Unmanned aerial vehicle intelligent scheduling method and system based on multi-service appeal construction

Abstract

The invention discloses an unmanned aerial vehicle intelligent scheduling method and system based on multi-business appeal construction, and the method comprises the steps: designing different inspection regions according to a plurality of business appeals, carrying out the analysis of the regions with different appeals of a plurality of business lines, and carrying out the intelligent arrangement of a plurality of unmanned aerial vehicles in a target region, intelligent scheduling in the inspection area is performed based on the initial position and the flight radius of the unmanned aerial vehicle in the area, effective scheduling and flight efficiency of the unmanned aerial vehicle are ensured, the utilization rate of the corresponding flight gap is improved, and the idle time of equipment is reduced. The flight time and the flight estimated electric quantity in the multiple flight tasks are planned, and task shifts are scheduled and arranged by using a simulated annealing algorithm in cooperation with the flight task planning time. And the optimal scheduling of the flight task of the unmanned aerial vehicle in the planning time is realized.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) route planning technology, and particularly relates to an intelligent UAV scheduling method and system based on multiple business requirements. Background Technology

[0002] In existing technologies, the scheduled execution of drone flight missions is constrained by objective factors such as battery power and airspace conditions. This can easily lead to situations where, after completing a certain route, the drone lacks sufficient power to perform subsequent routes, requiring it to return to recharge. Alternatively, there may be situations where there are no drones available for specific flight missions at certain times, which greatly affects flight efficiency. Most existing solutions are based on time-sharing scheduling after the drone has sufficient power, such as waiting half an hour after a flight before resuming flight. This results in excessively long idle time for the drone, or the drone may continue to perform subsequent routes, but due to insufficient power, it may have to return to recharge halfway through the route, failing to complete the flight mission and resulting in incomplete business processing and data analysis.

[0003] As business operations become more complex and business scenarios become increasingly diverse across departments, different government entities within the same administrative region have different jurisdictions and requirements, necessitating frequent dispatch of drones for flight missions. This leads to an increase in drone flight frequency and impacts the drone's operational capabilities.

[0004] In summary, the current practice of multiple drones flying along preset routes presents the following problems: 1. In the scheduling of multiple drone nests, it is necessary to manually monitor the battery level of each drone in each nest. Assigning tasks based on battery level can lead to situations where some devices are idle while others are in contact to perform tasks.

[0005] 2. Charging time should be fully considered in scheduled tasks, and tasks should not be executed continuously.

[0006] 3. When performing continuous missions, the drone's battery level needs to be manually monitored to avoid insufficient battery power during the second mission, which could prevent the flight path from being completed.

[0007] 4. After completing one route, a second route cannot be intelligently selected based on existing conditions and battery level. Summary of the Invention

[0008] To address the aforementioned technical issues, this invention discloses an intelligent drone scheduling method and system based on multiple business requirements. Different inspection areas are designed according to multiple business requirements, and multiple drones in the target area are intelligently orchestrated by analyzing the areas with different requirements of multiple business lines. Intelligent scheduling is performed within the inspection area based on the starting position and flight radius of the drones in the area, ensuring effective scheduling and flight efficiency of drones, improving the utilization rate of corresponding flight gaps, and reducing equipment idle time.

[0009] 1. Conduct flight path analysis for each inspection area to assess the drone's flight area, required operation time, flight duration, and power consumption; 2. The system autonomously calculates flight time and battery consumption based on the area and the drone's starting distance. 3. When objective conditions are met, such as sufficient power supply, suitable time and airspace for the operation, etc., the system will intelligently select a suitable flight route for flight. 4. If there is no flight route that meets the objective conditions, conduct a delay assessment, such as charging, waiting for the operation time to arrive, etc., and dispatch the drone to perform the task when the objective conditions are met. 5. Conduct performance analysis based on daily execution data to determine the drone's idle time, and increase the number of shifts based on the idle time.

[0010] To achieve the above objectives, the technical solution adopted in this invention is as follows: a UAV intelligent scheduling method based on multiple service requirements. This method utilizes multi-task intelligent orchestration technology to achieve intelligent orchestration of general UAV flight missions. It plans the flight time and estimated battery power for multiple flight missions, and uses simulated annealing algorithm to schedule and orchestrate mission shifts in conjunction with the planned flight time. This achieves optimal scheduling of UAV flight missions within the planned time. Implementation process: (1) Obtain all flight plans that require flight and UAV nest information that conforms to the flight plan, and build a basic data source, including nest coordinates, basic standard power (this standard power is based on the lowest power of the UAV among multiple UAVs as 100%), safe return value power (default 20%), charging efficiency; route start point, route end point, flight duration, energy consumption, priority, and time window.

[0011] Flight path origin: The point where the drone's operational flight path begins; Flight path endpoint: The location where the drone completes its operational flight path; Flight duration: The flight time of the drone from the starting point to the end point of the flight path; Energy consumption: The amount of electricity consumed during flight; Priority: Flight missions are prioritized based on their urgency; Time windows are mostly sorted based on time windows.

[0012] (2) Analyze the flight distance, maneuver and starting point position data in the flight plan, match them with the nest position, sort the distances from each nest to the starting point position, and construct a nest sort set that matches the flight plan.

[0013] (3) Construct the constraints of the drone, including maximum flight time and pre-charging time.

[0014] (4) Construct a scheduler based on the above data, calculate route priority first, and generate initial sorting rules based on this.

[0015] (5) Initial solution production: Based on the initial sorting rules in (4), process each route one by one, find the nest sorting set in (2) above, and calculate the total cost of executing the route one by one (total energy consumption = energy consumption to the starting point + energy consumption of the route flight). Select the optimal UAV for each route. At this time, two situations may occur: ① Current battery level of the drone ≥ Total energy consumption: Directly allocate resources to the drone, where the mission execution cost is the flight time.

[0016] ② Current drone battery level < Total energy consumption: Calculate charging time, where mission execution cost is flight time + charging time.

[0017] Select the drone allocation with the lowest total cost, and calculate the production allocation result.

[0018] (6) Construct a cost calculation function. The total production cost is constructed based on a weighted formula (1× total completion time + 2× total charging time [seconds] + 10× priority task delay [number]) + penalty term (1000 × power violation penalty). Select the one with the fewest penalty terms as the basic cost solution, and temporarily set it as the optimal cost and optimal solution.

[0019] (7) Simulate the main annealing loop, construct a temperature decrease, and perform loop iteration. The initial temperature is preset to 1000. Within 150 iterations per temperature, randomly generate neighbor solutions through the basic cost solution in (6), calculate the neighbor solution cost, and the neighbor solution production strategy is as follows (8). If the neighbor cost is lower, the neighbor solution replaces the current solution, and the neighbor cost replaces the current cost. If the neighbor cost is higher, calculate the acceptance probability, acceptance probability = exp([current cost - neighbor cost] / current temperature). Compare the generated random number with the acceptance probability. If the random number is less than the probability, the neighbor solution replaces the current solution, and the neighbor cost replaces the current cost.

[0020] (8) Based on the results in (7) loop, compare the current cost with the optimal cost. If the current cost is less than the optimal cost, then replace the optimal cost with the current cost and the optimal solution becomes the current solution.

[0021] Furthermore, the neighbor-based production strategy includes the following disturbance operations: 1. Swap flight path order: randomly select two flight paths and swap their execution order; 2. Reassign drone nests: randomly select a flight path, replace the drone nest, and rearrange the charging time considering the new drone's battery level and current position; 3. Adjust start time: randomly select a flight path and adjust the start time by ±30%; 4. Insert charging tasks: specifically adjust the time for flight paths that require charging to better coordinate charging with tasks; 5. Swap charging order: start the swapping of charging order from high-energy-consuming flight paths to optimize charging station utilization.

[0022] Furthermore, cooling is performed simultaneously in step (8). Considering our current computational scale and cost, the termination temperature should make the final acceptance rate close to 0, ensuring that the temperature is low enough that it almost no longer accepts different solutions, thus ensuring that the algorithm has fully converged. The current temperature cooling rate is preset between 0.85 and 0.95, until the current temperature is less than 0.1 and the loop ends. The temperature cooling rate is a key parameter for controlling the rate of temperature decrease. In the algorithm implementation, the temperature decreases exponentially, that is, after each iteration, the temperature is updated to the current temperature multiplied by the cooling rate (e.g., temperature := temperature × cooling rate). This multiplicative cooling method ensures that the temperature gradually decreases with each iteration, but remains positive, avoiding premature termination of the algorithm.

[0023] A higher temperature cooling rate results in a lower current temperature, more iterations, and higher computational costs. As a preferred approach, a temperature cooling rate of 0.93 is preferred, and the current temperature can be set to a decimal infinitely close to 0.

[0024] This invention also discloses an intelligent drone scheduling system based on multiple service requirements, used to execute the aforementioned intelligent drone scheduling method based on multiple service requirements, comprising: Input layer: Provides the initial scheduling of the required data, including basic data sources such as charging efficiency, drones, flight paths, and drone nests; Processing layer: Based on the data from the input layer, it builds a scheduler and generates initial sorting rules; Algorithm layer: Perform simulated annealing algorithm scheduling, iterative calculation, and obtain the optimized route scheduling order; Business layer: Provides business output to the terminal, outputting the final route scheduling plan, including the execution order of the routes and the execution nest, and counts the number of charging times and charging time required by the drone. Based on the above data, it summarizes the drone's efficiency, cost indicators, and quality indicators to determine the drone's flight performance. The drone's efficiency includes total completion time and utilization rate of each drone; the cost indicators include total cost and total charging time; and the quality indicators include route coverage and charging waiting time.

[0025] The present invention has the following beneficial effects: 1. It can intelligently schedule drone flight missions, reducing the total completion time and improving drone utilization.

[0026] 2. Reduce the cost of drone flights, ensure that the total energy consumption of daily flight time is within an effective threshold, and reduce charging time and number of charging sessions.

[0027] 3. Ensure flight quality, achieve maximum route coverage, and reduce the number of routes that cannot be flown due to scheduling issues.

[0028] 4. Ensure flight safety and avoid problems such as drones running out of power and being unable to continue operating during missions. Attached Figure Description

[0029] Figure 1 This is a flowchart illustrating the simulated annealing process of the UAV intelligent scheduling method based on multiple service requirements, as described in this embodiment of the invention.

[0030] Figure 2 This is an overall architecture diagram of an intelligent drone scheduling system built based on multiple service requirements, according to an embodiment of the present invention.

[0031] Figure 3 This is a schematic diagram of a multi-service drone task based on a drone intelligent scheduling method constructed according to multiple service requirements, as described in an embodiment of the present invention. Detailed Implementation

[0032] To facilitate understanding by those skilled in the art, the present invention will be further described below in conjunction with embodiments and accompanying drawings.

[0033] This implementation example Figure 1 The diagram shows two airports and a drone that need to perform six flight route tasks, demonstrating the process of intelligent scheduling based on multiple flight task scheduling.

[0034] like Figure 2 As shown, the intelligent drone scheduling system built based on multiple business requirements consists of four parts, which initialize data and output it after business processing: Input layer: Provides the initial scheduling of the required data, including charging efficiency, drones, flight paths, and drone nests.

[0035] Processing Layer: This is the pre-processing part of the program. It integrates the characteristics of UAV flight services, builds a scheduler based on the data from the input layer, and generates initial sorting rules. It includes a scheduling algorithm engine, a charging management module, a constraint detection module, and a cost calculation module.

[0036] Algorithm layer: Performs simulated annealing algorithm scheduling, iterative calculations, and obtains the optimized route scheduling order. This includes a simulated annealing optimizer, a neighbor solution generator, objective function calculation, and acceptance criterion judgment.

[0037] Business layer: Provides business output to the terminal, outputs the final route scheduling plan, including the execution order of the route and the execution nest, and counts the number of charging times and charging time required by the drone. Based on the above data, it derives the drone's efficiency (total completion time, utilization rate of each drone), cost indicators (total energy consumption, total charging time), and quality indicators (route coverage, charging waiting time), and summarizes the drone's flight performance.

[0038] like Figure 3 The flight paths to be executed and the initial points of the drone nests, nest A and nest B, are defined. There are 6 flight paths, and the task allocation begins through the above implementation process: (1) Obtain flight plan routes and nest locations to construct a basic dataset: Nest A, located at coordinates (0, 0), has a base standard power of 100%, a safe return value of 20%, and a charging efficiency of 5% per minute; Nest B, located at coordinates (30, 0), has a base standard energy of 120%, a safe return value of 25%, and a charging efficiency of 4% per minute; Route 1: Origin: (5, 5), Destination: (10, 10); Flight Duration: 30 minutes, Energy Consumption: 25%; Priority: 2, Time Window: Flexible Route 2: Origin: (15, 5), Destination: (25, 10); Flight Duration: 40 minutes, Energy Consumption: 35%; Priority: 4 (Highest), Time Window: Immediate Route 3: Origin: (35, 5), Destination: (40, 15); Flight Duration: 50 minutes, Energy Consumption: 45%; Priority: 3, Time Window: As soon as possible Route 4: Origin: (45, 0), Destination: (50, 10); Flight duration: 60 minutes, Energy consumption: 55%; Priority: 2, Time window: All day Route 5: Origin: (20, 15), Destination: (10, 5); Flight duration: 35 minutes, Energy consumption: 30%; Priority: 2, Time window: Afternoon Route 6: Origin: (25, 20), Destination: (30, 25); Flight Duration: 25 minutes, Energy Consumption: 20%; Priority: 2, Time Window: Flexible (2) Perform key distance data calculations, using both nest A and nest B to calculate the closest routes to them: Distance from Nest A to each starting point: - Route 1: Distance: 7.07 → Time: 3.5 min, Energy Consumption: 5.7% - Route 2: Distance: 15.81 km → Time: 7.9 min, Energy consumption: 12.6% - Route 5: Distance: 25.00 → Time: 12.5 min, Energy Consumption: 20.0% - Route 6: Distance: 32.02 → Time: 16.0 min, Energy consumption: 25.6% Distance from Nest B to each starting point: - Route 3: Distance: 5.00 → Time: 2.5 min, Energy Consumption: 4.0% - Route 4: Distance: 15.00 → Time: 7.5 min, Energy Consumption: 12.0% - Route 6: Distance: 5.00 → Time: 2.5 min, Energy Consumption: 4.0% - Route 2: Distance: 14.14 → Time: 7.1 min, Energy consumption: 11.3% (3) Write the above calculation results into the initial data to generate the initial dataset.

[0039] (4) Generate the initial solution using a greedy algorithm. Greedy strategy: ① Sorting tasks by priority: Sorting by priority, we get route 2 > route 3 > routes 1, 4, 5, 6.

[0040] ② Select the lowest-cost nest for each task: 1) Route 2: Select Nest B (7.1min < 7.9min) 2) Route 3: Select Nest B (closest distance) 3) Route 1: Select Nest A (closest distance) 4) Route 4: Select Navigation B (closest distance, but battery level needs to be checked) 5) Route 5: Choose Nest A (moderate distance) 6) Route 6: Select Nest B (closest distance) (5) Verify the feasibility of the initial solution by simulating the initial allocation: ① Nest A executes route 1: 0-3.5min: head to the starting point of route 1, 3.5-33.5min: execute route 1, energy consumption: 5.7%+25%=30.7%, remaining 69.3%; ② Nest A executes route 5: It needs to travel from the endpoint of R1 (10, 10) to the starting point of R5 (20, 15): distance 11.18, travel energy consumption: 8.9%; total demand: 8.9% + 30% + 20% = 58.9%, currently 69.3% > 58.9%, feasible but requires charging planning. ③ Nest B was operating routes 2, 3, 4, and 6, and its battery power was clearly insufficient, requiring intermittent charging. (6) Calculate the objective function value of the initial solution. Objective function calculation: (7) Total cost = w1 (time weight 1.0) × total time + w2 (time weight 2.0) × total charging time + w3 (priority weight 10) × priority delay + w4 (default weight 1000) × default. Initial solution cost estimate: ① Total completion time: ~250 minutes ② Total charging time: ~35 minutes ③ Emergency Mission R2 start time: 7.1 minutes ④ Violations: 0 times (assuming feasible) Initial cost = 250 × 1.0 + 35 × 2.0 + 7.1 × 10 = 250 + 70 + 71 = 391 points (8) Simulate the main annealing loop and set the initial state: current solution = initial solution, current cost = 391 points, optimal solution = current solution, optimal cost = 391 points, current temperature T = 1000, and the iteration counter is 0.

[0041] (9) The high-temperature stage iteration (T=1000~500) begins, generating neighbor solutions through 5 stirring strategies. Many iterations are required. The following is a simple example: ① Iteration 1: T=1000, generate neighbor solution: swap route 2 allocation (nest B → nest A), new solution cost: 410 points (worse), ΔC (point difference) = 410 - 391 = 19, acceptance probability P = exp(-19 / 1000) = 0.981, random number r = 0.85 < 0.981 → accept the new solution. Update: current solution = new solution, current cost = 410; ② Iteration 2: T=1000, generate neighbor solutions: adjust the task order so that route 6 is before route 3, new solution cost: 380 points (better), ΔC=-30 → always accept better solutions, update the current solution and the optimal solution; ③ Iteration 3: T=950, generate neighbor solutions: reallocate route 4 to nest A. The new solution requires D1 to return and recharge, increasing the cost. The cost of the new solution is 430 points. P=exp(-(430-380) / 950)=0.948, r=0.92<0.948 → accept the new solution. Update: Current solution = new solution, current cost = 430.

[0042] (10) Iteration in the intermediate temperature stage (T=500~100): Similar to the high temperature iteration, neighbor solutions are generated through 5 stirring strategies. Many iterations are required. The following is a simple example: ① Iteration 50: T=500, current optimal cost: 320 points, generate neighbor solutions: finely adjust the charging timing, new solution cost: 325 points, ΔC=5, P=exp(-5 / 500)=0.990, likely accept the small variation; ② Iteration 100: T=250, produce neighbor solution: let nest B execute multiple tasks continuously and then recharge. New solution: nest B executes R3→(charge)→R36→R1, the cost is reduced to 280 points, update the optimal solution.

[0043] (11) Low temperature stage iteration (T=100~0.1), in the later stage of the low temperature stage, it almost only accepts better solutions, and finally builds a stable cost.

[0044] (12) Produce the optimal solution output, and the final result is as follows: ① Nest A Scheduling: Execute Route 2, Start: 7.9 minutes (departing from Nest A), End: 47.9 minutes, Energy Consumption: 47.6%, Remaining: 52.4%, Execute Route 5, Arrive at Route 5 starting point (20, 15) in 3.5 minutes, Cumulative completion time 51.4 minutes, Route 5 completed, Remaining battery close to 16.7%, Return time 5.6 minutes, Cumulative time 92 minutes, Remaining battery 7.8%; Charge 8.6 minutes, Cumulative time 100.6 minutes, Remaining battery 50.7%, Execute Route 1, Arrive at Route 1 starting point (5, 5) in 3.5 minutes, Energy Consumption 5.7%, Remaining battery 45.0%, Route 1 completed, Cumulative time 134.1 minutes, Remaining battery 20.0%, Return.

[0045] ② Nest B Scheduling: Execute Route 3: Start: 3.5 minutes (departure from Nest B), End: 53.5 minutes, Energy Consumption: 50.7%, Remaining: 69.3%, Position at the end of Route 3 (40, 15), Execute Route 6, Arrive at the starting point of Route 6 (25, 20), Time taken: 7.9 minutes, Cumulative Time taken: 61.4 minutes, Energy Consumption: 12.6%, Remaining Battery: 56.7%, Route 6 completed, Cumulative Time taken: 86.4 minutes, Remaining Battery: 36.7% Return to Nest B, return time 12.5 minutes, total time 98.9 minutes, remaining battery 16.7%, start charging, charging for 19 minutes, total time 117.9 minutes, remaining battery 92.7%, execute route 4, arrive at the starting point of route 4 (45, 0), time 7.5 minutes, total time 125.4 minutes, energy consumption: 12%, remaining battery 80.7%, complete route 4, total time 185.2 minutes, remaining battery 25.7%, return to base.

[0046] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A method for intelligent scheduling of unmanned aerial vehicles (UAVs) based on multiple business requirements, characterized in that: Includes the following steps: S1: Obtain all flight plans that require flight and information on drone nests that conform to the flight plans to build a basic data source; Basic data source information includes: nest coordinates, basic standard power, safe return value power, charging efficiency, route origin, route destination, flight duration, energy consumption, route priority, and time window. S2, parse the flight distance, maneuver and starting point position data in the flight plan, match them with the nest positions, sort the distances from each nest to the starting point position, and construct a nest sort set that matches the flight plan; S3, establishes the constraints for the drone, including maximum flight time and pre-charge time; S4, build a scheduler based on the data of S1-S3, first calculate the route priority, and then generate the initial sorting rules based on the route priority; S5, initial solution production, based on the initial sorting rules in S4, processes each route one by one, finds the nest sorting set in S2, calculates the total cost of executing each route, and selects the optimal UAV for each route; total cost = energy consumption to the starting point + energy consumption of the route flight; If the drone's current battery level is greater than or equal to the total cost: allocate the drone directly, where the mission execution cost is the flight time; If the drone's current battery level is less than the total cost: calculate the charging time, where the mission execution cost is the sum of flight time and charging time; Select the drone allocation with the lowest total cost; production allocation result. S6. Construct a cost calculation function for the total production cost. The total cost is constructed based on a weighted formula (1 × total completion time + 2 × total charging time [seconds] + 10 × priority task delay [number]) + penalty term (1000 × power violation penalty). Select the solution with the fewest penalty terms as the basic cost solution, and define it as the optimal cost and optimal solution. S7 simulates the main annealing loop, constructing a decreasing temperature and performing iterative iterations. The initial temperature is preset to 1000. Within 150 iterations per temperature, neighboring solutions are randomly generated through the basic cost solution in S6. The cost of the neighboring solution is calculated. If the neighboring cost is lower, the neighboring solution replaces the current solution, and the neighboring cost replaces the current cost. If the neighboring cost is higher, the acceptance probability is calculated as exp([current cost - neighboring cost] / current temperature). A random number is generated and compared with the acceptance probability. If the random number is less than the probability, the neighboring solution replaces the current solution, and the neighboring cost replaces the current cost. S8, based on the results in the S7 loop, compares the current cost with the optimal cost. If the current cost is less than the optimal cost, then the optimal cost is replaced with the current cost, and the optimal solution becomes the current solution.

2. The intelligent scheduling method for unmanned aerial vehicles (UAVs) based on multiple service requirements as described in claim 1, characterized in that, The neighbor-to-producer strategy includes the following perturbation operations: Swap the order of the routes; randomly select two routes and swap their execution order. Reassign drone nests, randomly select a flight path, change the drone nest, and rearrange the charging time based on the new drone's battery level and current location. Adjust the start time by randomly selecting a flight route and adjusting the start time by ±30%. Insert charging tasks and adjust the timing specifically for routes that require charging to better coordinate charging with tasks; The charging sequence will be changed, starting with the high-energy-consuming routes.

3. The intelligent drone scheduling method based on multiple service requirements as described in claim 1, characterized in that: Cooling is performed in S8, with the current temperature cooling rate preset between 0.85 and 0.95, until the current temperature is less than 0.1 and the cycle ends.

4. The intelligent scheduling method for unmanned aerial vehicles (UAVs) based on multiple service requirements as described in claim 3, characterized in that: The preset current temperature cooling rate is 0.

93.

5. A drone intelligent scheduling system based on multiple business requirements, characterized in that, The method for executing the intelligent drone scheduling method based on multiple service requirements as described in claims 1-4 includes: Input layer: Provides the initial scheduling of the required data, including basic data sources such as charging efficiency, drones, flight paths, and drone nests; Processing layer: Based on the data from the input layer, it builds a scheduler and generates initial sorting rules; Algorithm layer: Perform simulated annealing algorithm scheduling, iterative calculation, and obtain the optimized route scheduling order; Business layer: Provides business output to the terminal, outputting the final route scheduling plan, including the execution order of the routes and the execution nest, and counts the number of charging times and charging time required by the drone. Based on the above data, it summarizes the drone's efficiency, cost indicators, and quality indicators to determine the drone's flight performance. The drone's efficiency includes total completion time and utilization rate of each drone; the cost indicators include total cost and total charging time; and the quality indicators include route coverage and charging waiting time.

Images