UAV charging path planning method based on hybrid artificial bee colony ant colony algorithm

Abstract

The present invention discloses a method for planning a charging path for a drone based on a hybrid artificial bee colony and ant colony algorithm. The method belongs to the field of drone technology and includes the following steps: first, mathematical modeling based on all basic data of a selected scenario; second, determining the scale of the charging scenario based on the number of all sensor nodes to be charged in the selected scenario; wherein, when the number of sensor nodes is less than 500, the drone's hovering point position, flight trajectory, and number of hovering points are optimized using a hybrid artificial bee colony and ant colony algorithm to minimize the drone's total energy consumption; and when the number of sensor nodes is greater than 500, the drone's hovering point position, flight trajectory, and number of hovering points are optimized using an improved non-dominated genetic algorithm to minimize the drone's total energy consumption; and third, the drone charges the sensor nodes and collects data using the calculated optimal path and optimal hovering point. The method can plan charging paths for different charging scenarios.

Description

Technical Field

[0001] The present invention relates to a method for planning a charging path for an unmanned aerial vehicle (UAV) based on a hybrid artificial bee colony and ant colony algorithm, and belongs to the technical field of UAVs. Background Art

[0002] A Wireless Sensor Network (WSN) is a wireless network composed of a large number of stationary or mobile sensors in a self-organizing and multi-hop manner. It can collaboratively sense, collect, process and transmit information about the sensed objects within the geographical area covered by the network, and ultimately send this information to the owner of the network.

[0003] In recent years, wireless sensor networks (WSNs) have become ubiquitous in fields such as commerce, detection, and healthcare. They are network systems composed of a large number of distributed micro-sensor nodes, often equipped with both sensing and communication capabilities. However, due to the size and cost constraints of sensors, energy replenishment has become a critical issue in determining the overall system's operating time. Energy replenishment, which theoretically can extend the network's lifespan indefinitely, has attracted extensive research. Currently, the most advanced and reliable method for achieving this is through wireless power transmission technology.

[0004] Traditional artificial bee colony algorithms are primarily suitable for optimization problems in continuous space, while traditional ant colony algorithms excel at discrete combinatorial optimization. We combine these two algorithms to overcome their limitations in solving the joint optimization problem of drone scheduling and trajectories, and introduce corresponding improvement strategies. This improved hybrid artificial bee colony and ant colony algorithm (HABC-ACO) not only adapts to the dimensionality of the solution during iteration, but also accelerates convergence compared to traditional algorithms. However, due to its single-solution nature, the algorithm suffers from some shortcomings when handling large-scale problems. We propose new solutions to this problem and subsequently introduce new optimization objectives to improve the performance of the entire wireless rechargeable sensor network. To address the unsuitability of the traditional NSGA-II algorithm for solving the joint optimization problem of drone scheduling and trajectories in large-scale scenarios, we incorporate a series of improvement strategies into the algorithm, including an adaptive population dimensionality mechanism, an adversarial resolution strategy, a PMX factor in the crossover process, and a 2-OPT factor and a cross-path improvement mechanism in the mutation process. Through these improvements, an improved non-dominated genetic algorithm (IMNSGA-II) was proposed, which significantly improved the global search capability and solution accuracy of the algorithm compared with the original NSGA-II algorithm. Summary of the Invention

[0005] The present invention designs and develops a UAV charging path planning method based on a hybrid artificial bee colony ant colony algorithm, which can plan charging paths for different charging scenarios and improve the wireless charging efficiency of UAVs.

[0006] The technical solution provided by the present invention is:

[0007] A UAV charging path planning method based on a hybrid artificial bee colony ant colony algorithm includes:

[0008] Step 1: Based on the number and location of sensor nodes in the selected scenario, the basic data on the power required by the sensor nodes, and the environmental parameters of the application scenario, the power parameters of the drone, and the corresponding charging parameters, mathematical modeling of the scenario is performed;

[0009] Step 2: Determine the scale of the charging scenario based on the number of all sensor nodes to be charged in the selected scenario;

[0010] Among them, when the number of sensor nodes is less than 500, it is a small-scale scenario;

[0011] Based on the hybrid artificial bee colony and ant colony algorithm, the drone's hovering point position, flight trajectory and number of hovering points are optimized to minimize the drone's total energy consumption.

[0012] When the number of sensor nodes is greater than 500, it is a large-scale scenario;

[0013] Based on the improved non-dominated genetic algorithm, the UAV hovering point position, UAV flight trajectory and the number of hovering points are optimized to minimize the total energy consumption of the UAV.

[0014] Step 3: The drone charges the sensor nodes and collects data through the calculated optimal path and optimal hovering point.

[0015] Preferably, the above-mentioned UAV charging path planning method based on hybrid artificial bee colony and ant colony algorithm is characterized in that the hybrid artificial bee colony and ant colony algorithm includes:

[0016] Step a1, initializing the relevant parameters of the hybrid artificial bee colony and ant colony algorithm;

[0017] Random Generation The hover position sets constitute the total set , each set Include Hover points, 、 , and initialize it;

[0018] Step a2: Hired bees perturb each coordinate to obtain a new hovering point set for each hovering point set and add it to the total set. , making ;

[0019] Step a3: Enter the inner loop and use the ant colony to complete a loop for each hovering point set, updating the pheromone concentration of each edge;

[0020] When the number of inner loops is reached, each hovering point set selects the Hamiltonian circuit with the largest sum of pheromone concentrations as the optimal path, otherwise it will continue to enter the inner loop;

[0021] Step a4: Calculate the total energy consumption, fitness function and retention probability of each hovering point set, and form a new total set based on the hovering point sets retained by probability. and number of collections ; Select the total energy consumption of the hovering set with the largest retention probability, recorded as ; Set the optimal solution for energy consumption under the current situation to ;

[0022] when , after updating the optimal solution, if the number of outer loops is reached, the global optimal solution is output;

[0023] when , if the number of outer loops is reached, the global optimal solution is directly output;

[0024] If the number of cycles has not been reached, go back to step a2 and reset the number of inner cycles;

[0025] Step a5: Output the optimal result and the calculation ends.

[0026] Preferably, the improved non-dominated genetic algorithm includes:

[0027] Step b1, initializing the population, parameters and hovering position set;

[0028] Step b2: Generate a new hovering position set based on the adaptive population dimension mechanism and the confrontation elimination strategy, and set the evolutionary generation , the maximum evolutionary generation , population size ;

[0029] Step b3: After performing non-dominated sorting and crowding calculation on the initial population, the first generation population is generated;

[0030] Perform a crossover operation with PMX factor improvement on the first generation population to generate a crossover subpopulation;

[0031] Step b4: performing a mutation operation on the first generation population with the 2-OPT factor and the cross-path improvement mechanism to generate a mutant population;

[0032] Step b5: merge the first generation population, crossover subpopulation and mutation population into a set, and perform non-dominated sorting and crowding calculation on the elements in the set before truncating them.

[0033] Step b6: When , the global optimal solution is output;

[0034] when , return to the crossover step until the global optimal solution is output.

[0035] Preferably, the step a3 includes:

[0036] Enter the inner loop and for each and , , generated using the ant colony algorithm The optimal path ;

[0037] Set each ant to , , from a random starting point Start and select the next hover point The probability is: ;

[0038] Where, For any ant, is the total number of ants, For the set of all remaining points, 、 is the heuristic factor weight, For two points 、 The pheromone concentration between the edges, For two points 、 The pheromone concentration between the edges, for 、 The reciprocal of the distance between two points, for 、 The reciprocal of the distance between two points;

[0039] All ants return to their starting point After that, update the information concentration , pheromone enhancement is ;

[0040] Where, It's an ant The total length of the path, Add weight to pheromones;

[0041] Update pheromone matrix ;

[0042] If the inner cycle is reached , select the pheromone concentration and the largest Hamiltonian circuit as Go to the next step, otherwise return to this step.

[0043] Preferably,

[0044] The calculation formula of the fitness is: ;

[0045] The calculation formula of the retention probability is: ;

[0046] Where, is the total energy consumption, is the traversal index of all individuals in the population, and its value range is , representing the total number of individual, is the size of the population after merging.

[0047] The beneficial effects of the present invention are as follows: the WRSNS UAV charging path planning method based on the hybrid artificial bee colony ant colony algorithm provided by the present invention can apply different optimization methods to scenarios of different scales to minimize the total energy consumption of UAV charging, shorten the charging time, and improve the charging efficiency.

[0048] The traditional artificial bee colony algorithm and the traditional ant colony algorithm are combined to overcome the limitations of the original algorithm in solving the joint optimization problem of drone scheduling and trajectory. This not only adapts to the dimension of the solution during the iterative process, but also accelerates the convergence of the algorithm.

[0049] The improved non-dominated genetic algorithm introduces an adaptive population mechanism and an adversarial elimination strategy, and introduces the PMX factor in the crossover process, and introduces the 2-OPT factor and the crossover path improvement mechanism in the mutation process. Through these improvements, the global search capability and solution accuracy of the algorithm are significantly improved.

[0050] In this method, based on the adaptive population mechanism, the method can maintain a diversity index of 0.18-0.35 in the later stage of iteration, which can effectively avoid the problem of premature convergence. Especially when dealing with heterogeneous node distribution, compared with traditional algorithms, it can significantly reduce the standard deviation of charging efficiency, showing superior robustness and scenario adaptability.

[0051] In smart agricultural monitoring systems, this method can shorten the total flight distance of drones and reduce the time it takes to complete charging tasks. It can also reduce peak energy consumption and reduce ineffective path planning. In disaster response, it can cover the optimal charging path for the monitored area within seconds. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of the drone-assisted wireless rechargeable sensor network described in the present invention.

[0053] Figure 2 This is a flowchart of the multi-objective optimization-based charging path planning algorithm for the UAV-assisted wireless rechargeable sensor network described in the present invention.

[0054] Figure 3 This is a schematic diagram of the optimization results of the hovering point and trajectory points of the UAV when the number of ground sensor nodes is 300 according to the present invention.

[0055] Figure 4 This is a comparison chart of the fitness function values ​​of different algorithms when the number of ground sensor nodes is 100 according to the present invention.

[0056] Figure 5 This is a schematic diagram of the optimization results of the drone hovering point and trajectory using the IMNSGA-II algorithm when the sensor node scale is 500 as described in the present invention.

[0057] Figure 6 This is a diagram showing the effect of the drone hovering point and trajectory using the IMNSGA-II algorithm described in the present invention when the sensor node scale is 500.

[0058] Figure 7 This is a schematic diagram of the optimization results of the drone hovering point and trajectory using the IMNSGA-II algorithm described in the present invention when the sensor node scale is 1000.

[0059] Figure 8 This is a diagram showing the effect of the drone hovering point and trajectory using the IMNSGA-II algorithm described in the present invention when the sensor node scale is 1000.

[0060] Figure 9 This is a schematic diagram of the optimization results of the drone hovering point and trajectory using the IMNSGA-II algorithm when the sensor node scale is 1500 as described in the present invention.

[0061] Figure 10 This is a diagram showing the effect of the drone hovering point and trajectory using the IMNSGA-II algorithm described in the present invention when the sensor node scale is 1500.

[0062] Figure 11 This is a comparison chart of the effectiveness verification of the improved algorithm described in the present invention and the existing algorithm. DETAILED DESCRIPTION

[0063] The present invention will be described in further detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0064] like Figure 1-11 As shown, the present invention provides a UAV charging path planning method based on a hybrid artificial bee colony ant colony algorithm. By using a hybrid artificial bee colony and ant colony algorithm and an improved non-dominated genetic algorithm, charging paths are planned for different charging scenarios to improve the wireless charging efficiency of the UAV, including:

[0065] Step 1: Based on all the basic data of the selected scenario, mathematical modeling is performed according to the environmental parameters of the application scenario, the power parameters of the drone, and the corresponding charging parameters;

[0066] Among them, the basic data include: the number and location of sensor nodes, the power required by sensor nodes;

[0067] UAV power parameters include: charging UAV transmission power and propulsion speed, and charging parameters include: charging energy consumption and effective charging distance;

[0068] Step 2: Determine the scale of the charging scenario based on the number of all sensor nodes to be charged in the selected scenario;

[0069] Among them, when the number of sensor nodes is less than 500, it is a small-scale scenario.

[0070] Based on a hybrid artificial bee colony and ant colony algorithm, the drone's hovering point location, flight trajectory, and number of hovering points are optimized to minimize the drone's total energy consumption, including:

[0071] Step a1, initializing the relevant parameters of the hybrid artificial bee colony and ant colony algorithm;

[0072] Random Generation The hover position sets constitute the total set , each set Include Hover points, 、 , and initialize it;

[0073] Step a2: Hired bees perturb each coordinate to obtain a new hovering point set for each hovering point set and add it to the total set. , making ;

[0074] Step a3: Enter the inner loop and use the ant colony to complete a loop for each hovering point set, updating the pheromone concentration of each edge;

[0075] When the number of inner loops is reached, each hovering point set selects the Hamiltonian circuit with the largest sum of pheromone concentrations as the optimal path, otherwise it will continue to enter the inner loop;

[0076] Step a4: Calculate the total energy consumption, fitness function and retention probability of each hovering point set, and form a new total set based on the hovering point sets retained by probability. and number of collections ; Select the total energy consumption of the hovering set with the largest retention probability, recorded as ; Set the optimal solution for energy consumption under the current situation to ;

[0077] when , after updating the optimal solution, if the number of outer loops is reached, the global optimal solution is output;

[0078] when , if the number of outer loops is reached, the global optimal solution is directly output;

[0079] If the number of cycles has not been reached, go back to step a2 and reset the number of inner cycles;

[0080] Step a5: Output the optimal result and the calculation ends.

[0081] When the number of sensor nodes is greater than 500, it is a large-scale scenario;

[0082] Based on the improved non-dominated genetic algorithm, the hovering point position, flight trajectory and number of hovering points of the UAV are optimized to minimize the total energy consumption of the UAV, including:

[0083] Step b1, initializing the population, parameters and hovering position set;

[0084] Step b2: Generate a new hovering position set based on the adaptive population dimension mechanism and the confrontation elimination strategy, and set the evolutionary generation

[0085] , the maximum evolutionary generation , population size ;

[0086] Step b3: After performing non-dominated sorting and crowding calculation on the initial population, the first generation population is generated. , and set ;

[0087] Perform a crossover operation with PMX factor improvement on the first generation population to generate a crossover subpopulation;

[0088] Step b4: performing a mutation operation on the first generation population with the 2-OPT factor and the cross-path improvement mechanism to generate a mutant population;

[0089] Step b5: merge the first generation population, crossover subpopulation, and mutation population into a set, perform non-dominated sorting and crowding calculation on the elements in the set, and then perform truncation;

[0090] Step b6: When , the global optimal solution is output;

[0091] when , return to the crossover step until the global optimal solution is output.

[0092] Step 3: The drone charges the sensor nodes and collects data through the calculated optimal path and optimal hovering point.

[0093] Step 1: Determine the number of sensor nodes , sensor node location Flight altitude of the drone .

[0094] Step 2: Calculate the charging power transmission efficiency between the drone and the node , data transmission rate Energy consumption of drones .

[0095] (a) The relationship between the node and the drone at any time distance :

[0096]

[0097] in, It is a sensor node The three-dimensional coordinates of is the coordinate of the drone at any moment;

[0098] (b) Charging power transmission efficiency :

[0099] ;

[0100] in, is the UAV transmission power, is the node receiving power, 、 、 These are all related constants for charging drones.

[0101] (c) Data transmission rate between nodes and drones ;

[0102] ;

[0103] in, is the system bandwidth, is the UAV transmission power, is the channel gain, is the white noise power, For any time between the node and the drone distance.

[0104] Will Defined as The data transmission time is for:

[0105] ;

[0106] (d) Total energy consumption of the drone during the charging cycle :

[0107]

[0108]

[0109]

[0110]

[0111] in, Energy consumption for charging drones, is the energy collected by all nodes, Each node Required electrical energy. and They are respectively the UAV flight energy consumption and hovering energy consumption, and are flight power and hovering power respectively, and are flight time and hovering time respectively. The propulsion power of the UAV is calculated as follows:

[0112] ;

[0113] in, and are two constants, namely the blade power and induced power of the drone in hovering state. is the rotor blade speed, For the moment The speed of the drone, is the average rotor induced speed in hover, and are the fuselage drag ratio and rotor solidity, and are air density and rotor disc area respectively. When the UAV propulsion speed is When the propulsion power ; When the drone propulsion speed is 0, that is, hovering, the hovering power .

[0114] Drone at the hovering point The hovering time is ,in, A binary variable representing the , Can the first sensor node connect to the If the drone is on the first hovering point The first hover point can be Nodes are connected, then , otherwise 0. for sensor nodes to the The data transmission time of the drone at each hovering point, and the total hovering time of the drone can be expressed as: , then the hovering energy consumption of the UAV is for ,From the above, the total energy consumption of the drone during the entire charging cycle is: .

[0115] Step 3: Construct different optimization problems for small-scale and large-scale scenarios and use different solutions to solve them, as follows:

[0116] (a) For small-scale scenarios, the objective function is designed as follows:

[0117]

[0118]

[0119]

[0120]

[0121] in, To optimize the total energy consumption under the target, Indicates the hover point The coordinates of the hover point, Indicates the Hover point to The distance between the hovering points, is a binary variable, indicating and Is there a trajectory between them? is the total number of hover points, is the total energy received by the sensor node, the objective function It means minimizing the total hovering energy consumption of the drone at all hovering points. The fewer the number of hovering points and the more appropriate the positions, the lower the hovering energy consumption.

[0122] Objective function It means minimizing the total flight energy consumption of the UAV between hovering points. The UAV flies at a constant speed. The shorter the total length of the trajectory, the lower the flight energy consumption. Then it is the charging energy consumption in step 2. Based on these objective functions, the hovering point position, UAV flight trajectory and the number of hovering points are optimized by hybrid artificial bee colony and ant colony algorithm (HABC-ACO) to minimize the total energy consumption of the UAV, which is the sum of hovering energy consumption, flight energy consumption and charging energy consumption.

[0123] The specific calculation process is as follows:

[0124] (i) Initialize the maximum number of outer loop iterations , maximum number of inner loop iterations , Hover point number range , Initialize the swarm size , scout bee trigger threshold ; Number of ants , pheromone volatility coefficient , pheromone enhancement weight , heuristic factor weight 、 ; Randomly generated The hover position sets constitute the set , each set Include Hover points, 、 ; Initialize each with the same value Every two points 、 The pheromone concentration between the edges ;

[0125] (ii) Enter the outer loop and for each set Each coordinate of Perturbation generates new coordinates , get the new hover position set and , updated through the adversarial elimination strategy and Make sure it meets the requirements and add it to Middle, last The number of elements in ;

[0126] (iii) Enter the inner loop and for each and , , generated using the ant colony algorithm (ACO) The optimal path The specific process is: each ant , , from a random starting point Start and select the next hover point The probability of ;

[0127] in, for 、 The reciprocal of the distance between two points, for 、 The reciprocal of the distance between two points, For the set of all remaining points, all ants return to their starting points After that, update the information concentration , pheromone enhancement is ,in It's an ant Total path length, update the pheromone matrix If the inner cycle is reached , select the pheromone concentration and the largest Hamiltonian circuit as and go to (iv), otherwise return to (iii);

[0128] (iv) For each According to the optimal path Calculate total energy consumption , calculate the fitness By probability Choose whether to keep ,form and , and select The largest Recorded as ,make 、 ,like , then record 、 、 .like continuous If all are retained, the scout bee will randomly reset its coordinates and ensured by adversarial elimination strategy and Meet requirements; update 、 If the external circulation is reached , then turn to (v), otherwise return to (ii);

[0129] ( ) Output 、 、 , the calculation ends.

[0130] (b) For large-scale scenarios, the objective function is designed as follows:

[0131]

[0132]

[0133]

[0134]

[0135]

[0136] Similar to the representation of small-scale scenes, is the total hovering energy consumption, is the total flight energy consumption, is the total charging energy consumption, is the energy utilization rate of the UAV, is the total energy received by the sensor node, is the total energy emitted by the drone. The overall objective function uses the IMNSGA-II algorithm to optimize the drone's hovering point location, flight trajectory, and number of hovering points to minimize the drone's total energy consumption (the sum of hovering energy consumption, flight energy consumption, and charging energy consumption) and maximize the drone's energy utilization. The specific calculation process is as follows:

[0137] (i) Parameter and population initialization, initialization of the maximum number of hovering point iterations , set the evolutionary generation , the maximum evolutionary generation , population size , cross ratio , variation ratio , coefficient of variation , hover position set , Number of hover points ;

[0138] (ii) Apply the adaptive population dimension mechanism to obtain the updated hover position set , and the number of hover points at this time and through the counter-elimination strategy and Satisfy requirements and iterate times, and then generate the initial population in the form of a structure;

[0139] (iii) After performing non-dominated sorting and crowding calculation on the initial population, the first generation population is generated. And set ;

[0140] (iv) For populations Perform a crossover operation with the PMX operator to generate a crossover subpopulation ;

[0141] (v) For populations Perform a mutation operation with a 2-OPT factor and a cross-path reduction mechanism to generate a mutant subpopulation ;

[0142] (vi) Fusion population 、 、 For collection ,right After the elements in the array are sorted and the congestion degree is calculated, they are truncated and the top Individuals, gather at this time The individuals in ;

[0143] (vii) If , output the result; otherwise Then, proceed to (iv);

[0144] (viii) Output the global optimal path , minimum total energy consumption , maximum system efficiency .

[0145] Step 4: The drone charges the sensor nodes and collects data through the calculated optimal path and optimal hovering point.

[0146] Example 1

[0147] In order to verify the effectiveness of IMSA in solving the problem, a comparative experiment was carried out when the number of sensor nodes was 100. The comparison algorithms used were Genetic Algorithm (GA), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and Simulated Annealing (SA). Figure 4 The comparison results are shown, where the horizontal axis represents the number of experimental iterations and the vertical axis represents the global total energy consumption. Figure 4 It can be seen that the HABC-ACO algorithm has a smaller fitness function value and a faster convergence speed than other algorithms.

[0148] Example 2

[0149] Although the improved algorithm has achieved a significant improvement in convergence performance, it is also necessary to ensure the stability of the algorithm. In order to eliminate randomness, the IMSA algorithm and other comparison algorithms were calculated 50 times in different scenarios. Table 1 compares the maximum, minimum, mean and standard deviation of each algorithm (Note: The data in the table are divided by ).

[0150] Table 1 Numerical comparison of various algorithms

[0151]

[0152] As can be seen from Table 1, HABC-ACO has better stability than other algorithms. After 50 experimental results, all the algorithm data are optimal, indicating that the algorithm has excellent stability in scenarios of different scales.

[0153] Example 3

[0154] In order to verify the computing time of the proposed HABC-ACO algorithm in different scale scenarios, the IMNSGA-II algorithm provided by the present invention is compared with the HABC-ACO algorithm to prove the advantages of the HABC-ACO algorithm in small-scale scenarios and the advantages of the IMNSGA-II algorithm in large-scale scenarios.

[0155] In order to eliminate random errors, the data in the table are the average values ​​of 50 calculation results.

[0156] Table 2 Comparison of operation time of two improved algorithms under different scale conditions

[0157]

[0158] As can be seen from Table 2, the HABC-ACO algorithm takes less computing time in small-scale scenarios, which means that the algorithm is more suitable for small-scale scenarios. As the number of nodes increases, the time required for the algorithm to complete the iteration increases significantly. The IMNSGA-II algorithm performs more strongly in large-scale scenarios, so the IMNSGA-II algorithm is used in large-scale scenarios.

[0159] Example 4

[0160] like Figure 5-10 The figure shows the two-dimensional and three-dimensional results of the IMNSGA-II algorithm for the motion trajectory of the drone hovering point set when the sensor node scale is 500, 1000, and 1500. As can be seen from the figure, while the increase in the node scale also increases the number of hovering points, the algorithm still optimizes complex paths effectively. The drone paths do not intersect, proving that the algorithm is very effective in optimizing the problem.

[0161] Example 5

[0162] In order to verify the effectiveness of this algorithm in multi-objective problems, comparative tests were conducted when the number of nodes was 500, 1000, and 1500. The comparative algorithms used were NSGA-II (Nondominated Sorting Genetic Algorithm II), NSGA-III (Nondominated Sorting Genetic Algorithm III), MOPSO (Multi-Objective Particle Swarm Optimization), and MOALO (Multi-Objective Antlion Optimization). Table 3 shows the results of the two objectives after optimization (Note: The data in the table are divided by ), and The lower the two values, the more effective the improvement. As can be seen from Table 3, the IMNSGA-II algorithm has a better performance in the two objectives. and Each has the best results.

[0163] Table 3 Multi-objective optimization results of the algorithm in different scenarios

[0164]

[0165] Example 6

[0166] In order to verify whether the performance of the algorithm in solving problems is significantly improved after adding improvement points, 500 nodes are taken as an example.

[0167] like Figure 11 As shown, the horizontal axis represents the number of algorithm iterations, and the vertical axis represents + Calculation results. IMNSGA-II is the algorithm proposed in this paper; IMPROVE indicates that the proposed improvements were introduced in the 150th iteration; NSGA-II represents the unmodified algorithm. The graph shows that the algorithm converges significantly faster after the improvements are introduced, ultimately converging to a result similar to IMNSGA-III, effectively demonstrating the effectiveness of the improvements.

[0168] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the description and implementation methods. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A WRSNS UAV charging path planning method based on a hybrid artificial bee colony ant colony algorithm, characterized in that: include: Step 1: Based on the number and location of sensor nodes in the selected scenario, the basic data on the power required by the sensor nodes, and the environmental parameters of the application scenario, the power parameters of the drone, and the corresponding charging parameters, mathematical modeling of the scenario is performed; Step 2: Determine the scale of the charging scenario based on the number of all sensor nodes to be charged in the selected scenario; Among them, when the number of sensor nodes is less than 500, it is a small-scale scenario; Based on the hybrid artificial bee colony and ant colony algorithm, the drone's hovering point position, flight trajectory and number of hovering points are optimized to minimize the drone's total energy consumption. When the number of sensor nodes is greater than 500, it is a large-scale scenario; Based on the improved non-dominated genetic algorithm, the hovering point position, flight trajectory and number of hovering points of the UAV are optimized to minimize the total energy consumption of the UAV. Step 3: The drone charges the sensor nodes and collects data through the calculated optimal path and optimal hovering point; The hybrid artificial bee colony and ant colony algorithm includes: Step a1, initializing the relevant parameters of the hybrid artificial bee colony and ant colony algorithm; Random Generation The hover position sets constitute the total set , each set Include Hover points, 、 , and initialize it; Step a2: Hired bees perturb each coordinate to obtain a new hovering point set for each hovering point set and add it to the total set. , making = ; Step a3: Enter the inner loop and use the ant colony to complete a loop for each hovering point set, updating the pheromone concentration of each edge; When the number of inner loops is reached, each hovering point set selects the Hamiltonian circuit with the largest sum of pheromone concentrations as the optimal path, otherwise it will continue to enter the inner loop; Step a4: Calculate the total energy consumption, fitness function and retention probability of each hovering point set, and form a new total set based on the hovering point sets retained by probability. and number of collections ; Select the total energy consumption of the hovering set with the largest retention probability, recorded as ; Set the optimal solution for energy consumption under the current situation to ; when , after updating the optimal solution, if the number of outer loops is reached, the global optimal solution is output; when , if the number of outer loops is reached, the global optimal solution is directly output; If the number of cycles has not been reached, go back to step a2 and reset the number of inner cycles; Step a5: Output the optimal result and the calculation ends.

2. The WRSNS UAV charging path planning method based on hybrid artificial bee colony ant colony algorithm according to claim 1 is characterized in that: The improved non-dominated genetic algorithm includes: Step b1, initializing the population, parameters and hovering position set; Step b2: Generate a new hovering position set based on the adaptive population dimension mechanism and the confrontation elimination strategy, and set the evolutionary generation , the maximum evolutionary generation , population size ; Step b3: After performing non-dominated sorting and crowding calculation on the initial population, the first generation population is generated; Perform a crossover operation with PMX factor improvement on the first generation population to generate a crossover subpopulation; Step b4: performing a mutation operation on the first generation population with the 2-OPT factor and the cross-path improvement mechanism to generate a mutant population; Step b5: merge the first generation population, crossover subpopulation and mutation population into a set, and perform non-dominated sorting and crowding calculation on the elements in the set before truncating them. Step b6: When , the global optimal solution is output; when , return to the crossover step until the global optimal solution is output.

3. The WRSNS UAV charging path planning method based on hybrid artificial bee colony ant colony algorithm according to claim 1 or 2 is characterized in that: The step a3 comprises: Enter the inner loop and for each and , , generated using the ant colony algorithm The optimal path ; Set each ant to , , from a random starting point Start and select the next hover point The probability is: ; Where, is any ant, is the total number of ants, For the set of all remaining points, 、 is the heuristic factor weight, For two points 、 The pheromone concentration between the edges, For two points 、 The pheromone concentration between the edges, for 、 The reciprocal of the distance between two points, for 、 The reciprocal of the distance between two points; All ants return to their starting point After that, update the information concentration , pheromone enhancement is ; Where, It's an ant The total length of the path, Add weight to pheromones; Update pheromone matrix ; If the inner cycle is reached , select the pheromone concentration and the largest Hamiltonian circuit as Go to the next step, otherwise return to this step.

4. The WRSNS UAV charging path planning method based on hybrid artificial bee colony ant colony algorithm according to claim 3 is characterized in that: The calculation formula of the fitness is: ; The calculation formula of the retention probability is: ; Where, is the total energy consumption, is the traversal index of all individuals in the population, and its value range is , representing the total number of individual, is the size of the population after merging.

Images