CN110264026B

CN110264026B - Rescue aircraft task allocation based on two-stage planning in earthquake rescue

Info

Publication number: CN110264026B
Application number: CN201910322250.XA
Authority: CN
Inventors: 张洪光; 王趱; 梁子涵; 刘元安; 谢刚; 冉静; 刘宇泓; 王瑞
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2019-04-22
Filing date: 2019-04-22
Publication date: 2021-07-13
Anticipated expiration: 2039-04-22
Also published as: CN110264026A

Abstract

The invention discloses a rescue aircraft task allocation method (TLP) based on a two-stage planning algorithm in earthquake rescue. The method comprises the following steps: earthquake rescue simulators and two-stage planning rescue algorithms (TLPs). The earthquake rescue simulator simulates the conditions of disaster-stricken personnel in ground medical areas and disaster-stricken areas in earthquakes. The TLP algorithm is used for performing online optimization on task allocation of the rescue aircraft, and comprises three parts, namely rescue task level planning, rescue aircraft level planning and rescue aircraft position allocation. And the first part is used for finishing the clustering of rescue tasks and the task allocation of rescue airplanes. The second part is used for carrying out rescue target sequence planning on each task group and the corresponding rescue aircraft group based on the particle swarm optimization, and the third part is used for carrying out position distribution on each rescue aircraft after the task is completed, so that the rescue work in the next rescue period is facilitated. The TLP algorithm maximizes the total number of successful rescue under the condition that the occurrence of earthquake and aftershock is random and unpredictable, and improves the rescue effect.

Description

Rescue aircraft task allocation based on two-stage planning in earthquake rescue

Technical Field

The invention belongs to the technical field of computer application, particularly relates to task allocation of a rescue airplane in earthquake rescue, and particularly relates to a task allocation method of an earthquake rescue airplane based on two-stage planning. The method can be used for realizing effective task allocation of the rescue aircraft on line so as to evacuate disaster-stricken personnel from a disaster area as soon as possible after an earthquake occurs and ensure the life safety of the personnel.

Background

Earthquake disasters have the characteristics of strong burst and large destructiveness, and meanwhile, the extrusion and collision among the plates cause the dislocation and the breakage of the inner parts of the plates and the edges of the plates, so that strong secondary disasters are caused, and the defense is extremely difficult. Therefore, in order to reduce social and economic losses as much as possible, a response needs to be made in time, and emergency rescue of disaster-stricken personnel in the earthquake-stricken area is planned. Compared with emergency rescue under other scenes, the earthquake rescue has the difficulties that the position of an earthquake occurrence point has randomness, the number of people suffered from a single earthquake area is large, and the high efficiency of response corresponding to the emergency rescue has higher requirements. Therefore, rescue resources need to be reasonably scheduled, an effective disaster relief mechanism is established, and reasonable configuration is carried out according to local rescue resources.

In recent years, rescue airplanes have been widely used in disaster management, and have an important role in dealing with natural disasters, accident disasters, social security events, and the like. Such as ALTI Transition VTOL of unmanned vertical take-off and landing hybrid aircraft, which is introduced by ALTI UAS of America, and mdSA of miniature water rescue aircraft, which is produced by Microdrones of Germany. In China, there are also many examples of applying rescue airplanes to earthquake rescue. For example, in the extreme large earthquake of Wenchuan, China successfully practices a' large airplane rescue mode, and armed police use a multi-rotor rescue airplane to participate in earthquake relief work for the first time in 2014. In addition, the mapping system based on the rescue airplane can also acquire image data of earthquake-stricken areas, meets the requirements of acquiring real-time disaster-stricken data and ensuring rescue effectiveness, and is beneficial to improving the efficiency of rescue airplane scheduling in earthquake rescue.

The existing earthquake rescue algorithm mainly comprises the following steps: (1) based on the study of the group behavior, a stable group pattern can be determined by studying the movement behavior of the group in an emergency. (2) The determination of the position of the temporary rescue center is an important component of city planning and emergency management, and plays an important role in ensuring the efficiency of emergency medical service and other rescues. (3) The post-earthquake emergency logistics scheduling is an important task after a large earthquake disaster, and provides real-time guidance for post-earthquake logistics scheduling by considering the influence of a dynamic disaster environment. (4) The rescue decision system can provide effective emergency management decisions, improve emergency rescue efficiency and reduce disaster loss.

The occurrence of earthquake has great uncertainty, so that an exact formula or model cannot be used for simulating real-time earthquake and other data information such as the position, time, magnitude and the like of the occurrence of aftershock. Therefore, an earthquake rescue model needs to be established according to actual rescue requirements. Most of the existing earthquake rescue methods focus on group behavior research in a certain area, emergency selection of a temporary rescue center after a disaster occurs, efficient implementation of rescue material scheduling, rapid decision making to achieve earthquake loss reduction and the like, and how to rescue disaster-stricken people generated in real time in an earthquake is not considered. A rescue algorithm needs to be designed on the basis of a rescue model, and real-time rescue planning of disaster-stricken people is achieved.

Disclosure of Invention

The embodiment of the invention provides a rescue aircraft online task allocation algorithm based on a two-stage optimization algorithm (TLP) in earthquake rescue. Effective task allocation of the rescue aircraft is realized through online optimization, and people in a disaster are evacuated from the disaster area as soon as possible after an earthquake occurs, so that the life safety of people is ensured.

In order to achieve the above object, an embodiment of the present invention provides a two-stage optimization algorithm (TLP) based rescue aircraft online task allocation method, and fig. 1 is a system architecture diagram. The earthquake rescue simulator is designed to simulate the generation of a disaster area, the distribution of disaster-stricken persons and the rescue situation of a rescue airplane in the process of sustaining earthquake disasters, and the effective task allocation of the rescue airplane is completed through the planning of a rescue task level and a rescue airplane level according to the disaster-stricken situation. The method comprises the following steps:

the earthquake rescue simulator is designed and used for simulating disaster conditions in a ground medical area and a severe disaster area in the earthquake occurrence process and simultaneously realizing rescue of a rescue airplane in a certain area after generating disaster-stricken personnel in the area, and the frame of the earthquake rescue simulator is shown as the attached drawing 2. The rescue simulator comprises the interaction between the rescue aircraft and various factors influencing the number of people suffering from a disaster (such as the relation between rescue time and the generation time of the people suffering from the disaster, the survival rate of the people suffering from the disaster and the like). The earthquake rescue simulator mainly comprises:

(1) and (3) carrying out map simulation on the earthquake-stricken area, wherein a two-dimensional map related to the size of an actual map is used and corresponds to a population density map and an earthquake intensity map. Grid maps are grids that define seismic maps, each grid having corresponding map attributes, such as: population density attribute, earthquake intensity condition, number of people suffering from a disaster and generation time of people suffering from a disaster. The ground medical area in the grid map is a rescue point for simple rescue in the earthquake area, and one ground medical area is a rescue task. The number and location distribution of the ground medical areas depends on the population density map and the seismic intensity map. In areas with high population density and large earthquake intensity, the number of ground medical areas is large and the distribution is dense; for areas with low population density and low seismic intensity, the number of ground medical areas will be small and the distribution will be sparse. The safe area refers to a hospital, a rehabilitation area and the like for providing the wounded with treatment outside the earthquake area. The distribution of the safety zones may be non-uniform, and the location of the safety zones may be determined according to the distribution of the ground medical treatment zones and the environment outside the seismic area.

(2) The generation and distribution of the disaster-stricken persons, each grid in the map covers a disaster-stricken area which is possible to have earthquakes, when one grid is in a disaster, the surrounding grids with people can be caused to generate a certain number of disaster-stricken persons, and the grids containing the disaster-stricken persons are a rescue task. According to the population density graph and the seismic intensity graph, disaster victims are generated periodically, and the method comprises the following steps:

ground medical area disaster recovery personnel: according to the population density chart, the actual casualties RAN, the simulation times SN, the set MA of the ground medical areas and the number n of the ground medical areas_maCalculating the initial point of each ground medical treatment area in each periodThe number of persons who have suffered a disaster AN (MAi). First, an SN row n is generated_maNormalized matrix of columns, M, requires:

and is

The initial number of people in disaster in each ground medical area in the nth rescue cycle can be represented as:

disaster-stricken personnel in the disaster area: according to the set MA of the ground medical treatment area and the distance threshold value R_thNumber threshold value N of severe disaster areas_thDisaster-stricken number threshold AN of severe disaster-stricken area_thAnd obtaining a set SAA of the heavily affected areas.

Step 1: the number of severe disaster areas generated is as follows: n ═ 1, N_th]Inner uniform random integer

Step 2: randomly generating coordinates (x, y) in a grid, and calculating the closest distance d between (x, y) and MA_minIf d is_min>R_thThen (x, y) is included in the SAA, and AN (x, y) is made [1, AN [ ]_th]Inner uniform random integer

And step 3: repeating the step 2 until the number of the elements in the SAA is N

(3) The survival rate, also called survival rate, of a disaster-stricken person is the ratio of the number of people that a certain patient or wounded person survives in a period of time in the future to the total number of people, and can be generally described as:

survival rate (number of persons the mth ST survived/total number at the beginning) × 100%.

Where ST is the survival statistic cycle. To visually represent this trend of decreasing survival rate over time, the EI-based piecewise function is used to describe: SR (EI)^(m-1)(

m

1,2,3, …), i.e. of the m ST disaster-stricken personSurvival was EI-fold higher than m-1 ST. The smaller the EI, the more injury the person suffering from the disaster is, and the more obvious the survival rate of the person suffering from the disaster is reduced along with the time. FIG. 3 is a graph showing the survival rate at 15 minutes ST, 0.7 EI and 10 minutes ST, 0.5 EI. Different statistical periods ST and impact parameters EI can be set according to the actual situation.

(4) Rescue airplane rescue, which is divided into single airplane rescue and multi-airplane rescue scheduling. Rescue of a single rescue airplane: firstly, the number AN of disaster-stricken persons at the time (x, y) of t is calculated^t(x, y), and then calculating the time t + t for the rescue aircraft u to reach the rescue target^u _pTiAnd the number of remaining disaster victims at arrival

If the load of u

When the residual capacity is not enough to complete the rescue target, the user needs to return to the safety area nearby. When it can be done, calculate rescue plane u goes to rescue target (x)_i,y_i) Number of people who succeed in post-rescue N_u,iAnd u is required to satisfy the cycle time constraint during rescue.

Rescue scheduling of a plurality of airplanes: fig. 4 shows a rescue schedule for the co-rescue of U-frame rescue planes. Firstly, according to the disaster situations of the ground medical area and the severe disaster area in the nth period and the position information of each rescue aircraft, a task-level planning algorithm of a TLP (transient liquid phase) rescue algorithm is used for distributing a rescue target sequence TS (transport stream) to each rescue aircraft_u＝[T_u,1,T_u,2,…,T_u,i,…,T_u,nu]. And then, the rescue aircrafts complete rescue tasks in sequence according to the sequence of the respective rescue target sequences.

The TLP rescue algorithm, the motivation of TLP, is to maximize the total number of successful rescues and reduce the total number of failed rescues as much as possible. The TLP is mainly composed of three parts, namely rescue task level planning, rescue airplane level planning and rescue airplane position distribution. Figure 5 shows a block diagram of the TLP rescue algorithm and an example of the on-line task allocation of the rescue aircraft.

(1) And (4) planning the rescue task level, wherein the rescue task level finishes the optimal clustering of task groups and the distribution of the rescue airplane groups corresponding to the task groups according to the rescue task condition in the nth rescue period and the distribution condition of the rescue airplanes. Through proper clustering, the global optimization problem is simplified into a regional centralized optimization problem, and unnecessary flight distance of the rescue aircraft in the rescue process can be effectively reduced. And the rescue task level planning is based on the grid coordinates of each rescue task as clustering basis, completes the clustering of the rescue tasks and determines the clustering center of each task group. And then determining the number of rescue airplanes required by each task group according to the proportional relation of the total number of the disaster-stricken persons of each group of rescue tasks. And calculating the distance between each rescue aircraft and the clustering center, and dividing the rescue aircraft groups to complete corresponding rescue tasks. Rescue mission level planning can be divided into two steps.

Firstly, clustering rescue tasks by adopting a K-means clustering method. K value in K-means clustering is set to [2, K_max]Random integer of (a), wherein K_maxThe number of rescue missions in the nth rescue period is the minimum value of the total number U of the rescue aircrafts. Rescue center m for calculating each K value_i(i ═ 1,2, …, K), and the sum of the squares of errors (SSE) for each value of K is recorded,

wherein, C_iIs of the ith class, p is C_iCoordinate value of (1), m_iIs C_iThe center of mass of the rescue center. Finally, selecting the K value corresponding to the elbow as the optimal clustering number K by using an elbow method_bestAnd is combined with K_bestAnd setting the clustered output result as the optimal clustering scheme of the rescue task.

And secondly, completing the division of the rescue aircraft groups according to the proportion of the disaster-suffered total number of people of each task group and the distance from each rescue aircraft to the rescue center. Generation of K_bestAnd the rescue aircraft groups correspond to the task groups one by one.

(2) And (4) rescue plane level planning, wherein the rescue plane level planning sequentially finishes the planning of the rescue task sequence of each rescue plane in each rescue aircraft group. Fig. 6 is an operational flow diagram of a rescue aircraft-level planning of a TLP rescue algorithm in the nth rescue cycle. And optimizing the rescue task sequence of the rescue aircraft based on the particle swarm algorithm.

And (3) an encoding mode: a two-section coding method is adopted in TLP rescue airplane-level planning. Firstly, a rescue factor is set

Wherein L is_maxThe maximum load of the aircraft. Alpha represents the number of rescue planes needed without casualties. And expanding each rescue task into alpha rescue targets according to the alpha of the rescue task, and numbering the rescue targets. The maximum number of people to be rescued contained in each rescue target is L_max. And a two-part coding method is adopted to distribute the rescue target to the rescue aircraft. The first part is a target part, the coding length is the number of rescue targets, and the rescue targets are arranged by adopting symbol coding. The second part is a breakpoint part, the coding length is the number of rescue airplanes-1, and each section of target divided by the breakpoint value is the rescue target distributed to each rescue airplane. Wherein the setting of the individual breakpoints should satisfy the constraint conditions.

A random operator: four random operators are used for carrying out deep search on the neighbors of all individuals in the population, and the wide optimization of a space solution set is realized. When a new individual is generated, two parts of a target and a breakpoint are comprehensively considered, and four random operators are as follows: two-point exchange, single-point insertion, section exchange and section reverse order. Fig. 7 is a schematic diagram of four random operators.

Fitness function: the fitness function in the earthquake rescue process is described as follows:

F(X)＝max(β×RN(X,P^t)-(1-β)×FN(X,P^t))

wherein the weight coefficient beta is ∈ [0,1 ]]，RN(X,P^t) And FN (X, P)^t) Respectively as the total of successful rescue of all rescue airplanes in the nth periodNumber and total number of rescue failures. Let AN (x, y) be the initial distribution of disaster-stricken people, AN^t(x, y) is the distribution of the disaster stricken persons at the end of the rescue period.

Constraint conditions are as follows:

(3) the rescue aircraft position is distributed, and in the nth rescue period, the rescue termination time rt of the rescue aircraft u exists_uIf the time is shorter than the end time nT of the nth period, a rescue aircraft position distribution method is needed to fully utilize the residual nT-rt_uTo assign the rescue aircraft u to the appropriate location.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.

Fig. 1 is an architecture diagram of a rescue aircraft mission planning system provided in an embodiment of the present invention

FIG. 2 is a frame diagram of an earthquake rescue simulator provided in the embodiment of the present invention

FIG. 3 is a graph illustrating survival rate variation according to an embodiment of the present invention;

FIG. 4 is a rescue schedule for a U-rack rescue plane for a common rescue according to an embodiment of the present invention;

fig. 5 is a block diagram of a TLP rescue algorithm and an example of on-line task allocation for a rescue aircraft provided by an embodiment of the invention.

Fig. 6 is a flowchart of the operation of a rescue aircraft-level plan for a TLP rescue algorithm in the nth rescue cycle according to the embodiment of the invention.

Fig. 7 is an example of four random operators provided in the embodiment of the present invention.

Detailed Description

The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The technical scheme of the invention is specifically explained according to the attached drawings.

Referring to fig. 1, the earthquake rescue simulator first generates an earthquake-stricken area map according to a population density map and an earthquake intensity map, and determines the position distribution of a ground medical area and a safety area. According to the actual seismic data, disaster-stricken personnel data of each ground medical area can be generated in each period. Meanwhile, in each rescue cycle, the number and the positions of the severe disaster areas and the data of the disaster victims in the severe disaster areas can be randomly generated, and the positions of the severe disaster areas in the map meet the distance threshold constraint with the ground medical area. Factors influencing the number of people suffering from a disaster mainly include survival rate and rescue of rescue airplanes, and under the action of the two factors, the number of people suffering from a disaster and the number of people successfully rescued in each rescue period can be obtained. Fig. 2 shows the survival rate curves for 15 min ST, 0.7 EI and 10 min ST, 0.5 EI. The smaller the EI, the more injury the person suffering from the disaster is, and the more obvious the survival rate of the person suffering from the disaster is reduced along with the time. Different statistical periods ST and influence parameters EI can be set according to actual conditions and differences of the ground medical area and the heavily stricken area.

An example rescue schedule for a plurality of rescue planes is given with reference to figure 3,firstly, a rescue task is generated according to the conditions of the disaster-stricken personnel in the ground medical area and the severe disaster-stricken area in the nth period and the position of the rescue airplane. Allocating rescue target sequences TS for all rescue airplanes according to TLP rescue algorithm_u＝[T_u,1,T_u,2,…,T_u,i,…,T_u,nu]. And then, the rescue aircrafts complete rescue tasks in sequence according to the sequence of the respective rescue target sequences. At the beginning, all rescue airplanes fly to the first rescue target (namely, the rescue airplane 1 flies to T_1,1Rescue aircraft 2 flying direction T_2,1And so on) to calculate the time t when they reach the corresponding target_u+t^u _pTiAnd t_u+t^u _pTiAnd the remaining number of people suffering from the disaster in the ground medical area or the severe disaster area corresponding to the rescue target at any moment. Then, the statistics are according to t_u+t^u _pTiAnd sequentially completing rescue tasks by the rescue airplanes from small to large, and calculating the rescue success number of each rescue airplane at the corresponding rescue target. And then updating the information of each rescue task and the rescue aircraft, and repeatedly executing the steps until the rescue tasks of the U-frame rescue aircraft are completely finished or are terminated overtime. And finally, the positions of the rescue airplanes are distributed by using a rescue airplane position distribution method, so that the rescue work of the (n + 1) th period is facilitated.

Referring to fig. 4, an example of the structure of the TLP algorithm and the task allocation of the rescue aircraft is given, and the TLP mainly consists of three parts, namely rescue task level planning, rescue aircraft level planning and rescue aircraft position allocation.

And (3) rescue task level planning:

(1) clustering the rescue tasks by adopting a K-means clustering method to generate K_bestAnd (4) determining a rescue task group and determining a rescue center. Set the value of K to [2, K_max]Any random integer of (a), wherein K_maxClustering each possible K value from small to large for the minimum value of the total number U of the rescue aircrafts and the number of rescue tasks in the nth rescue period, and calculating the rescue center m of each K value_i(i ═ 1,2, …, K), and the sum of the squares of errors (SSE) for each value of K is calculated according to the formula,

wherein C is_iIs of the ith class, p is C_iCoordinate value of (1), m_iIs C_iThe center of mass of the rescue center.

Finally, selecting the optimal clustering number K by using an elbow method_best。

(2) According to the proportion of the total number of the people suffered from the disaster and the distance between each rescue aircraft and the rescue center, K is generated_bestA rescue aircraft group, wherein the rescue aircraft group and the rescue task group are in one-to-one correspondence

Rescue plane level planning:

the rescue airplane level is used for sequentially finishing the planning of the rescue task sequence of each rescue airplane in each rescue airplane set. Fig. 5 shows a flow chart of an algorithm for rescue airplane-level planning.

(1) And inputting information of the task groups and the unmanned aerial vehicle groups, and expanding each rescue task to a rescue target according to the maximum load L of the unmanned aerial vehicle for each rescue task group and the corresponding rescue aircraft group.

(2) And a two-part coding method is adopted to distribute the rescue target to the rescue aircraft. Initializing a population pos (t), wherein the number of the population is P. Fitness was calculated at each round and gBest was recorded.

Four random operators are adopted to carry out deep search on the neighbors of all individuals in the population, including: two-point exchange, single-point insertion, section exchange and section reverse order.

The fitness function is: f (X) max (β × RN (X, P)^t)-(1-β)×FN(X,P^t))

Wherein RN (X, P)^t) And FN (X, P)^t) The total number of successful rescue passengers and the total number of failed rescue passengers of all rescue airplanes in the nth period are weighted by the weight coefficient beta belonging to [0,1 ]]The method can be used for adjusting the proportional relation between the total number of people who succeed in rescue and the total number of people who fail in rescue. Determining a rescue target sequence TS to be completed for each rescue aircraft in each rescue aircraft group_uAnd rescue end time rt_uAnd respectively using X and RT to assemble the first rescue planeOptimal allocation and rescue termination time, X ═ TS₁,TS₂,…,TS_u,…,TS_U]，RT＝[rt₁,rt₂,…,rt_u,…,rt_U]。

And (3) rescue aircraft position allocation:

in the nth rescue period, there is a rescue termination time rt for rescuing the aircraft u_uIn the case that the time is shorter than the end time nT of the nth period, the method for allocating the positions of the rescue aircrafts is used to fully utilize the residual nT-rt_uTo assign the rescue aircraft u to the appropriate location.

Clustering rescue tasks by using a rescue task level planning method to generate K_best' rescue mission group and rescue aircraft group, and determine the rescue center. And updating the position coordinates of the rescue aircraft according to the position of the rescue center and the position of the rescue aircraft u.

A schematic diagram of the four random operators is given with reference to fig. 6. And comprehensively considering two parts of the target and the breakpoint when generating a new individual. The specific random operator is defined as follows:

(1) two-point exchange: randomly selecting two different positions r of a target portion₁、r₂Exchange r of₁、r₂。

(2) Single-point insertion: randomly selecting two different positions r of a target portion₁、r₂If r is₁<r₂Then r will be₂Is inserted into r₁Before, otherwise r₂Is inserted into r₁And then.

(3) Section exchange: randomly selecting four positions r of the target portion₁、r₂、r₃、r₄And r is₁≤r₂<r₃≤r₄Exchange section r₁-r₂And section r₃-r₄The position of (a). If r₁＝r₂And r is₃＝r₄Then the segment exchange is equivalent to a two-point exchange.

(4) Segment reverse order: randomly selecting two different positions r of a target portion₁、r₂Section r of reverse order₁-r₂All positions in (1) are r₂-r₁。

Claims

1. The rescue airplane task allocation method based on the two-stage planning algorithm in earthquake rescue is characterized by comprising an earthquake rescue simulator and the two-stage planning rescue algorithm, wherein the two-stage planning is abbreviated as TLP, the earthquake rescue simulator simulates the conditions of a disaster area and disaster-stricken persons in earthquake disasters, and simultaneously simulates the change of the disaster-stricken persons under the influence of survival rate and various rescue factors of the rescue airplane; the method mainly comprises the following steps:

simulating a map of the earthquake-stricken area, namely simulating a disaster-stricken area by using a grid map related to the size of an actual map, wherein each grid has population density attributes, earthquake intensity conditions, the number of people suffering from a disaster and map attributes of the generation time of people suffering from the disaster, and the attributes of the map grids and the positions of a ground medical area and a safety area are generated according to the population density map and the earthquake intensity map;

generating distribution simulation of disaster-stricken persons, wherein when a grid representing a disaster-stricken area in a map is in a disaster, a certain number of disaster-stricken persons are generated by surrounding grids, the disaster-stricken persons can be periodically generated according to a population density map and an earthquake intensity map, the number of the disaster-stricken persons and the distribution condition are changed along with time under the influence of the survival rate, and the disaster-stricken persons in a ground medical area and the disaster-stricken persons in a disaster-stricken area are generated according to a distance threshold value Rth and the positions of new disaster-stri;

simulating the survival rate of the disaster-stricken person, wherein the survival rate is also called survival rate, and the simulation of casualty conditions of the disaster-stricken person along with time is described as follows: survival rate (number of m ST survivors/total number at the beginning) × 100%, where ST is the statistical period of survival rate, the trend of gradually decreasing survival rate with increasing time, described using a piecewise function based on EI: SR (EI)^(m-1)(m is 1,2,3, …), namely the survival rate of the disaster-stricken in the mth ST is more than the EI of the m-1 ST, because m is 1,2,3, …, the SR is in a step-shaped segmented form, the smaller the EI is, the larger the injury to the disaster-stricken is, and the more obvious the survival rate of the disaster-stricken is reduced along with the time;

rescue airplane rescue, which is divided into single airplane rescue and multiple airplanesRescue scheduling, rescue of a single rescue airplane: calculating the number of remaining disaster suffered by rescue aircraft u when the rescue aircraft u reaches a rescue target

Wherein AN^t(x, y) represents the number of persons in disaster at time (x, y) t, t^u _pTiIndicating rescue aircraft u from current position p_u ^tFlying to target position T_u,iTime of (t + t)^u _pTiIndicating the time at which the aircraft reaches the rescue objective, if the load of u

Is insufficient, returns to the safe area nearby, and when the completion is possible, calculates the heading of the rescue aircraft u to the rescue target (x)_i,y_i) Number of people who succeed in post-rescue N_u,iU is required to satisfy the constraint of cycle time in the rescue process;

rescue scheduling of a plurality of airplanes: firstly, according to the disaster situations of the ground medical area and the severe disaster area in the current period and the position information of the rescue aircrafts, a task-level planning algorithm of a TLP (transient liquid phase) rescue algorithm is used for distributing a rescue target sequence TS (transport stream) for each rescue aircraft_u＝[T_u,1,T_u,2,…,T_u,i,…,T_u,nu]Then, each rescue aircraft completes rescue tasks in sequence according to the sequence of the rescue target sequence

The TLP rescue algorithm is used for optimizing task allocation of the rescue aircraft to maximize the proportion of the number of successful rescue people, and the TLP mainly comprises the following steps:

the rescue task level planning completes the optimal clustering of task groups and the distribution of rescue airplane groups corresponding to the task groups according to the rescue tasks and the distribution conditions of rescue airplanes in the current rescue period, and mainly comprises the following steps: (1) clustering rescue tasks by using the grid coordinates of each rescue task as a clustering basis and adopting a K-means clustering method, and determining a clustering center of each task group; (2) the rescue aircraft set is divided according to the proportion of the disaster-suffered total number of people of each task group and the distance from each rescue aircraft to the rescue center;

the rescue aircraft level planning is realized, the planning of the rescue task sequence of each rescue aircraft in each rescue aircraft set is sequentially finished, in the nth rescue period, the task planning in the rescue aircraft sets is optimized based on a particle swarm algorithm, in a coding mode, the rescue tasks are expanded and numbered according to the maximum load L of the rescue aircraft, the rescue targets are obtained, the rescue targets are distributed to the rescue aircraft by adopting a method of coding a target part and a breakpoint part, wherein the coding length of the target part is the number of the rescue targets, the rescue targets are arranged by adopting a symbol coding method, the coding length of the breakpoint part is the number-1 of the rescue aircraft, the breakpoint value represents the division condition of the target part, each divided section of targets is the rescue target distributed to each rescue aircraft, the population is randomly initialized, and four random operators are designed to carry out depth search on the neighbors of all individuals in the population, four random operators are defined:

(1) two-point exchange: randomly selecting two different positions r of a target portion₁、r₂Exchange r of₁、r₂；

(2) Single-point insertion: randomly selecting two different positions r of a target portion₁、r₂If r is₁<r₂Then r will be₂Is inserted into r₁Before, otherwise r₂Is inserted into r₁Then;

(3) section exchange: randomly selecting four positions r of the target portion₁、r₂、r₃、r₄And r is₁≤r₂<r₃≤r₄Exchange section r₁-r₂And section r₃-r₄If r is₁＝r₂And r is₃＝r₄Then segment exchange is equivalent to two-point exchange;

(4) segment reverse order: randomly selecting two different positions r of a target portion₁、r₂Section r of reverse order₁-r₂All positions in (1) are r₂-r₁(ii) a The fitness function is:

F(X)＝max(β×RN(X,P^t)-(1-β)×FN(X,P^t))

RN(X,P^t) And FN (X, P)^t) The total number of successful rescue passengers and the total number of failed rescue passengers of all rescue airplanes in the nth period respectively, and the weight coefficient beta belongs to [0,1 ]]；

Rescue airplane position distribution in a rescue period T_nWithin, there is a rescue end time rt of the rescue aircraft u_uIf the time is shorter than the end time nT of the nth period, the rescue aircraft u is distributed to a proper position by using the residual time, and the rescue tasks are clustered by using a rescue task level planning method to generate K_bestThe rescue mission group and the rescue aircraft group determine a rescue center, and update the position coordinates of the rescue mission group and the rescue aircraft according to the position of the rescue center and the position of the rescue aircraft u.

2. The rescue airplane task allocation method based on the two-stage planning algorithm in earthquake rescue as claimed in claim 1, characterized in that the simulation of the earthquake-stricken area map uses a two-dimensional grid map simulation related to the size of the actual map, and each grid has population density attribute, earthquake intensity condition, number of people suffering from a disaster, generation time map attribute of people suffering from a disaster and corresponds to the population density map and the earthquake intensity map; dividing grids into three types of urban occupied areas, rural occupied areas and unmanned areas according to the distribution of population density from high to low, determining the position distribution of ground medical areas and safety areas according to population density graphs and earthquake intensity graphs, and then enabling the distances between the grids and the ground medical areas to be larger than a threshold value R in each rescue cycle_thThe grid in disaster can become a heavily disaster area.

3. The rescue aircraft task allocation method based on the two-stage planning algorithm in earthquake rescue according to claim 1, characterized by simulating the generation and distribution of disaster-stricken personnel;

according to the population density chart, the actual casualties RAN, the simulation times SN, the set MA of the ground medical areas and the number n of the ground medical areas_maCalculating the initial number of people suffering from a disaster A N (MAi) in each ground medical area in each period, and firstly generating an SN row n_maNormalized matrix of columns, M, requires:

and is

adjusting the initial number of people suffering from a disaster in each regional medical area according to the population density map;

according to the set MA of the ground medical treatment area and the distance threshold value R_thNumber threshold value N of severe disaster areas_thDisaster-stricken number threshold AN of severe disaster-stricken area_thThe set SAA of heavily affected areas is calculated, and is described by means of an EI-based segmentation function: SR (EI)^(m-1)(m is 1,2,3, …), namely the survival rate of the disaster-stricken in the mth ST is more than the EI of the m-1 ST, because m is 1,2,3, …, the SR is in a step-shaped segmented form, the smaller the EI is, the larger the injury to the disaster-stricken is, and the more obvious the survival rate of the disaster-stricken is reduced along with the time.

4. The rescue airplane task allocation method based on the two-stage planning algorithm in earthquake rescue according to claim 1, characterized by simulating rescue airplane rescue;

the single airplane continuously flies to the rescue target, flies to a safety area to place personnel when the residual capacity is insufficient, and updates the number N of successful rescue personnel_u,iAnd rescue end time rt_u；

When a plurality of airplanes are rescued, firstly, a rescue planning algorithm is used for distributing rescue target sequences TS for all rescue airplanes_u＝[T_u,1,T_u,2,…,T_u,i,…,T_u,nu]The rescue airplanes are sequentially in the sequence of the respective rescue target sequencesAnd finishing the rescue task.

5. The rescue aircraft task allocation method based on the two-stage planning algorithm in earthquake rescue as claimed in claim 1, characterized in that a K-means clustering method is adopted to cluster rescue tasks for rescue task level planning, and the K value in the K-means clustering is set as [2, K ] K_max]Random integer of (a), wherein K_maxA rescue center m for calculating each K value for the minimum value of the total number U of the rescue aircrafts and the number of rescue tasks in the nth rescue period_i(i ═ 1,2, …, K), and the sum of the squares of errors (SSE) for each value of K is recorded,

wherein, C_iIs of the ith class, p is C_iCoordinate value of (1), m_iIs C_iThe center of mass is the rescue center, and finally, an elbow method is used for selecting the K value corresponding to the elbow as the optimal clustering number K_bestAnd is combined with K_bestSetting the output result after clustering as the optimal clustering scheme of the rescue tasks, and then generating K according to the proportion of the total number of the disaster suffered people of each task group and the distance from each rescue aircraft to the rescue center_bestThe rescue aircraft groups correspond to the task groups one by one.

6. The rescue airplane task allocation method based on the two-stage planning algorithm in earthquake rescue as claimed in claim 1, wherein a two-stage coding method is adopted in TLP rescue airplane stage planning, and a rescue factor α ═ initial disaster receiver/L is firstly set_max]Wherein L is_maxAlpha represents the number of rescue airplanes needed under the condition of no casualties, each rescue task is expanded into alpha rescue targets according to alpha, the rescue targets are numbered sequentially, and the maximum number of the persons to be rescued in each rescue target is L_max；

The rescue target is distributed to the rescue airplane by adopting a two-part coding method, the first part is a target part, the coding length is the number of the rescue targets, and the rescue targets are arranged by adopting symbol coding; the second part is a breakpoint part, the coding length is the number of rescue airplanes-1, each section of target divided by the breakpoint value is a rescue target distributed to each rescue airplane, and the setting of the individual breakpoints meets the constraint condition.

7. The rescue aircraft task allocation method based on the two-stage planning algorithm in earthquake rescue according to claim 1, characterized in that the fitness function describing the earthquake rescue process is as follows:

F(X)＝max(β×RN(X,P^t)-(1-β)×FN(X,P^t))

wherein the weight coefficient beta is ∈ [0,1 ]]，RN(X,P^t) And FN (X, P)^t) The total number of people for successful rescue and the total number of people for failed rescue of all rescue airplanes in the nth period are respectively;

constraint conditions are as follows:

wherein AN (x, y) is the initial distribution of disaster-stricken people, AN^t(x, y) is the distribution of the disaster-stricken personnel at the end of the rescue period, and U is the total number of rescue airplanes.

8. The rescue airplane task allocation method based on the two-stage planning algorithm in earthquake rescue as claimed in claim 1, wherein after rescue is completed, the rescue airplane position allocation scheme is adopted, and in the nth rescue period, the rescue termination time rt of the rescue airplane u exists_uShorter than the end of the nth periodIn the case of nT, a rescue aircraft position distribution method is required to be used, and the residual nT-rt is fully utilized_uThe rescue aircraft u is distributed to a proper position, the rescue tasks are clustered by using a rescue task level planning method, and K is generated_best' a rescue task group and a rescue aircraft group, and determining a rescue center; each rescue aircraft flies towards the rescue center and calculates the final position P^nT。