Multi-unmanned aerial vehicle charging time equalization method and system for wireless charging sensor network, storage medium and electronic device

Abstract

The application discloses a kind of multi-unmanned aerial vehicle charging time equalization methods, systems, storage medium and electronic equipment for wireless charging sensor network, method includes the following steps: S1, inside two-dimensional plane deployment one wireless chargeable sensor network WRSN, by three parts: sensor node, base station, unmanned aerial vehicle;S2, in the WRSN scene of step S1 deployment, by planning the flight trajectory of multiple unmanned aerial vehicles, the time of each unmanned aerial vehicle to complete charging task is balanced, and is modeled as the minimum problem of maximum completion charging task time;S3, solve the problem in step S2, realize the equalization of unmanned aerial vehicle charging time.The application solves the path planning problem of existing multi-unmanned aerial vehicle (UAV) in WRSN to charge node, and realizes the relative balance of each UAV charging task time.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless rechargeable sensor networks, and specifically proposes a multi-unmanned aerial vehicle charging time balancing method, system, storage medium and electronic device for wireless rechargeable sensor networks. BACKGROUND

[0002] With the rapid development of wireless technology, wireless sensor networks (WSN) play a crucial role in real life, especially in the field of sensing and monitoring, such as environmental perception, target tracking and health monitoring. Traditional wireless sensor networks are usually powered by batteries, but limited by battery capacity, energy shortage not only limits the large-scale deployment of sensors, but also adversely affects the service life of the entire network.

[0003] With the continuous progress of wireless power transmission (WPT) technology, a new network mode - wireless rechargeable sensor networks (WRSN) has emerged. In WRSN, sensor nodes are equipped with wireless energy transceiver devices, enabling wireless charging. Early WRSN charging methods mainly rely on wireless charging vehicles (WCV) to power nodes. WCV moves in the network according to the preset path and charges the sensor nodes by staying at a specific location. This method is suitable for sensor nodes distributed in cities or areas accessible by vehicles, and thus has higher requirements for road conditions. However, for WRSN deployed in wild environments, the applicability of this method is very low. The reason is that in complex terrain such as mountains and forests, most sensor nodes are distributed in areas that are difficult to access, and cannot provide good road conditions. The traditional charging method is difficult to meet the actual demand.

[0004] In recent years, unmanned aerial vehicles (UAV) have gradually become a research hotspot for solving the charging problem of WRSN nodes due to their high mobility, flexibility, rapid deployment, ability to adapt to diverse environments, and lower cost. Using unmanned aerial vehicles to carry large-capacity charging modules to provide energy supply for WSN nodes shows great potential. In order to effectively solve the node charging problem in WRSN, researchers have proposed various charging schemes based on unmanned aerial vehicles, which can be mainly divided into two categories: single unmanned aerial vehicle charging and multi-unmanned aerial vehicle cooperative charging. The single unmanned aerial vehicle scheme is limited by its own power capacity and is difficult to meet the charging demand of large-scale WRSN, and is only suitable for small-scale area charging task planning. For large areas, multi-unmanned aerial vehicle cooperative charging is usually used to complete the charging task.

[0005] In the multi-UAV cooperative charging mode, it is necessary to reasonably allocate the charging task area for each UAV. However, there may be a problem of time imbalance in task allocation, which will cause some UAVs to run out of power due to too long task time, and cannot complete the charging task, and even may have the risk of crashing. Therefore, how to optimize the charging path planning of multi-UAV to ensure that the task completion time of each UAV is relatively balanced has become a key problem in designing the charging strategy of WRSN node. SUMMARY

[0006] To solve the path planning technical problem of multi-UAV in WRSN node charging, the present application provides a multi-UAV charging time balancing method, system, storage medium and electronic equipment for a wireless rechargeable sensor network WRSN. The present application aims to realize the relative balance of the charging task time of each UAV. The UAV charging task time is composed of flight time and charging time, and the UAV hovers above the node to charge the node.

[0007] The present application adopts the following technical solutions:

[0008] The multi-UAV charging time balancing method for a wireless rechargeable sensor network includes the following steps:

[0009] S1, deploying a wireless rechargeable sensor network WRSN in a two-dimensional plane, the WRSN including sensor nodes, a base station and UAVs;

[0010] S2, in the WRSN scene deployed in step S1, the flight trajectory of the multi-UAV is planned to balance the time of each UAV to complete the charging task, and the problem is modeled as a minimum problem of maximum charging task completion time;

[0011] S3, solving the problem in step S2 to balance the charging time of the UAV.

[0012] Preferably, in step S1, the WRSN scene is modeled as follows:

[0013] A WRSN is deployed in any two-dimensional plane, which is composed of three parts: sensor nodes S, a base station CBS and UAVs. The two-dimensional plane area of the WRSN is represented as G(S, CBS, UAVs). Each part is introduced as follows:

[0014] Base station: The base station CBS is deployed at any position in the two-dimensional plane, and its coordinates are (x0, y0). As an energy supply source, the base station is not limited by energy and can provide charging support for the UAVs performing tasks, while monitoring the position information and remaining power state of the sensor nodes in real time.

[0015] Sensor nodes: There are n sensor nodes in total, forming a set S = {s1, s2, ..., sn}. n These nodes are evenly distributed in a two-dimensional plane. Their main function is to monitor and upload surrounding environmental data in real time, and to achieve wireless charging through their own wireless charging receivers. Each sensor node s i Let i represent the node with the number i, where i∈[1,n],i∈Z. The set of electrical charges of the nodes is represented as E={E1,E2,...,E...} n}, where E i Let E represent the current power level of the i-th node, and let E represent the maximum power capacity of the sensor node. max Data is transmitted between nodes via multi-hop communication. Since each node consumes power differently, the required charging energy E varies. max -E is also different.

[0016] Drones: There are J drones in the network. Each drone departs from the base station carrying sufficient energy and hovers above the sensor nodes to charge them in a one-to-one manner (meaning that when a single drone is performing a charging task, it only charges one node). Once a node has finished charging, the drone departs at the optimal flight speed v. * Proceed to the next node to continue charging. Repeat this process until all sensor nodes requiring charging have been charged, then return to the base station. When the UAV performs the charging task, the default flight altitude is fixed at H (in meters), and the acceleration and deceleration processes before and after the UAV charges the nodes are not considered.

[0017] Preferably, step S2, problem planning and optimization problem construction, is as follows:

[0018] In the WRSN scenario shown in step S1, the problem to be solved is: how to plan the flight trajectories of multiple drones so that the time for each drone to complete the charging task is relatively balanced. This is modeled as a problem of minimizing the maximum charging task completion time, where the charging task completion time consists of flight time and charging time. Since each drone has the same task nature, a mathematical model of the charging task completion time of a single drone is performed as an example.

[0019] (1) Power transfer model for charging task

[0020] Wireless charging between the drone and sensor nodes is primarily achieved through line-of-sight (LoS) transmission (an existing technology where wireless signals propagate in a straight line between the transmitter and receiver). According to the free-space path loss model, the link transmission loss P... w Represented as:

[0021] Pw = 20 log(f) + 20 log(d) - 147.55 dB

[0022] where d represents the transmission distance from the sensor node to the UAV, in meters; f is the charging frequency of the UAV, in Hz. Since the UAV is charging directly above the node, d = H, in meters.

[0023] Let the UAV transmit with a constant transmit power P t The energy is transferred. At each sensor node, the received radio frequency signal is converted to a direct current signal and energy is collected through a rectifier, then the received radio frequency power P r can be expressed as:

[0024] P r = P t + G UAV + G SN - P w

[0025] where G UAV represents the transmit antenna gain of the UAV, G SN represents the receive antenna gain of the sensor, P w is the path loss of the power.

[0026] Assuming the energy conversion efficiency at the sensor node is η, then the power obtained by the receiving end (sensor node) after conversion through the rectifier is

[0027] (2) Charging task time model

[0028] The total time for the UAV to complete the charging task is composed of the flight time and the charging time (i.e., hovering time). The following describes these two parts of time in formulas:

[0029] If a UAV has m charging nodes in the charging area, and is less than the total number of nodes n. To get the total flight time of the UAV, the time of each segment of the UAV flight path in the task area needs to be calculated. The flight time from the i-1th node to the ith node is :

[0030]

[0031] where l i represents the flight distance from the i-1th node to the ith node, in meters. l1 represents the flight distance of the UAV from the charging base station to the first node, l m+1 represents the flight distance of the UAV from the last node after completing charging to return to the charging base station; the flight distance v* represents the flight speed of the UAV in the most energy-saving state, which can minimize the energy consumption of the UAV in the uncharged state. Therefore, the total flight time of the UAV is:

[0032]

[0033] Since the initial energy of each node requiring charging is different, the time required by the UAV to charge each node also varies. The amount of charge required to supplement each node can be represented as:

[0034]

[0035] During wireless charging, the actual power converted by the node is Thus, the charging time of the UAV at the i-th node is:

[0036]

[0037] The total charging time T of the UAV in the task area is h :

[0038]

[0039] After obtaining the total flight time and charging time, the total time T of the UAV to complete the charging task is UAV :

[0040] T UAV = T f + T h

[0041] Assuming there are j UAVs, to achieve the balanced charging task time of each UAV, the present application models this problem as a min-max problem, i.e., a problem of minimizing the maximum charging task time of the UAV, as follows:

[0042]

[0043] Preferably, step S3 is based on the problem solving of the Crown-Hoar optimization algorithm; specifically, the Crown-Hoar optimization algorithm (CPO) is used to solve the min-max problem in step S2, and the detailed steps are as follows:

[0044] S3.1 initialize the population and randomly generate a certain number of individuals

[0045] First, determine the population size N, the maximum number of iterations T max , the lower boundary L of the solution space, the upper boundary U of the solution space, and the dimension dim of the solution. CPO initializes the process from the initial individual set:

[0046] X k= L + r x (U - L) k = 1, 2,..., N

[0047] where N represents the population size, also represented as N solutions, X k is the kth candidate solution in the search space, each candidate solution represents a multi-UAV task node allocation manner, L and U are the upper and lower limits of the search range, respectively, and r is a random number between 0 and 1. After initializing all N individuals in the population using the above formula, the initialized population is obtained as:

[0048]

[0049] S3.2 Initialization of candidate solution matrix

[0050] In the initialization phase of the optimization algorithm, each row in the matrix X represents an individual (candidate solution), and each individual corresponds to a complete path planning scheme. The column number represents the label of the node. Assuming there are n nodes in total, the first column represents the label of the first node in all populations, the second column represents the label of the second node in all populations, and the nth column represents the label of the nth node in all populations.

[0051] S3.3 Sorting candidate solutions

[0052] Before calculating the fitness function, each candidate solution needs to be sorted for subsequent task allocation. Each node in each candidate solution (each row of the matrix) is sorted in ascending order according to its label.

[0053] S3.4 Splitting tasks

[0054] After sorting, each candidate solution is split. The goal of splitting is to evenly distribute the nodes in each solution to the J UAVs. Let the quotient of n / J be a and the remainder be b, then the first b UAVs are allocated a+1 nodes, and the last J-b UAVs are allocated a nodes to ensure that the number of nodes served by each UAV is as balanced as possible.

[0055] S3.5 Adding start and end coordinates

[0056] In each UAV task, the start coordinate and end coordinate, i.e. the CBS coordinate (x0, y0), are added. These coordinates are the starting point and ending point of the path planning and are also considered when calculating the task time. The start and end coordinates mark the starting position and ending position of each UAV task, making the path calculation more accurate.

[0057] S3.6 Calculate the total task time of each UAV task

[0058] According to the task sequence after splitting, the total task time of each UAV from the starting point to the end point is calculated by a path calculation formula (such as the formula described in S2).

[0059] S3.7 Calculate the fitness of each individual

[0060] Then the fitness of each individual in the population is calculated according to the above method, which is the maximum task time of the UAV in the individual, and the individual with the lowest fitness is taken as the current global optimal solution. The specific fitness function is:

[0061] Fitness(X i )=1 / max{T j},j∈[1,J]

[0062] Where T j represents the charging task time of the jth UAV, and the goal of the fitness function is to minimize the maximum task time by selecting the individual with the minimum maximum UAV task time in the population. The fitness function of each individual is calculated to select the individual X i with the highest fitness in the population as the global optimal solution.

[0063] S3.8 Adjust the position of the individual using the defense mechanism, and update the individual according to the fitness.

[0064] In order to find better solutions, the solution space is then searched and the position of each individual X i in the population is updated. The specific search process can be divided into two parts: global exploration and local development. The size of two random numbers τ9 and τ8 is generated randomly to determine whether to perform global search or local search. Each individual will select one of the four defense strategies to change. The four defense strategies are as follows:

[0065] S3.8.1 First defense strategy

[0066] Global exploration is performed when τ8 < τ9, and global exploration can be performed in two ways, namely the first defense strategy and the second defense strategy, which can be determined by comparing the size of two randomly generated random numbers. When τ6 < τ7, the first defense strategy is executed, and its mathematical model is:

[0067]

[0068] Where r is a random number between 1 and N, is the current total group of other random individuals. τ1 is a random number following a normal distribution; τ2 is a random number between 0 and 1; is the current global optimal solution; the current individual ​will be updated according to the global optimal solution and the reference position , the random factors τ1 and τ2 are introduced to guide the individual in the search process and to explore a certain range to avoid premature convergence. After an individual is adjusted according to the first defense strategy, the fitness of and is calculated. If the fitness of is greater than the fitness of , the modified individual is replaced in the population X, otherwise, no replacement is made. Then the fitness of and the global optimal solution is compared. If it is greater, then is replaced by which is called the new global optimal solution.

[0069] S3.8.2 Second defense strategy

[0070] When τ6>τ7, the second defense strategy is executed. The mathematical model is as follows:

[0071]

[0072] wherein, U1 is a randomly generated 1-row n-column vector, which only contains 0 or 1. Therefore, the calculation of the first half will keep the column node corresponding to the value of (1-U1) being 1. and are two random individuals in the population X, and τ3 is a random number in the interval of 0 to 1. This updating strategy combines local information and global information. The next position of the individual will be affected by the current position and the difference between the two random individuals. After an individual is adjusted according to the second defense strategy, the fitness of and is calculated. If the fitness of is greater than the fitness of , the modified individual is replaced in the population X, otherwise, no replacement is made. Then the fitness of and the global optimal solution is compared. If it is greater, then is replaced by which is called the new global optimal solution.

[0073] S3.8.3 Third defense strategy

[0074] When τ8>τ9, local exploration is executed. Local search can also adopt two methods, which are the third defense strategy and the fourth defense strategy. Specifically, two randomly generated random numbers τ 10With T f The value is determined by the magnitude of τ. 10 <T f When the third defense strategy is implemented, its mathematical model is as follows:

[0075]

[0076] Where r3 is a random number between 1 and N; t is the iteration number, which is incremented by 1 after each individual in the population has traversed the four defense strategies once; T max τ is the maximum number of iterations; τ3 is a random number between 0 and 1; and Let X be three random individuals from population X. The odor diffusion factor is calculated as follows:

[0077]

[0078] in, Let be the fitness of the i individuals at the t-th iteration, and ε be an arbitrarily small positive number. From the above equation, it can be seen that when... The larger, the more it means The better the fitness, the more inclined the individual will be to move closer to the optimal solution. The formula for updating the third defense mechanism combines the current individual position, global information, fitness information, and random factors. After adjustment using the above method, proceed to steps S3.3-S3.7. and Perform fitness calculations if The fitness is greater than If the fitness of the modified individual is high, it will be replaced in population X; otherwise, no replacement will be made. Then compare... and the global optimal solution If the fitness is greater than 1, then... Replace with This is called the new global optimal solution.

[0079] S3.8.4 Fourth Defense Strategy

[0080] When τ 10 >T f At that time, the fourth defense strategy is implemented, and the mathematical model is as follows:

[0081]

[0082] Where α is the velocity convergence factor set to 0.2; τ4 and τ5 are random numbers between 0 and 1; F i t It is the inelastic collision force generated when an individual attacks a predator with its body, and it is calculated as follows:

[0083]

[0084]

[0085] wherein, denotes is the other random individual of the current total population, is the current individual m i is the quality of the ith individual at iteration t. After adjustment by the above method, the quality of and is calculated, if the quality of is greater than the quality of , the modified individual is replaced into the population X, otherwise, it is not replaced. Then the quality of and the global optimal solution is compared, if it is greater, then is replaced by which is called the new global optimal solution.

[0086] When all individuals of the original population have been traversed through the four defense mechanisms, since each individual has been changed and the new individual has a better quality than the original one, the new individual will be saved into the population, thus a new population is generated, and the population size is still N. In order to accelerate the convergence speed, the cyclic population reduction technique is also adopted, and the population size is periodically changed with the increase of the iteration number t.

[0087] S3.9 Cyclic population reduction technique

[0088] In order to improve the convergence speed, the CPO adopts the cyclic population reduction strategy. After the tth iteration is completed, the population size is correspondingly reduced in order to accelerate the convergence speed of the population; when the population size is reduced to a certain value, the population size is increased again in order to increase the diversity of the population, and the cycle is repeated until the maximum iteration number is reached. In this way, the algorithm avoids the waste of computing resources, and at the same time accelerates the optimization process. Therefore, some individuals are obtained from the population during the optimization process, the selection of the individuals is related to the quality in order to accelerate the convergence speed, and they are reintroduced into the population, thereby improving the diversity and avoiding falling into a local minimum; the cycle is based on a loop variable to determine the number of times the process is executed during the optimization process, and the mathematical model of the cyclic population size reduction is as follows:

[0089]

[0090] wherein, T is a variable for determining the number of cycles, t is the current iteration number, T max is the maximum iteration number, N minN is the minimum number of individuals in the newly generated population, so the population size cannot be less than N min When the population is reduced, the population needs to be reduced according to the fitness of each individual calculated in the foregoing, the fitness is sorted, the individuals with good fitness are retained, and the individuals with poor fitness are deleted.

[0091] S3.10 gradually converges to the optimal solution by guiding the search process through the globally optimal position

[0092] According to steps S3.8-S3.9, with the increase of the number of iterations, new individuals are constantly searched through the four defense mechanisms, and the corresponding fitness values are calculated, and the best individuals are selected to be retained, and the process is repeated. When the population reaches the maximum number of iterations, the optimal individual saved in the algorithm is returned. The individual saves the node data of the task allocation area of each unmanned aerial vehicle, and the minimum fitness value represents the maximum task time in this case, so as to achieve the goal of balancing the charging time.

[0093] The application also discloses a multi-unmanned aerial vehicle charging time balancing system for a wireless charging sensor network, which is used for executing the method and comprises the following modules.

[0094] A wireless chargeable sensor network scene modeling module: deploying a wireless chargeable sensor network WRSN in a two-dimensional plane, which is composed of three parts: sensor nodes, base stations and unmanned aerial vehicles;

[0095] A problem construction module: in the deployed WRSN scene, the flight trajectory of the multiple unmanned aerial vehicles is planned to balance the charging task completion time of each unmanned aerial vehicle, and the problem is modeled as a maximum charging task completion time minimization problem;

[0096] A problem solving module: solving the problem in the problem construction module to balance the charging time of the unmanned aerial vehicles.

[0097] The application also discloses a storage medium storing computer instructions, and the computer instructions are used for causing a computer to execute the method or the system.

[0098] The application also discloses an electronic device, characterized in that the electronic device comprises:

[0099] A processor;

[0100] A memory for storing a program, when the program is called and executed by the processor, the program causes the processor to execute the method or the system.

[0101] The application solves the path planning problem of the existing multi-unmanned aerial vehicle (UAV) in the WRSN node charging, and realizes the relative balance of the charging task time of each UAV. BRIEF DESCRIPTION OF DRAWINGS

[0102] Figure 1 is a charging model diagram of a preferred embodiment of a multi-UAV wireless rechargeable sensor network.

[0103] Figure 2 is a complete node allocation schematic diagram of a preferred embodiment.

[0104] Figure 3 is a total flow chart of multi-UAV charging path planning based on the CPO algorithm of a preferred embodiment.

[0105] Figure 4 is a path planning diagram for four UAVs.

[0106] Figure 5 is a maximum task time iteration diagram.

[0107] Figure 6 is a total time diagram of the sum of flight time and charging time after path planning for each UAV.

[0108] Figure 7 is a multi-UAV charging time balancing system block diagram for a wireless rechargeable sensor network of a preferred embodiment. DETAILED DESCRIPTION

[0109] In order to describe the present application in more detail and facilitate the understanding of those skilled in the art, the present application will be further described below in conjunction with the accompanying drawings and examples. The examples in this part are used to explain the present application and are for the purpose of understanding, and do not limit the present application.

[0110] As shown in Figure 1 , a wireless rechargeable sensor network (WRSN) model is established, which is composed of a base station, multiple UAVs and sensor nodes. The UAVs start from the base station, fly to the sensor nodes with low battery after being fully charged, and use wireless charging technology to charge the nodes. When the nodes in the network are fully charged, the UAVs return to the base station.

[0111] Since the UAVs have limited power, their energy consumption includes flight energy consumption, hovering energy consumption and charging energy consumption, so it is necessary to reasonably plan the charging path to achieve the purpose of multi-UAV charging task time balancing strategy. As shown in Figure 3 , the present embodiment is a multi-UAV charging time balancing method for a wireless rechargeable sensor network, and the specific steps are as follows:

[0112] S1, WRSN scene modeling

[0113] As shown in Figure 1 , a WRSN is deployed in an arbitrary two-dimensional plane, which is composed of three parts: sensor nodes S, base station CBS, and UAVs UAVs. The two-dimensional plane area of the WRSN is represented as G(S, CBS, UAVs).

[0114] Base Station: A base station (CBS) is deployed at any location in a two-dimensional plane, with coordinates (x0, y0). As a power source, the base station is not limited by energy and can provide charging support for drones performing missions, while simultaneously monitoring the location information and remaining power status of sensor nodes in real time.

[0115] Sensor nodes: There are n sensor nodes in total, forming a set S = {s1, s2, ..., sn}. n These nodes are evenly distributed in a two-dimensional plane. Their main function is to monitor and upload surrounding environmental data in real time, and to achieve wireless charging through their own wireless charging receivers. Each sensor node s i Let i represent the node with the number i, where i∈[1,n],i∈Z. The set of electrical charges of the nodes is represented as E={E1,E2,...,E...} n}, where E i Let E represent the current power level of the i-th node, and let E represent the maximum power capacity of the sensor node. max Data is transmitted between nodes via multi-hop communication. Since each node consumes power differently, the required charging energy E varies. max -E is also different.

[0116] Drones: There are J drones in the network. Each drone departs from the base station carrying sufficient energy and hovers above the sensor nodes in a one-to-one manner to charge the nodes. Once a node has finished charging, the drone departs at the optimal flight speed v. * Proceed to the next node to continue charging. Repeat this process until all sensor nodes requiring charging have been charged, and finally return to the base station. When the UAV performs the charging task, the default flight altitude is fixed at H (in meters), and the acceleration and deceleration processes of the UAV before and after charging the nodes are not considered.

[0117] S2, Problem Planning and Optimization: Problem Formulation

[0118] In the WRSN scenario shown in step S1, the problem to be solved is: how to plan the flight trajectories of multiple drones so that the time for each drone to complete the charging task is relatively balanced. This problem can be modeled as a minimization problem of the maximum charging task completion time, where the charging task completion time consists of flight time and charging time. Since each drone has the same task nature, a mathematical model of the charging task completion time of a single drone is used as an example.

[0119] Power transfer model for charging tasks

[0120] Wireless charging between UAV and sensor nodes is mainly through Line-of-Sight (LoS) transmission. According to the free space path loss model, the transmission loss P w is expressed as:

[0121] P w = 20log(f) + 20log(d) - 147.55 dB

[0122] where d represents the transmission distance from the sensor node to the UAV, in m; f is the charging frequency of the UAV, in Hz. Since the UAV is charging directly above the node, d = H, in m.

[0123] Suppose the UAV transmits energy at a constant transmit power P t . At each sensor node, the received radio frequency signal is converted to a direct current signal and energy is collected through a rectifier, then the received radio frequency power P r of the sensor node can be expressed as:

[0124] P r = P t + G UAV + G SN - P w

[0125] where G UAV represents the transmit antenna gain of the UAV, G SN represents the receive antenna gain of the sensor, and P w is the path loss of the power.

[0126] Assuming that the energy conversion efficiency at the sensor node is η, then the power obtained by the receiving end (sensor node) after conversion through the rectifier is

[0127] Charging task time model

[0128] The total time for the UAV to complete the charging task is composed of the flight time and the charging time (i.e., hovering time). The following describes these two parts of time in formulas:

[0129] If a UAV has m charging nodes in the charging area, which is less than the total number of nodes n. To obtain the total flight time of the UAV, the time of each UAV flight path in the task area needs to be calculated. The flight time from the i-1th node to the ith node is :

[0130]

[0131] where l irepresents the flight distance from the i-1th node to the ith node, in meters. l1represents the flight distance of the UAV from the charging base station to the first node, and li represents the flight distance of the UAV from the i-1th node to the ith node. m+1 represents the flight distance of the UAV from the last node to the charging base station after completing charging; the flight distance of the UAV from the last node to the charging base station after completing charging is represented by li+1. v * represents the flight speed of the UAV in the most energy-saving state, which can minimize the flight energy consumption of the UAV in the uncharged state. Therefore, the total flight time of the UAV is:

[0132]

[0133] Since the initial energy of each node that needs to be charged is different, the time required by the UAV to charge each node also differs. The amount of charge required to supplement each node can be represented as:

[0134]

[0135] During wireless charging, the actual power converted by the node is Therefore, the charging time of the UAV at the ith node is:

[0136]

[0137] The total charging time T of the UAV in the task area is: h

[0138]

[0139] After obtaining the total flight time and charging time, the total time T of the UAV to complete the charging task is: UAV

[0140] T UAV = T f + T h

[0141] Assuming that there are j UAVs, to achieve the balance of the charging task time of each UAV, the present embodiment models this problem as a min-max problem, i.e., a problem of minimizing the maximum charging task time of the UAV, as follows:

[0142]

[0143] S3, problem solving based on the Crown-Hoar optimization algorithm

[0144] The present embodiment uses the Crown-Hoar optimization algorithm (CPO) to solve the min-max problem in S2. The specific steps are as follows:

[0145] S3.1 initialize the population and randomly generate a certain number of individuals ​​

[0146] Firstly, the population size N, the maximum number of iterations T, the lower bound L of the solution space, the upper bound U of the solution space, and the dimension dim of the solution are determined. max The CPO initializes the process from the initial individual set:

[0147] X k = L + r x (U - L) k = 1, 2,..., N

[0148] where N represents the population size, also represented as N solutions, X k is the kth candidate solution in the search space, each candidate solution represents a multi-UAV task node allocation manner, L and U are the upper and lower limits of the search range, respectively, and r is a random number between 0 and 1. After initializing all N individuals in the population using the above formula, the initialized population is obtained:

[0149]

[0150] S3.2 Initialization of candidate solution matrix

[0151] In the initialization phase of the optimization algorithm, each row in the generated matrix X represents an individual (candidate solution), and each individual corresponds to a complete path planning scheme. The column number represents the label of the node. Assuming there are n nodes in total, the first column represents the label of the first node in all populations, the second column represents the label of the second node in all populations, and the nth column represents the label of the nth node in all populations.

[0152] S3.3 Sorting candidate solutions

[0153] Before calculating the fitness function, each candidate solution needs to be sorted for subsequent task allocation. Each node in each candidate solution (each row of the matrix) is sorted in ascending order according to its label.

[0154] S3.4 Splitting tasks

[0155] After sorting, each candidate solution is split. The goal of splitting is to evenly distribute the nodes in each solution to the J UAVs. Let the quotient of n / J be a and the remainder be b, then the first b UAVs are allocated a+1 nodes, and the last J-b UAVs are allocated a nodes to ensure that the number of nodes served by each UAV is as balanced as possible.

[0156] S3.5 Adding start and end coordinates

[0157] In each UAV's task, the initial coordinate and the end coordinate, namely the CBS coordinate (x0, y0), are added. These coordinates are the starting point and the end point of the path planning, and are also considered when calculating the task time. The coordinates of the initial and end points mark the starting position and the ending position of each UAV's task, making the path calculation more accurate.

[0158] S3.6 Calculate the total task time of each UAV task

[0159] According to the split task sequence, the total task time of each UAV from the starting point to the end point is calculated by the path calculation formula (as described in S2).

[0160] S3.7 Calculate the fitness of each individual

[0161] Then calculate the fitness of each individual in the population according to the above method, which is the maximum task time of the UAV in the individual. The individual with the lowest fitness is taken as the current global optimal solution. The specific fitness function is:

[0162] Fitness(X i )=1 / max{T j},j∈[1,J]

[0163] Where T j represents the charging task time of the jth UAV. The goal of this fitness function is to minimize the maximum task time by selecting the individual with the smallest maximum UAV task time in the population. The fitness function of each individual is calculated to find the individual X i with the highest fitness in the population, which is retained as the global optimal solution.

[0164] S3.8 Adjust the position of the individual using the defense mechanism and update the individual according to the fitness.

[0165] In order to find better solutions, the solution space is then searched and the position of each individual X i in the population is updated. The specific search process can be divided into two parts: global exploration and local development. The size of two randomly generated random numbers τ9 and τ8 is used to determine whether to perform global search or local search. Each individual will choose one of the four defense strategies to change. The four defense strategies are as follows:

[0166] S3.8.1 First defense strategy

[0167] When τ8 < τ9, global exploration is performed. Global exploration can be divided into two methods, namely the first defense strategy and the second defense strategy. The size of two randomly generated random numbers can be used to determine which strategy to use. When τ6 < τ7, the first defense strategy is executed, and its mathematical model is:​

[0168]

[0169] where r is a random number from 1 to N, are other random individuals in the current total population. τ1 is a random number from a normal distribution; τ2 is a random number from the interval [0, 1]; is the current global optimal solution; the current individual will be updated according to the global optimal solution and the reference position and a random factor τ1 and τ2 are introduced during the updating, so that the individual is guided by the global optimal solution and can also explore within a certain range during the search process, thereby avoiding premature convergence. After an individual is adjusted according to the first defense strategy, the fitness of and is calculated according to steps S3.3-S3.7, and if the fitness of is greater than the fitness of , the modified individual is replaced into the population X, otherwise, no replacement is performed. Then, the fitness of and the global optimal solution is compared, and if it is greater, the is replaced by which is the new global optimal solution.

[0170] S3.8.2 Second defense strategy

[0171] When τ6> τ7, the second defense strategy is executed. The mathematical model is as follows:

[0172]

[0173] where U1 is a randomly generated 1-row n-column vector, which only contains 0 or 1. Therefore, the calculation of the first half will retain the column nodes corresponding to the value 1 of (1-U1). and are two random individuals in the population X, and τ3 is a random number from the interval [0, 1]. This updating strategy combines local information and global information. The next position of the individual will be affected by the current position and the difference between the two random individuals. After an individual is adjusted according to the second defense strategy, the fitness of and is calculated according to steps S3.3-S3.7, and if the fitness of is greater than the fitness of , the modified individual is replaced into the population X, otherwise, no replacement is performed. Then, the fitness of and the global optimal solution is compared, and if it is greater, the is replaced by called the new global optimal solution.

[0174] S3.8.3 The third defense strategy

[0175] When τ8>T9, local exploration is performed, and the local search can adopt two methods, namely the third defense strategy and the fourth defense strategy. Specifically, it can be determined by comparing the sizes of two randomly generated random numbers τ 10 and T f . When τ 10 T f , the third defense strategy is performed, and the mathematical model is as follows:

[0176]

[0177] wherein r3is a random number between 1 and N; t is the iteration number, and each individual in the population is traversed once every four defense strategies, and then the iteration number is incremented by 1, T max is the maximum iteration number; τ3is a random number between 0 and 1; and are three random individuals in the population X. is the odor diffusion factor, and the calculation method is as follows:

[0178]

[0179] wherein, is the fitness of the i-th individual at the t-th iteration, and ε is an arbitrarily small positive number. From the above formula, it can be seen that the larger is, the better the fitness of is, and the individual will be more inclined to approach the optimal solution. The formula for updating the third defense mechanism combines the current individual position, global information, fitness information, and random factors. After adjusting using the above method, the fitness of and is calculated according to steps S3.3-S3.7. If the fitness of is greater than the fitness of , the modified individual is replaced in the population X, otherwise, no replacement is performed. Then, the fitness of and the global optimal solution is compared, and if it is greater, then is replaced by called the new global optimal solution.

[0180] S3.8.4 The fourth defense strategy

[0181] When τ 10 T f , the fourth defense strategy is performed, and the mathematical model is as follows

[0182]

[0183] where, a is the velocity convergence factor set to 0.2; τ4, τ5 are random numbers between 0 and 1; F i t is the inelastic collision force generated when the individual attacks the predator with the body, and its calculation method is:

[0184]

[0185]

[0186] where, represents the other random individuals in the current total population, the current individual m i is the mass of the ith individual at iteration t. After adjusting using the above method, the fitness of and is calculated according to steps S3.3-S3.7, if the fitness of is greater than the fitness of , the modified individual is replaced into the population X, otherwise, no replacement is performed. Then the fitness of and the global optimal solution is compared, if it is greater, then is replaced by which is called the new global optimal solution.

[0187] When all individuals of the original population have traversed the four defense mechanisms, since each individual has changed and the fitness of the new individual is better than the original, the new individual will be saved to the population, so a new population is generated, and the population size is still N. In order to accelerate the convergence speed, the circulating population reduction technique is also used, and the population size is periodically changed with the increase of the iteration number t.

[0188] S3.9 Circulating population reduction technique

[0189] To improve the convergence speed, CPO adopts the cyclic population reduction strategy. When the tth iteration is completed, the population size is reduced accordingly to speed up the convergence of the population; when the population size is reduced to a certain value, the population size is increased again to increase the diversity of the population, and the cycle is repeated until the maximum iteration number is reached. In this way, the algorithm avoids wasting computing resources while speeding up the optimization process. Therefore, some individuals are obtained from the population during the optimization process, the selection of the individuals is related to the fitness to speed up the convergence, and the individuals are reintroduced into the population to improve the diversity and avoid falling into a local minimum; the cycle is based on a loop variable to determine the number of times the process is executed during the optimization process, and the mathematical model for reducing the population size in the cycle is as follows:

[0190]

[0191] Where T is a variable that determines the number of cycles, t is the current iteration number, T max is the maximum iteration number, N min is the minimum number of individuals in the newly generated population, so the population size cannot be less than N min . When the population is reduced, the fitness of each individual calculated in the previous step is used to reduce the population, the fitness is sorted, the individuals with good fitness are retained, and the individuals with poor fitness are deleted.

[0192] S3.10 gradually converges to the optimal solution by guiding the search process with the global optimal position

[0193] According to steps S3.8-S3.9, as the number of iterations increases, new individuals are constantly searched for through the four defense mechanisms, and the corresponding fitness values are calculated, the best individuals are selected to repeat the process. When the population reaches the maximum iteration number, the optimal individual saved in the algorithm is returned. The individual saves the node data of the task allocation area of each unmanned aerial vehicle, and the minimum fitness value represents the maximum task time in this case, thereby achieving the goal of balancing the charging time.

[0194] As Figure 5 shown, the fitness function with the maximum task time decreases and gradually converges as the number of iterations increases.

[0195] As Figure 2 , 3 , 4, 5, and 6 show that the present application can quickly converge the fitness function, i.e., find the optimal time-balanced multi-unmanned path faster. As Figure 7 described, the embodiment discloses a multi-unmanned aerial vehicle charging time balancing system for a wireless charging sensor network, which is used for the above method embodiment and includes the following modules:

[0196] A wireless rechargeable sensor network scenario modeling module: deploying a wireless rechargeable sensor network WRSN in a two-dimensional plane, which is composed of three parts: sensor nodes, base stations, and unmanned aerial vehicles;

[0197] A problem construction module: in the deployed WRSN scenario, the flight trajectory of multiple unmanned aerial vehicles is planned to balance the charging time of each unmanned aerial vehicle, and the problem is modeled as a maximum charging task completion time minimization problem;

[0198] A problem solving module: the CPO algorithm is used to solve the problem in the problem construction module to balance the charging time of the unmanned aerial vehicles.

[0199] Other contents of the embodiment can refer to the above method embodiments.

[0200] The embodiment discloses a storage medium which stores computer instructions for causing a computer to execute the above method or system.

[0201] The embodiment discloses an electronic device, comprising:

[0202] A processor;

[0203] A memory for storing a program, when the program is called and executed by the processor, the processor executes the above method or system.

[0204] The above has described the preferred embodiments of the present application in detail, and the protection scope of the present application is not limited to the embodiments shown in the text.

Claims

1. A method for equalizing the charging time of multiple drones in wireless charging sensor networks, characterized by: Includes the following steps: S1. Deploy a wireless rechargeable sensor network (WRSN) in a two-dimensional plane. The WRSN includes sensor nodes, base stations, and drones. S2. In the WRSN scenario deployed in step S1, by planning the flight trajectories of multiple drones, the time for each drone to complete the charging task is balanced, and the model is modeled as the problem of minimizing the maximum time to complete the charging task. In step S2, a mathematical model is performed to determine the completion time of a single drone charging task, as follows: The power transfer model for the charging task is as follows: Wireless charging is achieved between the drone and sensor nodes via line-of-sight transmission; based on the free-space path loss model, the transmission loss P of the link... w Represented as: Where d represents the transmission distance from the sensor node to the drone; f represents the drone charging frequency; Assume the UAV transmits at a constant power P. t Energy transfer is performed; at each sensor node, the received radio frequency signal is converted into a DC signal and energy is harvested, thus the radio frequency power received by the sensor node is... P r Represented as: Among them, G UAV G represents the transmit antenna gain of the drone. SN Indicates the sensor's receiving antenna gain; Let the energy conversion efficiency at the sensor node be η, then the power obtained by the sensor node is: ; The charging task time model is as follows: The total time for a drone to complete its charging task is the sum of its flight time and charging time. Given j UAVs, to achieve a balanced charging task time for each UAV, we construct a problem to minimize the maximum charging task completion time, as follows: S3. The CPO (Cuora Optimization Probability) algorithm is used to solve the problem of minimizing the maximum charging task completion time in step S2, thereby achieving a balance in drone charging time. This step is detailed below: S3.1 Initialize the population by randomly generating a certain number of individuals; S3.2 Initialize the candidate solution matrix In the initialization phase of the optimization algorithm, each row in the generator matrix X represents an individual, i.e., a candidate solution. Each individual corresponds to a complete path planning scheme, and the column number represents the label of the node. If there are a total of n nodes, then the first column represents the label of the first node in all populations, the second column represents the label of the second node in all populations, and so on, until the nth column represents the label of the nth node in all populations. S3.3, Sort candidate solutions Before calculating the fitness function, each candidate solution needs to be sorted in order to facilitate subsequent task allocation; S3.4, Task Splitting After sorting, each candidate solution is split; the goal of splitting is to evenly distribute the nodes in each solution to J UAVs; let the quotient of n / J be a and the remainder be b, then the first b UAVs are allocated a+1 nodes and the last Jb UAVs are allocated a nodes, so as to ensure that the number of nodes served by each UAV is as balanced as possible. S3.5, Add start and end coordinates; S3.6 Calculate the total mission time for each drone mission; S3.7 Calculate the fitness of each individual. The fitness is the maximum task time of the drones in the individual, and the individual with the lowest fitness is taken as the current global optimal solution; S3.

8. Use defense mechanisms to adjust the position of individuals and update individuals based on fitness; S3.9, Cyclic population reduction technology; S3.

10. The search process is guided by the global optimal position and gradually converges to the optimal solution.

2. The method for equalizing the charging time of multiple drones for wireless charging sensor networks as described in claim 1, characterized in that, Step S1 is as follows: A WRSN is deployed in a two-dimensional plane. The WRSN consists of three parts: sensor nodes. S Base stations CBS and drones UAVs The two-dimensional planar region of WRSN is represented as G(S,CBS,UAVs); Base station: with coordinates (x0, y0), serves as an energy source, providing charging support for drones, while also monitoring the location information and remaining power status of sensor nodes in real time; Sensor Nodes: Suppose there are n sensor nodes in total, forming a set S = {s1, s2, ..., sn}. n The nodes are evenly distributed in a two-dimensional plane, used for real-time monitoring of surrounding environmental data and information upload, and can also achieve wireless charging; each sensor node s i This represents the node with number i, where... The set of electricity of a node is represented as E i Let E represent the current power level of the i-th node, and let E represent the maximum power capacity of the sensor node. max Data is transmitted between nodes through multi-hop communication; Unmanned aerial vehicles (UAVs): [This section appears to be incomplete and requires further context.] J Each drone carries energy from the base station and hovers above the sensor nodes in a one-to-one manner to charge them. Once a sensor node is fully charged, the drone departs at an optimal flight speed v. * Proceed to the next sensor node to continue charging until all sensor nodes that require charging have completed their tasks, and finally return to the base station.

3. The method for equalizing the charging time of multiple drones for wireless charging sensor networks as described in claim 2, characterized in that, In step 2, the charging task time model is as follows: Assume a drone has a charging area of... m There are (i-1)th charging nodes, and the number of charging nodes is less than the total number of sensor nodes n; the flight time from the (i-1)th sensor node to the ith sensor node. for: Among them, l i l1 represents the flight distance from the (i-1)th node to the ith node; l1 represents the flight distance of the drone from the charging base station to the first node. m+1 This indicates the flight distance the drone travels from the last charging node back to the charging base station; flight distance Represents node coordinates; v * This represents the drone's flight speed in its most energy-efficient state; the total flight time of the drone is: The amount of additional charging required for each sensor node is expressed as follows: The charging time for the drone at the i-th node is: Total charging time T for the drone within the mission area h for: The total time T for the drone to complete the charging task UAV for: Among them, T f This indicates the total flight time of the drone.

4. The multi-UAV charging time equalization method for wireless charging sensor networks as described in claim 3, characterized in that, Step S3: The CPO (Cuora Optimization Probability) algorithm is used to solve the problem of minimizing the maximum charging task completion time in step S2, as detailed below: S3.1 Determine the population size N and the maximum number of iterations T. max The solution space is defined by its lower boundary L, upper boundary U, and dimension dim; initialization begins with the initial set of individuals. Where N represents the population size, X k Let be the k-th candidate solution in the search space, where each candidate solution represents a multi-UAV task node allocation method. L and U are the upper and lower bounds of the search range, respectively, and r is a random number between 0 and 1. Using the above formula, after initializing all N individuals in the population, the initialized population is obtained as follows: S3.2 In the initialization phase, each row of the generation matrix X represents an individual, and each individual corresponds to a complete path planning scheme. The column number represents the label of the node. If there are a total of n nodes, then the first column represents the label of the first node in all populations, the second column represents the label of the second node in all populations, and the nth column represents the label of the nth node in all populations. S3.3 Sort each node in each candidate solution in ascending order according to its label; S3.4, Split each candidate solution; let... n / J The business is a The remainder is b Then the former b UAV allocation a +1 node, then Jb UAV allocation a One node; S3.

5. Add start and end coordinates to the mission of each drone; S3.

6. Based on the order of the split tasks, calculate the path using the formula. Calculate the total mission time for each drone from the starting point to the destination; S3.7 Calculate the fitness of each individual in the population. This fitness is the maximum mission time of the UAVs among the individuals. The individual with the lowest fitness is taken as the current global optimum. The fitness function is: Among them, T j Let X represent the charging time of the j-th drone; calculate the fitness function for each individual, and select the individual with the highest fitness in the population. i Retained to the global optimal solution middle; S3.8 Search the solution space and update the X of each individual in the population. i Location; S3.

9. Based on the loop variable, to determine the number of times the process is executed during optimization, the mathematical model for iteratively reducing the population size is as follows: Where T is a variable that determines the number of iterations, and t is the current iteration number. max It is the maximum number of iterations, N min It is the minimum number of individuals in the newly generated population; the population size cannot be less than N. min ; S3.

10. The search process is guided by the global optimal position and converges to the optimal solution.

5. The multi-UAV charging time equalization method for wireless charging sensor networks as described in claim 4, characterized in that, In step S3.8, the search process is divided into two parts: global exploration and local exploration; two random numbers are generated randomly. and The size determines whether to perform a global search or a local search; each individual chooses one of four defense strategies to modify.

6. The multi-UAV charging time equalization method for wireless charging sensor networks as described in claim 5, characterized in that, In step S3.8, the four defense strategies are as follows: First defense strategy: when At that time, a global exploration is executed. The global exploration is divided into a first defense strategy and a second defense strategy, which is determined by comparing the size of two randomly generated random numbers; when The first defense strategy is executed at that time, and the mathematical model is as follows: in, r From 1 to N random numbers, Other random individuals in the current population; It is a random number that follows a normal distribution; It is a random number located in the interval between 0 and 1; It is the current global optimal solution; the current individual It will be based on the global optimal solution and reference position An update is performed, during which a random factor is introduced. Once an individual has adjusted according to the first defense strategy, proceed with steps S3.3-S3.

7. Perform fitness calculations if The fitness is greater than If the fitness of the modified individual is high, it will be replaced in population X; otherwise, no replacement will be made. Then compare... and the global optimal solution If the fitness is greater than 1, then... Replace with This is called the new global optimal solution; Second defense strategy: when When the second defense strategy is implemented, the mathematical model is as follows: Wherein, U1 is a randomly generated vector of 1 row and n columns, which contains only 0 or 1; The calculation results will be retained The column node corresponding to the value 1 in (1-U1); Let X be two random individuals from population X. It is a random number located in the interval between 0 and 1; after an individual adjusts according to the second defense strategy, proceed according to steps S3.3-S3.

7. and Perform fitness calculations if The fitness is greater than If the fitness of the modified individual is high, it will be replaced in population X; otherwise, no replacement will be made. Then compare... and the global optimal solution If the fitness is greater than 1, then... Replace with This is called the new global optimal solution; Third defense strategy: when Local exploration is performed, and the local search is divided into a third defense strategy and a fourth defense strategy, which compare two randomly generated random numbers. The size determines; when At that time, the third defense strategy is implemented, and the mathematical model is as follows: Where r3 is located at 1 to N The random number between T and T; t is the current iteration number. After each individual in the population has traversed all four defense strategies, the iteration number is incremented by 1. max This represents the maximum number of iterations. It is a random number between 0 and 1; Let X be three random individuals from population X. The odor diffusion factor is calculated as follows: in, Let i be the fitness of individuals at the t-th iteration. For any positive number; proceed according to steps S3.3-S3.

7. Perform fitness calculations if The fitness is greater than If the fitness of the modified individual is high, it will be replaced in population X; otherwise, no replacement will be made. Then compare... and the global optimal solution If the fitness is greater than 1, then... Replace with This is called the new global optimal solution; Fourth defense strategy: when At that time, the fourth defense strategy is implemented, and the mathematical model is as follows: in, It is the velocity convergence factor; It is a random number between 0 and 1; It is the inelastic collision force generated when an individual attacks a predator with its body, and the calculation formula is: in, This represents the final velocity of the i-th individual at the next iteration t+1. Let m be the initial velocity of the i-th individual at iteration t; i It is the mass of the i-th individual at iteration t; according to steps S3.3-S3.7... and Perform fitness calculations if The fitness is greater than If the fitness of the modified individual is high, it will be replaced in population X; otherwise, no replacement will be made. Then compare... and the global optimal solution If the fitness is greater than 1, then... Replace with This is called the new global optimal solution.

7. The method for equalizing the charging time of multiple drones for wireless charging sensor networks as described in claim 6, characterized in that, Step S3.10 is as follows: When the population reaches the maximum number of iterations, the optimal individual is saved; this individual saves the node data of the task allocation area for each UAV, and the minimum fitness value represents the maximum task time under this condition, so as to achieve the goal of equalizing charging time.

8. A multi-UAV charging time equalization system for wireless charging sensor networks, used to perform the method as described in any one of claims 1-7, characterized in that... Includes the following modules: Wireless Rechargeable Sensor Network Scene Modeling Module: Deploys a wireless rechargeable sensor network (WRSN) in a two-dimensional plane, consisting of three parts: sensor nodes, base stations, and drones; Problem building module: In the deployed WRSN scenario, by planning the flight trajectories of multiple drones, the time for each drone to complete the charging task is balanced, and the problem is modeled as minimizing the maximum time to complete the charging task. Problem Solving Module: Solve the problems in the Problem Building Module to achieve a balance in drone charging time.

9. A storage medium, characterized in that, The computer instructions are stored to cause the computer to perform the method according to any one of claims 1-7 or the system according to claim 8.

10. An electronic device, characterized in that, include: processor; A memory for storing a program that, when invoked and executed by a processor, causes the processor to perform the method as described in any one of claims 1-7 or the system as described in claim 8.

Images