatin level set But the crowdedness of chromosome i is d i If it is larger, it can also be determined that chromosome i is dominant.

5. The method for coordinating and allocating emergency resources based on the improved NSGA-II algorithm according to claim 4 is characterized in that: The objective function F = {f1, f2, f3} in step 7.1 is calculated as follows: f3=C1+C2 (7) In the above formula (5), f1 represents the comprehensive perceived time satisfaction of all disaster-affected points in all periods, It is the comprehensive perceived time satisfaction that integrates the waiting time satisfaction and the comparison level; in formula (6), f2 represents the comprehensive material loss pain effect of all disaster-stricken points in all periods. The comprehensive material perception loss of the disaster site that integrates the material waiting loss and the comparison loss, V(ρ j ) is the perceived loss of material damage at the disaster site under prospect theory; in formula (7), f3 represents the material transportation cost of all disaster sites in all periods, C1 represents the multi-period material collection cost and transportation cost, and C2 represents the multi-period post-disaster road repair cost.

6. The method for emergency resource coordination and deployment based on the improved NSGA-II algorithm according to claim 5, characterized in that: The process of step 8 is: Step 8.1, initialize k=1; Step 8.2: Select all chromosomes in the kth level and retain them in the Gen+1th generation parent population set middle; Step 8.3: Determine the Gen+1 generation parent population set Is the total number of chromosomes in greater than N? If not, proceed to step 8.5; if so, proceed to step 8.4; Step 8.4: For the k-th level chromosome Calculate the crowding degree and perform descending sorting, delete the individual with the smallest crowding degree within the k level; then recalculate The chromosome crowding degree in the set is sorted in descending order, and the individuals with the smallest crowding degree are deleted until the sum of the number of individuals retained in this level and the number of chromosomes in the previous k-1 levels is N, and then jump to step 9; Step 8.5: Assign k+1 to k and continue with step 8.

2.

7. The method for emergency resource coordination and deployment based on the improved NSGA-II algorithm according to claim 6, characterized in that: The adaptive crossover and adaptive mutation probabilities described in step 9 are calculated according to formulas (8) and (9) respectively; Among them, p c is the adaptive crossover probability of the current generation, p c1 、p c2 is the maximum and minimum crossover probability, Gen is the current iteration number, and maxgen is the maximum iteration number; Among them, p m is the probability of chromosome mutation in the current generation, p m1 、p m2 is the maximum and minimum mutation probability, Gen is the current iteration number, and maxgen is the maximum iteration number.

8. A method for coordinated deployment of emergency resources based on an improved NSGA-II algorithm according to any one of claims 1 to 7, characterized in that: The paper is applied to a three-level emergency dispatch environment consisting of a set of e types of relief supplies E = {1, 2, ..., e}, a set of l emergency material distribution points H = {1, 2, ..., l}, a set of m emergency material distribution centers I = {1, 2, ..., m}, a set of J disaster-affected points J = {1, 2, ..., n}, a set of K types of transportation tools K = K1∪K2, and a set of T periods T = {1, 2, ..., w}. Considering the transportation volume, damage volume, transportation time, transportation distance, and road repair conditions of emergency material dispatch, a comprehensive cost function and a comprehensive perception function of disaster victims are constructed. Among them, K1 = {1, 2, ...k1} represents the vehicle transportation set, K2 = {1, 2, ...k2} represents the helicopter transportation set, and there is no road network damage between the distribution point H and the distribution center I. Vehicle transportation is adopted, and its fixed transportation and unit transportation costs are respectively If there is road damage between distribution center I and the disaster site, a combined transportation method of vehicles and helicopters is used. The specific transportation method is determined by the road parameters. Its fixed cost and unit transportation cost are: The maximum transport volume of materials e from distribution center i to disaster site j in the tth period is The actual demand for emergency supplies e at disaster site j in period t is

Images