Multi-unmanned aerial vehicle assisted ship emission detection method

Abstract

The invention discloses a multi-unmanned-aerial-vehicle-assisted ship emission detection method, and belongs to the technical field of multi-unmanned-aerial-vehicle emission monitoring, and the method is suitable for unmanned aerial vehicles, and comprises the steps: receiving a trajectory plan of an unmanned aerial vehicle executing a detection task sent by a supervision center, and enabling the detection task to take a ship in a task sea area as a detection target, the trajectory plan is generated based on a trajectory planning method of a genetic algorithm; executing a detection task in the task sea area according to the trajectory plan, detecting the concentration of gas discharged by the ship and inhalable particles through a sensor pre-carried on the unmanned aerial vehicle, and transmitting the detected data to a supervision center to complete the detection task corresponding to the ship; when the task deadline is reached or all tasks are completed, the unmanned aerial vehicle returns to the base station; according to the invention, an unmanned aerial vehicle detection technology is adopted, and multiple unmanned aerial vehicles are deployed at the same time for auxiliary emission detection, so that manpower and time resources are saved, and the detection efficiency and the detection coverage rate are improved.

Description

Technical Field

[0001] The invention relates to a multi-UAV assisted ship emission detection method, belonging to the technical field of multi-UAV emission monitoring. Background Art

[0002] With the development of economic globalization, the demand for international cargo transportation has increased significantly. Among the many means of transportation, sea transportation is usually regarded as the most convenient choice because of its low cost and efficient long-distance transportation capacity. With the development of the shipping industry, the pollution problems it brings are becoming more and more serious, especially the pollution caused by ship emissions. The establishment of emission control areas is one of the key measures to alleviate environmental pollution in coastal areas, and the relevant management departments have taken various measures to regulate the emissions of ships sailing in the sea area. At the same time, there are stricter standards and greater supervision for ships in emission control areas. However, there are still a large number of ships pursuing economic benefits, taking risks and ignoring laws and regulations. At the same time, traditional emission monitoring methods mainly rely on coastal monitoring stations and technical personnel boarding ships. These methods have many limitations such as unsatisfactory monitoring results, requiring a lot of manpower and resources, and low efficiency and poor flexibility.

[0003] Drones have been widely used in various environmental detection tasks due to their excellent flexibility and on-demand scheduling capabilities. Their high maneuverability enables drones to efficiently cover a wide area and conduct real-time detection. Although a single drone has high maneuverability and flexibility in performing tasks, it also has some limitations. For example, a single drone has a limited operating range and endurance, and may not be able to complete the task efficiently when faced with a large area or complex environment. In addition, the failure or unstable performance of a single drone may cause mission interruption or data loss.

[0004] In contrast, the use of multiple drones can overcome these shortcomings. Through division of labor and cooperation, multiple drones can significantly improve the coverage and processing efficiency of the task. Multiple drones can monitor different areas at the same time, thus shortening the time to complete the task. In addition, multiple drones also have strong fault tolerance. If one of them has a problem, other drones can quickly make up for it, ensuring the continuity and stability of the overall task.

[0005] Multi-UAV-assisted emission detection can significantly save manpower and time resources, but considering the limited battery capacity of UAVs, the trajectory of the UAVs needs to be carefully planned so that they can complete as many detection tasks as possible.

[0006] Current research focuses on how drones can detect during ship navigation, but there is less research on multi-drone detection and ship detection coverage. For example, a ship emission detection method and system based on a flying platform provides a ship emission detection method and system based on a flying platform. This method uses a multi-rotor drone equipped with an infrared camera, a laser rangefinder, and a gas sensor module to detect ship emissions. However, when facing multiple ships in a large sea area, there is a lack of path planning and task allocation for multiple drones, and there are still deficiencies in detection efficiency and coverage.

[0007] Therefore, there is an urgent need for a multi-UAV-assisted ship emission detection method that can detect as many ships as possible in the ship navigation scenario in the mission sea area, taking into account the energy and time constraints, and improve the detection efficiency and coverage. Summary of the invention

[0008] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a multi-UAV assisted ship emission detection method, which adopts UAV detection technology and simultaneously deploys multiple UAVs for auxiliary emission detection, thereby saving manpower and time resources and improving detection efficiency and detection coverage.

[0009] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a multi-UAV-assisted ship emission detection method, the method being applicable to UAVs, comprising: Receive a trajectory plan for a drone to perform a detection task sent by a supervision center, wherein the detection task takes ships in the mission sea area as detection targets, and the trajectory plan is generated based on a trajectory planning method of a genetic algorithm; The drone will carry out inspection tasks in the mission sea area according to the trajectory plan, detect the concentration of gases and inhalable particulate matter emitted by the ship through the sensors pre-installed on the drone, and transmit the detected data to the supervision center to complete the corresponding inspection tasks of the ship; when the mission deadline is reached or all tasks are completed, the drone returns to the base station.

[0010] Furthermore, the method for generating the trajectory plan includes: Step 1-1: Obtain ship information in the mission area; deploy drones and obtain drone information; Step 1-2: Generate a chromosome by encoding it according to the drone information and the ship information through the chromosome construction method, where the chromosome represents the task sequence that the drone will perform; Step 1-3: Calculate the feasible solution of the chromosome through the feasible solution generation method, and update the best solution; the feasible solution includes the trajectory plan of the drone; Step 1-4: Repeat steps 1-2 to 1-3 to generate a first-generation population containing multiple chromosomes; Step 1-5: Based on the first generation population, chromosomes are selected by roulette wheel selection method, and a new generation of chromosomes is generated by chromosome reconstruction method; Step 1-6: Use crossover and mutation operations to generate new offspring of the new generation chromosomes to form a new generation population; Step 1-7: Update the feasible solution of each chromosome in the new generation population through the feasible solution generation method, and record the best solution; Step 1-8: Repeat steps 1-5 to 1-7, iteratively evolve the population, and obtain the best solution as the final generated trajectory plan.

[0011] Furthermore, in step 1-2, the chromosome is encoded and generated by a chromosome construction method according to the drone information and the ship information, including: Step 2-1: Adjust the order of tasks according to their urgency and importance. Drone Determine the task to be performed first, including , design reward value formula to sort all tasks in descending order, drone Once upon a time Randomly select a task from the tasks; Step 2-2: In the subsequent task allocation, design a new reward value formula and use a greedy strategy to select the task goal that maximizes the reward value; Step 2-3: Encode the chromosome according to the sequence of tasks to be performed. The chromosome represents The sequence of tasks that the drone will perform, chromosome Represented as a two-dimensional vector defined by the formula: ; , ; in, It's a drone The sequence of tasks to be executed, Defined as a drone Executed tasks, It's a drone The number of tasks to be executed.

[0012] Furthermore, in step 2-1, the reward value formula is: ; in, is the reward value, for The weight of Indicates drone With the task The distance between the initial positions, The higher the value, the more important it is. The smaller it is, the more urgent the task is.

[0013] Furthermore, in step 2-2, the new reward value formula is as follows: ; in, is the new reward value, for The weight of Indicates the current task With the next task The distance between the initial positions.

[0014] Furthermore, in the steps 1-3, the feasible solution of the chromosome is calculated by a feasible solution generation method, including: Step 3-1: Given a chromosome , define a failed task set , record the task failure due to lack of time or energy; Step 3-2: Continuous working state through time To build a drone Trajectory plan ,in Represents trajectory plan The total number of states; trajectory planning Status in by Indicates; among them, Indicates the state target; and Respectively indicate the status The starting and ending positions; Indicates drone In Status Flight speed under and Respectively indicate the status The start and end time of Indicates drone In Status The flight trajectory of the following; generate drones one by one Trajectory plan ; For drones , according to chromosome Task sequence in Execute tasks; Step 3-3: For the drone Tasks to perform ,for From current location Go to the mission execution location , create flight status ; Then create the detection state ; Check the drone status At the end of the energy level, there are two situations: low energy level, or sufficient remaining energy; Step 3-4: When the remaining energy of the drone is negative or insufficient to fly to the nearest base station, it is in a low energy level situation. Before executing the mission, you need to go to the base station to replace the battery, and four states will be generated accordingly: the flight state from the current position to the nearest base station , Charging status , the flight status of going back to the base station to perform the mission And the detection status ; If in state If the completion time constraint or deadline constraint is violated at the end, the task is temporarily abandoned and recorded as a failed task; if in state If the drone's energy level is still low at the end, the mission is temporarily abandoned and recorded as a failed mission; otherwise, the status It is feasible and preserved, and the drone is updated Trajectory plan ; Step 3-5: If the drone The remaining energy is enough to go to the nearest base station and check whether the completion time constraint and deadline constraint are met. If the constraint is violated, the mission is temporarily abandoned. Otherwise, it means that the drone Can successfully execute tasks and update Trajectory plan ; Step 3-6: Set the failed task The tasks in the corresponding chromosome , and we can get a feasible solution by deleting With failed task set , a feasible solution The trajectory plan of all drones The calculation formula is: .

[0015] Furthermore, in the steps 1-5, a new generation of chromosomes is generated by a chromosome reconstruction method, including: Step 4-1: Based on the failed task set through chromosome reconstruction method The failure history information of the task is used to rebuild the chromosome by inserting the failed task into the given chromosome; Step 4-2: Set the failed task The failed tasks iteratively insert a copy of the given chromosome ; Step 4-3: Select chromosomes based on sequence length Task sequence in ; Step 4-4: From the failed task set Randomly select a task ; Adjust according to the historical failure reasons; if the task fails due to insufficient time, there is a greater probability of putting it in the task sequence If a task fails due to insufficient energy, there is a greater probability that it will be inserted into the task sequence. After the more recent mission.

[0016] Furthermore, in the steps 1-6, the new generation of chromosomes is subjected to crossover and mutation operations to generate new offspring, including: Step 5-1: In the crossover operation, the two-dimensional structure of a given chromosome is converted into two linear structures, and random segments of the two linear structures are selected and exchanged; Step 5-2: Delete the repeated tasks in the two linear structures and then transform them into a two-dimensional structure as the new offspring; Step 5-3: In the mutation operation, given a chromosome, convert it into a linear structure, randomly select two tasks and exchange them, and convert the linear structure into a two-dimensional structure of the new chromosome.

[0017] Furthermore, the optimization goal of the method is to find the weights and The largest feasible solution , weights and The calculation formula is as follows: ; in, Contains all completed tasks, and the calculation formula is as follows: .

[0018] Furthermore, the constraints of the method include: Energy Constraints: Limiting Drones The energy level cannot be negative at any time; Continuity Constraints: Limiting Drones On Track Program Status in must be continuous over time; Completion time limit: Guaranteed task This must be completed before the ship arrives at the port; Deadline constraints: guaranteeing tasks Must be within a predefined deadline Otherwise, the task will be abandoned. Distance constraint: limit the distance between drones to maintain a safe distance ; Position constraint: restricts the drone and the ship to meet at the planned location.

[0019] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a multi-drone assisted ship emission detection method, which adopts drone detection technology and simultaneously deploys multiple drones for auxiliary emission detection, which not only improves the detection efficiency and detection coverage; at the same time, a chromosome construction method based on failure history is introduced to obtain a higher quality population, and a reward value formula based on task weight and distance optimization is introduced in the task allocation process to optimize the task execution order of the drone. Different from traditional single drone operations or multi-drone operations based on static algorithms, the present invention considers a dynamic scene, and considers multiple constraints of time and energy, and can dynamically adjust the task allocation strategy according to real-time task requirements, and is more adaptable to changes in complex environments and differences in task priorities, thereby achieving higher detection efficiency and coverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of a multi-UAV-assisted ship emission detection method provided by an embodiment of the present invention; Figure 2 is a comparison chart of performance and effectiveness results of different algorithms provided by embodiments of the present invention; Figure 3 is a schematic diagram of a UAV trajectory plan provided by an embodiment of the present invention; Figure 4 This is an overview of the multi-UAV-assisted ship emission detection method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0022] Embodiment 1: This embodiment introduces a multi-UAV-assisted ship emission detection method, which is applicable to UAVs and includes: Receive a trajectory plan for a drone to perform a detection task sent by a supervision center, wherein the detection task takes ships in the mission sea area as detection targets, and the trajectory plan is generated based on a trajectory planning method of a genetic algorithm; The drone will carry out inspection tasks in the mission sea area according to the trajectory plan, detect the concentration of gases and inhalable particulate matter emitted by the ship through the sensors pre-installed on the drone, and transmit the detected data to the supervision center to complete the corresponding inspection tasks of the ship; when the mission deadline is reached or all tasks are completed, the drone returns to the base station.

[0023] In order to make the content of the present invention more clearly understood, the following is a diagram of the radius. The present invention is further described in detail by taking a semicircular nearshore sea area of ​​9000m as the mission area.

[0024] like Figure 1 and Figure 4 As shown, the application process of the multi-UAV-assisted ship emission detection method provided in this embodiment specifically involves the following steps: Step 1: Construct a multi-UAV-assisted ship emission detection system. In this system, in a mission area under mild meteorological conditions (which does not affect ship navigation and UAV flight, for example, the sea surface wind speed does not exceed 10 meters per second; the horizontal visibility is not less than 5 kilometers; the effective wave height does not exceed 2 meters; there is no extreme weather such as lightning, hail, etc., where the specific values ​​depend on the conditions of the sea area, ships and UAVs), the supervision center will take the ships in the given mission area as the detection task, and deploy multiple UAVs to perform the detection task in the area, and require the UAVs to complete the task within the constraints of the completion time and deadline. At the same time, multiple base stations are set up in the mission area to provide stable signal support and efficient energy supply for the UAVs, effectively improving their endurance.

[0025] Step 2: The supervision center in the system constructs the working status of the UAV and generates a trajectory plan for the UAV to perform the task through the GA-CRO method based on GA trajectory planning.

[0026] Step 3: The drone performs the detection task in the mission area according to the given trajectory plan. The drone detects CO through the high-precision gas sensor it carries. 2 、NO X 、SO X The drones can detect gaseous emissions such as air pollution and various inhalable particulate matter concentrations with the help of highly sensitive optical sensors, so as to efficiently collect ship emission data and transmit the data to the supervision center to complete the corresponding inspection tasks of the ship. When the task deadline is reached or all tasks are completed, all drones return to the base station.

[0027] 2. In step 2, the GA-CRO trajectory planning method based on GA is used to construct the working state of the drone and generate the flight trajectory of the drone for detecting ship emissions. The specific process is shown in steps 1-1 to 1-7: Step 1-1: In the mission area, the system obtains Ships are considered as inspection tasks ,in Represents a set of tasks, . And represent the task as a four-tuple ,in Indicates the initial position of the ship; Indicates the ship's sailing speed; It represents the weight value defined according to the historical violation emission situation and is positively correlated with the violation situation; Indicates the detection time, and the task The corresponding ship is at time The position is defined as . And set up in the mission area base stations, the set of base stations is represented as Each base station The position is represented by Based on this information, the regulatory center deploys drones, all with the same flight speed , the same energy limit , the same charging time And the same energy consumption model , where the energy consumption model With speed According to the above information, chromosomes are generated by chromosome construction method. .

[0028] Assume that there are 5 ships in the area, namely , corresponding to 5 detection tasks, the specific information is shown in Table 1, and the target port of all ships is the same, located at coordinates At the same time, the monitoring center deployed two drones for detection, namely , and the speed of the drone , UAV energy limit , Drone charging time There are two base stations set up along the coast of the area. and , the locations are and .

[0029] Table 1 Detection information

[0030] Step 1-2: For the obtained chromosome , calculate the corresponding feasible solution through the feasible solution generation method , and update the best solution obtained so far .

[0031] Step 1-3: Repeat step 1-1 to step 1-2 until the system is complete. The first generation population of chromosomes .

[0032] Steps 1-4: The system iteratively generates a new population based on the previous generation population , two chromosomes are selected by roulette wheel selection and If the corresponding feasible solution of a chromosome has a higher weight sum, it has a greater chance of being selected. Reconstructing and To create two new chromosomes and .

[0033] Step 1-5: Generate new offspring through crossover and mutation operations and append them to In the new generation population .

[0034] Steps 1-6: Create new chromosomes , update the corresponding feasible solution through the feasible solution construction method , and update the best solution obtained so far .

[0035] Step 1-7: By repeating steps 1-4 to 1-6, the system retains the previous solution according to the weighted sum of the corresponding feasible solutions. chromosomes, and then continuously iterate and evolve to produce new populations, and finally get the best solution to the problem. .

[0036] 3. In step 1-1, according to the ship information, the chromosome is encoded and generated by the chromosome construction method , see steps 2-1 to 2-3 for the specific process: Step 2-1: In order to optimize task allocation, the system further adjusts the order of tasks according to their urgency and importance. Drone Determine the task to be performed first, including Design the following reward value formula to sort all tasks in descending order. Once upon a time Randomly select a task from the tasks.

[0037] ;

[0038] in, for The weight of Indicates drone With the task The distance between the initial positions, specifically, The higher the value, the more important it is. The smaller the value, the more urgent the task is. This reward value takes into account the urgency and importance of the task, ensuring that the drone can give priority to urgent and important tasks.

[0039] Step 2-2: In the subsequent task allocation, considering the weight and distance, the following reward value formula is designed. The subsequent task goal of , using the greedy strategy, selects the task goal that maximizes the reward value.

[0040] ; in, for The weight of Indicates the current task With the next task The distance between the initial positions. That is, in the selection of subsequent tasks, the task with the highest weight-to-distance ratio is selected, so that the drone can perform the task with higher weight with the least consumption.

[0041] Step 2-3: Encode the chromosome according to the sequence of tasks to be performed. The chromosome represents The sequence of tasks that the drone will perform. Represented as a two-dimensional vector defined by the formula: ; , ; in, It's a drone The sequence of tasks to be executed, Defined as a drone Executed tasks, It's a drone The number of tasks to be executed.

[0042] 4. In step 1-2, the corresponding feasible solution is calculated by the feasible solution generation method based on the chromosome information , see steps 3-1 to 3-6 for the specific process: Step 3-1: Given a chromosome , defines a set of failed tasks To record the failed missions and the reasons. There are two main reasons for mission failure: (1) the mission cannot be completed before the corresponding ship reaches the end of the area or the deadline; (2) the mission distance is too far and the remaining energy is insufficient to support such a long distance flight. The former is called insufficient time and the latter is called insufficient energy.

[0043] Step 3-2: Through a series of time-continuous working states To build a drone Trajectory plan ,in Represents trajectory plan The total number of states. Trajectory planning Status in by Indicates. Among them, Indicates the state target, which can be a base station or a task; and Respectively indicate the status The starting and ending positions; Indicates drone In Status Flight speed under and Respectively indicate the status The start and end time of Indicates drone In Status The system will generate the flight trajectory of the drones one by one. Trajectory plan For drones , try to sort by chromosome Task sequence in Execute the task. At the same time, for each failed task, record the task and the reason for failure.

[0044] Step 3-3: For the drone Tasks to perform ,for From current location Go to the mission execution location , create flight status ; Then create the detection state Then, check the status of the drone. Energy level at the end. There are two situations: low energy level, or plenty of energy remaining.

[0045] Step 3-4: If the remaining energy of the drone is negative or insufficient to fly to the nearest base station, it is in a low energy level situation. Before executing the mission, you need to go to the base station to replace the battery, and four states will be generated accordingly: the flight state from the current position to the nearest base station , Charging status , the flight status of going back to the base station to perform the mission And the detection status If in state At the end, the completion time constraint or deadline constraint is violated, and the task needs to be temporarily abandoned and recorded as a failed task. If the drone's energy level is still low at the end, it means the ship is too far away and the mission should be abandoned and recorded as a failed mission. Otherwise, the status It is feasible and preserved, and the drone is updated Trajectory plan .

[0046] Step 3-5: If the drone The remaining energy is enough to go to the nearest base station, that is, the energy level is sufficient. The system needs to check whether the completion time constraint and deadline constraint are met. If the constraint is violated, the task is temporarily abandoned. Otherwise, it means that the drone Can successfully execute tasks and update Trajectory plan .

[0047] According to the mission information described in Table 1, the monitoring center generates a trajectory plan for the UAV as shown in Table 2.

[0048] Table 2 UAV trajectory plan

[0049]

[0050] like Figure 3 As shown, the UAV performs the detection task in the mission area according to the pre-designed trajectory plan.

[0051] Step 3-6: Set the failed task The tasks in the corresponding chromosome , and we can get a feasible solution by deleting With failed task set , a feasible solution The trajectory plan of all drones constitute.

[0052] ; 5. Reconstruct the selected chromosomes in steps 1-4 by the chromosome reconstruction method to generate two new chromosomes and The specific process is shown in steps 4-1 to 4-4: Step 4-1: Based on the failed task set through chromosome reconstruction method The failure history information of the task is used to rebuild the chromosome by inserting the failed task into the given chromosome.

[0053] Step 4-2: Set the failed task The failed tasks iteratively insert a copy of the given chromosome .

[0054] Step 4-3: Select chromosomes based on sequence length Task sequence in , shorter task sequences will have a greater chance of being selected.

[0055] Step 4-4: From the failed task set Randomly select a task . Adjustments are made based on historical failure reasons. If a task fails due to insufficient time, there is a greater probability that it will be placed earlier in the task sequence, which means that the task has more chances to be completed before the deadline or before the completion time. If a task fails due to insufficient energy, there is a greater probability that it will be inserted after the nearest task in the task sequence.

[0056] 6. In steps 1-5, new offspring are generated through two genetic operations, crossover and mutation. The specific process is shown in steps 5-1 to 5-3: Step 5-1: In the crossover operation, the two-dimensional structure of a given chromosome is converted into two linear structures. Random segments of the two linear structures are selected and exchanged.

[0057] Step 5-2: Delete the duplicate tasks in the two linear structures and then transform them into a two-dimensional structure as the new offspring.

[0058] Step 5-3: In the mutation operation, given a chromosome, convert it into a linear structure, then randomly select two tasks and exchange them. Finally, convert the linear structure into a two-dimensional structure of the new chromosome.

[0059] Optimization goal of the algorithm And the constraints are as follows: ; The optimization goal is to find the weights and The largest feasible solution .in, Contains all completed tasks.

[0060] ; Constraints: Energy constraints limit drones The energy level cannot be negative at any time.

[0061] Continuity constraints mean that drones On Track Program Status in Must be continuous over time.

[0062] Complete the time limit guarantee task This must be completed before its corresponding vessel arrives at the port.

[0063] Deadline constraints ensure that tasks Must be within a predefined deadline Otherwise, the task is abandoned.

[0064] Distance constraints limit the safe distance between drones , preventing them from colliding unless they are located at the same base station.

[0065] The position constraint means that when the UAV is heading to the mission location, at the end of this flight state, the ship and the UAV should meet at the same planned location.

[0066] The present invention fully considers the energy consumption, time limit and dynamic change characteristics of the detected target of the drone in trajectory planning. In order to achieve efficient task execution, the trajectory planning of multiple drones is modeled as a series of time-continuous working states. Each state not only reflects the current position and speed of the drone, but also comprehensively considers its energy consumption and flight time, thereby ensuring that the drone completes the task under limited energy and time resources.

[0067] This paper designs an efficient flight trajectory planning framework based on genetic algorithm, aiming to optimize the flight paths of multiple drones to ensure that as many inspection tasks as possible are completed within a limited time. It improves the timeliness of drone operations and effectively improves the coverage and reliability of the entire detection system. The algorithm framework includes three main components: generating chromosomes, constructing feasible solutions, and performing chromosome reconstruction to improve population quality. In addition, crossover and mutation operations are defined.

[0068] In order to cope with the complex dynamic changes and multiple constraints in the problem, the present invention proposes an innovative strategy to reconstruct the chromosomes in the algorithm by using historical failure information. Specifically, the system will conduct an in-depth analysis and summary of the failures that occurred during past executions, identify the key factors that led to the failures, and optimize the current chromosomes based on this information. Through this process, the chromosome structure can be adjusted and improved to obtain higher quality chromosomes, thereby significantly improving the quality of the solution and the execution efficiency.

[0069] The present invention illustrates the effectiveness of the method through experimental comparison. In the experimental comparison, the relative percentage deviation ARPD is used to measure the effectiveness of the solution. Under the same test conditions, the smaller the ARPD, the better the performance and effectiveness. Figure 2 As shown, from the sea area radius , Number of ships , Number of base stations 、Number of drones , UAV speed , UAV energy limit ,expiration date In seven aspects, the ARPD of GA-CRO algorithm, ACO-DSP algorithm, RL-DDSP algorithm and QL algorithm are compared under the same instance. It can be seen from the figure that GA-CRO has better effect and robustness than other algorithms.

[0070] In summary, compared with the prior art, the present invention adopts a multi-UAV detection method based on GA-CRO trajectory planning, which not only improves the detection efficiency and coverage, but also introduces a chromosome construction method based on failure history to obtain a higher quality population, and also introduces a reward value formula based on task weight and distance optimization in the task allocation process to optimize the task execution order of the UAV. Different from traditional single-UAV operations or multi-UAV operations based on static algorithms, the present invention considers a dynamic scene, and considers multiple constraints of time and energy, and can dynamically adjust the task allocation strategy according to real-time task requirements, and is more adaptable to changes in complex environments and differences in task priorities, thereby achieving higher detection efficiency and coverage.

[0071] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multi-UAV-assisted ship emission detection method, characterized in that: The method is applicable to drones and includes: Receive a trajectory plan for a drone to perform a detection task sent by a supervision center, wherein the detection task takes ships in the mission sea area as detection targets, and the trajectory plan is generated based on a trajectory planning method of a genetic algorithm; The drone will carry out inspection tasks in the mission sea area according to the trajectory plan, detect the concentration of gases and inhalable particulate matter emitted by the ship through the sensors pre-installed on the drone, and transmit the detected data to the supervision center to complete the corresponding inspection tasks of the ship; when the mission deadline is reached or all tasks are completed, the drone returns to the base station.

2. The multi-UAV-assisted ship emission detection method according to claim 1 is characterized in that: The method for generating the trajectory plan includes: Step 1-1: Obtain ship information in the mission area; deploy drones and obtain drone information; Step 1-2: Generate a chromosome by encoding it according to the drone information and the ship information through the chromosome construction method, where the chromosome represents the task sequence that the drone will perform; Step 1-3: Calculate the feasible solution of the chromosome through the feasible solution generation method, and update the best solution; the feasible solution includes the trajectory plan of the UAV; Step 1-4: Repeat steps 1-2 to 1-3 to generate a first-generation population containing multiple chromosomes; Step 1-5: Based on the first generation population, chromosomes are selected by roulette wheel selection method, and a new generation of chromosomes is generated by chromosome reconstruction method; Step 1-6: Use crossover and mutation operations to generate new offspring of the new generation chromosomes to form a new generation population; Step 1-7: Update the feasible solution of each chromosome in the new generation population through the feasible solution generation method, and record the best solution; Step 1-8: Repeat steps 1-5 to 1-7, iteratively evolve the population, and obtain the best solution as the final generated trajectory plan.

3. The multi-UAV-assisted ship emission detection method according to claim 2 is characterized in that: In step 1-2, the chromosome is generated by encoding according to the drone information and the ship information through the chromosome construction method, including: Step 2-1: Adjust the order of tasks according to their urgency and importance. Drone Determine the task to be performed first, including , design reward value formula to sort all tasks in descending order, drone Once upon a time Randomly select a task from the tasks; Step 2-2: In the subsequent task allocation, design a new reward value formula and use a greedy strategy to select the task goal that maximizes the reward value; Step 2-3: Encode the chromosome according to the sequence of tasks to be performed. The chromosome represents The sequence of tasks that the drone will perform, chromosome Represented as a two-dimensional vector defined by the formula: ； , ； in, It's a drone The sequence of tasks to be executed, Defined as a drone Executed tasks, It's a drone The number of tasks to be executed.

4. The multi-UAV-assisted ship emission detection method according to claim 3 is characterized in that: In step 2-1, the reward value formula is: ； in, is the reward value, for The weight of Indicates drone With the task The distance between the initial positions, The higher the value, the more important it is. The smaller it is, the more urgent the task is.

5. The multi-UAV-assisted ship emission detection method according to claim 3 is characterized in that: In step 2-2, the new reward value formula is as follows: ； in, is the new reward value, for The weight of Indicates the current task With the next task The distance between the initial positions.

6. The multi-UAV-assisted ship emission detection method according to claim 2 is characterized in that: In the steps 1-3, the feasible solution of the chromosome is calculated by a feasible solution generation method, including: Step 3-1: Given a chromosome , define a failed task set , record the task failure due to lack of time or energy; Step 3-2: Continuous working state through time To build a drone Trajectory plan ,in Represents trajectory plan The total number of states; trajectory planning Status in by Indicates; among them, Indicates the state target; and Respectively indicate the status The starting and ending positions; Indicates drone In Status Flight speed under and Respectively indicate the status The start and end time of Indicates drone In Status The flight trajectory of the following; generate drones one by one Trajectory plan ; For drones , according to chromosome Task sequence in Execute tasks; Step 3-3: For the drone Tasks to perform ,for From current location Go to the mission execution location , create flight status ; Then create the detection state ; Check the drone status At the end of the energy level, there are two situations: low energy level, or sufficient remaining energy; Step 3-4: When the remaining energy of the drone is negative or insufficient to fly to the nearest base station, it is in a low energy level situation. Before executing the mission, you need to go to the base station to replace the battery, and four states will be generated accordingly: the flight state from the current position to the nearest base station , Charging status , the flight status of going back to the base station to perform the mission And the detection status ; If in state If the completion time constraint or deadline constraint is violated at the end, the task is temporarily abandoned and recorded as a failed task; if in state If the drone's energy level is still low at the end, the mission is temporarily abandoned and recorded as a failed mission; otherwise, the status It is feasible and preserved, and the drone is updated Trajectory plan ; Step 3-5: If the drone The remaining energy is enough to go to the nearest base station and check whether the completion time constraint and deadline constraint are met. If the constraint is violated, the mission is temporarily abandoned. Otherwise, it means that the drone Can successfully execute tasks and update Trajectory plan ; Step 3-6: Set the failed task The tasks in the corresponding chromosome , and we can get a feasible solution by deleting With failed task set , a feasible solution The trajectory plan of all drones The calculation formula is: .

7. The multi-UAV-assisted ship emission detection method according to claim 2 is characterized in that: In the steps 1-5, a new generation of chromosomes is generated by a chromosome reconstruction method, including: Step 4-1: Based on the failed task set through chromosome reconstruction method The failure history information of the task is used to reconstruct the chromosome by inserting the failed task into the given chromosome; Step 4-2: Set the failed task The failed tasks iteratively insert a copy of the given chromosome ; Step 4-3: Select chromosomes based on sequence length Task sequence in ; Step 4-4: From the failed task set Randomly select a task ; Adjust according to the historical failure reasons; if the task fails due to insufficient time, there is a greater probability of putting it in the task sequence If a task fails due to insufficient energy, there is a greater probability that it will be inserted into the task sequence. After the more recent mission.

8. The multi-UAV-assisted ship emission detection method according to claim 2 is characterized in that: In the steps 1-6, the new generation of chromosomes is subjected to crossover and mutation operations to generate new offspring, including: Step 5-1: In the crossover operation, the two-dimensional structure of a given chromosome is converted into two linear structures, and random segments of the two linear structures are selected and exchanged; Step 5-2: Delete the repeated tasks in the two linear structures and then transform them into a two-dimensional structure as the new offspring; Step 5-3: In the mutation operation, given a chromosome, convert it into a linear structure, randomly select two tasks and exchange them, and convert the linear structure into a two-dimensional structure of the new chromosome.

9. The multi-UAV-assisted ship emission detection method according to claim 2, characterized in that: The optimization goal of the method is to find the weights and The largest feasible solution , weights and The calculation formula is as follows: ； in, Contains all completed tasks, and the calculation formula is as follows: 。 10. The multi-UAV-assisted ship emission detection method according to claim 2, characterized in that: The constraints of the method include: Energy Constraints: Limiting Drones The energy level cannot be negative at any time; Continuity Constraints: Limiting Drones On-Trajectory Program Status in must be continuous over time; Completion time limit: Guaranteed task This must be completed before the ship arrives at the port; Deadline constraints: guaranteeing tasks Must be within a predefined deadline Otherwise, the task will be abandoned. Distance constraint: limit the distance between drones to maintain a safe distance ; Position constraint: restricts the drone and the ship to meet at the planned location.

Images